Inexpensive magnetic flux (gauss) meter [Archive]

Warske

Dec 06, 2004, 09:46 PM

After building the probe, I started looking on the Web for information.

This page shows a diagram of a 4 wire Hall Sensor and some information on measuring magnetic fields.
http://my.execpc.com/~rhoadley/magflux.htm

This is a similar project (written in Dutch), incorporating a 3v battery, current limiting resistor, and push button switch. Looks much better than mine!
http://www.modelbouwforum.nl/phpBB2/viewtopic.php?t=7834

Another meter using sensors made by Allegro.
http://my.execpc.com/~rhoadley/magmeter.htm

I had wanted to do more with this, but I'm out of time. In fact, I will be out of town for a few days and I won't have access to this forum until I'm back. If you have more information on this subject, please feel free to post it so others can benefit.

Warske
-----------------------------

Edit 2/7/2005 - Added
Links into this thread (by post number):

Gauss Meter
What is a Hall sensor & how does it work 122 (http://www.rcgroups.com/forums/showthread.php?p=3261978) 127 (http://www.rcgroups.com/forums/showthread.php?p=3269617)
Linearity & saturation 13 (http://www.rcgroups.com/forums/showthread.php?p=3025907)
Calibration
...Single turn coil 79 (http://www.rcgroups.com/forums/showthread.php?p=3191564)
...Earth's magnetic field 119 (http://www.rcgroups.com/forums/showthread.php?p=3247252)
Offset 119 (http://www.rcgroups.com/forums/showthread.php?p=3247252)
Regulated power supply 59 (http://www.rcgroups.com/forums/showthread.php?p=3143332)
Flux meter with detachable probe 69 (http://www.rcgroups.com/forums/showthread.php?p=3157990)
Comparing magnets 15 (http://www.rcgroups.com/forums/showthread.php?p=3038396) 59 (http://www.rcgroups.com/forums/showthread.php?p=3143332)
Measuring a solenoid 83 (http://www.rcgroups.com/forums/showthread.php?p=3198870)
Measuring Earth's magnetic field 119 (http://www.rcgroups.com/forums/showthread.php?p=3247252)
Destructive testing of permanent magnets 20 (http://www.rcgroups.com/forums/showthread.php?p=3071211)
Other gauss meters/sensors. Measuring stronger fields 111 (http://www.rcgroups.com/forums/showthread.php?p=3232239) 115 (http://www.rcgroups.com/forums/showthread.php?p=3234224)
Magnetic field calculators 111 (http://www.rcgroups.com/forums/showthread.php?p=3232239)

Magnetizing Neodymium Magnets
Magnetizing Neodymium 30 (http://www.rcgroups.com/forums/showthread.php?p=3097667)
Forming capacitors 32 (http://www.rcgroups.com/forums/showthread.php?p=3098761)
Resonant frequency of an LC circuit 45 (http://www.rcgroups.com/forums/showthread.php?p=3116839)
Storage oscilliscope alternative 48 (http://www.rcgroups.com/forums/showthread.php?p=3126177)
Schematic and WinSpice3 transient analysis of pulse 51 (http://www.rcgroups.com/forums/showthread.php?p=3131839)
Optimizing the coil 55 (http://www.rcgroups.com/forums/showthread.php?p=3136959) 62 (http://www.rcgroups.com/forums/showthread.php?p=3145883) 74 (http://www.rcgroups.com/forums/showthread.php?p=3182596) 118 (http://www.rcgroups.com/forums/showthread.php?p=3241471)
Capacitor ESR 57 (http://www.rcgroups.com/forums/showthread.php?p=3138267)
Flyback diode 66 (http://www.rcgroups.com/forums/showthread.php?p=3153685) 70 (http://www.rcgroups.com/forums/showthread.php?p=3170877) 100 (http://www.rcgroups.com/forums/showthread.php?p=3226099)
High voltage power supplies
...Electronic flash units 81 (http://www.rcgroups.com/forums/showthread.php?p=3195654) 89 (http://www.rcgroups.com/forums/showthread.php?p=3202160) 91 (http://www.rcgroups.com/forums/showthread.php?p=3219821) 94 (http://www.rcgroups.com/forums/showthread.php?p=3220097)
...Variac and transformer 118 (http://www.rcgroups.com/forums/showthread.php?p=3241471)
Switches
...High speed mechanical switch 48 (http://www.rcgroups.com/forums/showthread.php?p=3126177)
...SCR 100 (http://www.rcgroups.com/forums/showthread.php?p=3226099) 118 (http://www.rcgroups.com/forums/showthread.php?p=3241471)
Converting Gauss to Oersteds 83 (http://www.rcgroups.com/forums/showthread.php?p=3198870)
Measuring capacitance 95 (http://www.rcgroups.com/forums/showthread.php?p=3220317)
Field required to magnetize Neodymium (graph) 104 (http://www.rcgroups.com/forums/showthread.php?p=3227357)

MAVA

Dec 06, 2004, 10:38 PM

Warske,

Thank you for your post.

Very informative.

Martin

Andy W

Dec 07, 2004, 06:01 AM

Now, can you make a flux capacitor? :D
..a

(good posts, great info!)

empeabee

Dec 08, 2004, 10:25 AM

There are different ways to unsolder the chips. I built a little tool from hookup wire and used it with an 800 degree tip on my soldering iron as shown in the pictures.

When the solder melts, the chip comes away in the tool because of surface tension. You need to be ready to flick it into a can or remove it with tweezers before it cooks to death.

Very good poor mans ' branding Iron ' to remove the chip'
Very good job - I'll be building one RSN
Thanks

Mike

rumplestilskin

Dec 08, 2004, 11:42 AM

WOW, thanks!!

Warske

Dec 10, 2004, 05:09 PM

Thanks for all the feedback. I updated post #3 with pictures to better show how to remove the circuit board from the motor.

If anyone finds a datasheet for the hall sensor, please post the info for the rest of us. Also if you find other sources for cheap analog output hall sensors...

This is off topic, but if you are also interested in what to do with the motor, you might want to check out these links:

RC Groups Discussion > Airplanes - Electric > Power Systems
Brushless CD-ROM Motor SURPLUS
http://www.rcgroups.com/forums/showthread.php?t=203806
I think the above is the genesis of GoBrushless.com

RC Groups Discussion > Airplanes - Electric > Power Systems
One hour CD_ROM motor construction
http://www.rcgroups.com/forums/showthread.php?t=127606

Warske

Warske

Dec 13, 2004, 11:42 PM

I ran a test to check the linearity of my GoBrushless Hall sensor. As you can see from the graph, it is quite linear to 150 mV, and fairly linear to 300 mV.

Considering the Earth's magnetic field is about 0.5 Gauss, and the sensor reads the ambient field in my living room as 0.58 mV, I figure this gives me a usable linear range of up to about 250 gauss (0.5 * 300 / 0.58).

I've included the data if you want to play with it. Note that the sensor was tested at an applied voltage of 1.3987 V and the sensor has an offset of 3.1 mV.

Of course, the sensor is still very useful outside its linear range to check magnet field direction and to determine if one magnet is stronger than another.

Now that I know what the linear range is, I'm thinking it would be interesting to test the temperature sensitivity of the N45 magnets that came with my GoBrushless kit. What temperature causes them to degrade, and how much do they recover when they cool down?

Warske

Dec 13, 2004, 11:44 PM

Let me mention how I tested the sensor so you can test your own or come up with a better idea and post it here! :D

The concept was to use an electromagnet and a variable voltage power supply to provide the test field. Unfortunately, if I used an iron core to intensify the field, I might be measuring the saturation of the core instead of the sensor.

To get a coreless coil, I took apart a dead fluorescent light ballast. This worked great up to about 0.7 amps, but that only got me to 110 mV of sensor response. At higher currents, the coil starts to get hot, its resistance goes up, and the current isn't stable enough for a good reading. Plus I didn't want to heat up the sensor and be reading its temperature coefficient.

To get past that problem, I took the readings in 5 stages. After the first stage with 110.1 mV at 0.7 amps, I set the current to zero and biased the sensor back to 110.1 mV using a nice big rare earth magnet. I then recorded the response up to 0.7 amps again and repeated the procedure. Since magnetic fields add together in the absence of saturable material, this should work and the graph suggests that it did. Five stages let me get up to an effective current of 3.5 amps, which would have vaporized my coil. BTW, I figure it is important to make the test platform quite ridged so that the magnet won't move toward the coil as current is applied.

It was surprisingly easy to bias up the sensor to the exact last reading. I just positioned the magnet with blocks of wood to give a slightly high reading, then shifted it to one side to get the exact number.

Warske

Dec 16, 2004, 07:44 PM

Do you fly with matched batteries? How about your motor magnets? Are they all the same strength? How do you know if one of them is a dud?

Now that we've established the linear range of the flux sensor, its time to get to some practical applications.

As you know if you've looked at the CD ROM motor building threads, its important that you put the same number of turns on each tooth so that the ESC doesn't get confused. But what if one of the magnets is (gasp) *weaker* than the others? (This is what the marketing folks are into when they talk about creating a need where none previously existed. But no, I'm not selling these sensors. You have to build one yourself! :D )

Using my usual cardboard and masking tape, I rigged up a test jig. This allowed me to keep the sensor at a fixed distance from the magnet (a distance where the flux density is within its linear range), and let me accurately place the magnets relative to the sensor.

Warske

Dec 16, 2004, 07:46 PM

I then checked the magnets from an unused GoBrushless kit.

First I took 1 magnet and tried its 8 orientations in the holder (4 sides on the magnet, either up or down). I got a range of 123.10 to 125.27 with the North pole facing up, and -123.81 to -126.31 with the South pole up (the readings are milivolts). I also noted that the sensor has a histeresis of 0.5 mV when you flip poles, so I decided to check all the magnets with North pole up.

Since the last two digits were in the noise, I switched the meter to its 20 V scale so it would round the numbers for me. I then checked my stack of 14 magnets. The result was an average of 126.6, with a range of 122 to 131.

The kit was advertised as containing N45 magnets, but I also bought a stack of N50 magnets from GoBrushless. Since they phased out their N45 5x5x1 magnets, I thought there was a chance that the kit had N50's also so I was interested to see if my stack of N50's were different.

From the stack of N50s I tested 10 and got an average of 154.70 with a range of 147 to 161. Thus, 22% stronger on average. Apparently the kit had N45's after all.

From the chart at http://www.mmcmagnetics.com/ourproducts/pdf/SinteredNdFeB.pdf it appears that the N50's are only rated 5% stronger than N45's. Interesting.

Next: Comparing the GoBrushless plastic ring magnet to N45s. Perhaps eventually we'll even get to the flux capacitor! :D

Warske

Other methods of comparing magnets
Deflection of wire (http://www.rcgroups.com/forums/showthread.php?p=3064484)
The Counter Top Test (http://www.rcgroups.com/forums/showthread.php?p=2209467)

empeabee

Dec 16, 2004, 08:08 PM

I then checked the magnets from an unused GoBrushless kit.

First I took 1 magnet and tried its 8 orientations in the holder (4 sides on the magnet, either up or down). I got a range of 123.10 to 125.27 with the North pole facing up, and -123.81 to -126.31 with the South pole up (the readings are milivolts). I also noted that the sensor has a histeresis of 0.5 mV when you flip poles, so I decided to check all the magnets with North pole up.

Since the last two digits were in the noise, I switched the meter to its 20 V scale so it would round the numbers for me. I then checked my stack of 14 magnets. The result was an average of 126.6, with a range of 122 to 131.

The kit was advertised as containing N45 magnets, but I also bought a stack of N50 magnets from GoBrushless. Since they phased out their N45 5x5x1 magnets, I thought there was a chance that the kit had N50's also so I was interested to see if my stack of N50's were different.

From the stack of N50s I tested 10 and got an average of 154.70 with a range of 147 to 161. Thus, 22% stronger on average. Apparently the kit had N45's after all.

From the chart at http://www.mmcmagnetics.com/ourproducts/pdf/SinteredNdFeB.pdf it appears that the N50's are only rated 5% stronger than N45's. Interesting.

Next: Comparing the GoBrushless plastic ring magnet to N45s. Perhaps eventually we'll even get to the flux capacitor! :D

Warske

This is better than Dallas ;) :D
Mike ( waiting to c who shot JR :) )

Warske

Dec 21, 2004, 03:34 AM

This is better than Dallas ;) :D
Mike ( waiting to c who shot JR :) )
Wait 'till we get to destructive testing! :D

I used the flux meter to compare the ring magnet in the Gobrushless $4 motor to an end bell that had 12 of the 5x5x1 N45 magnets glued in.

To keep the flux in the range of my linearity chart (http://www.rcgroups.com/forums/showthread.php?p=3025907), I taped a piece of popsicle stick to the end of the probe to space it away from the surface of the magnets (see picture).

Then I moved the probe to a point where I obtained the highest reading in each case.

Results:
Ring 0.254 volts
N45 0.354 volts

Since 0.354 is in the nonlinear response area of the sensor, but still on my chart, I looked up the "effective current" on the chart and used that to compare the magnets. This gives me a ratio of 2.85 / 1.75 = 1.6285

That is, the N45 magnets create about 63% more flux in the area where the sensor was located.

Next, does it hurt a neodymium magnet to drop it on the floor? How about boiling it in water? Stay tuned!

Warske

Bill Harris

Dec 21, 2004, 10:52 AM

Now, can you make a flux capacitor?
..a

That one is easy:

http://users.rcn.com/zap.dnai/cnslt001.html

Use it on my DrWho Stick. Woiks like a charm.

--Bill

Warske

Dec 24, 2004, 09:11 PM

That one is easy
Flux capacitors are everywhere, once you start looking! :)

Now lets get to the fun stuff: Destructive testing.

Drop Testing
Remember when they told you if you drop a magnet, it will loose its strength? And that you aren't supposed to let them slap together for the same reason (not to mention you risk breaking the ceramic and neodymium ones)? Ever wonder if that was really true?

Recently I bought a HiMax outrunner motor, and in order to reconfigure it, I needed to drive the shaft through to the other side. Which I did using a hammer. Only later I learned you were supposed to use a vice and *press* the shaft through (there are no instructions with the motor or on their WEB site). Then I measured the motor's Kv, and found it to be a bit higher than advertised, which is not a good thing. (This is one reason, by the way, to measure and record the Kv of a new motor, so you can tell later if the magnets got weaker).

So I thought I would slap some of my neodymium magnets around a bit, to see what would happen.

I used the test setup shown in post 15 (http://www.rcgroups.com/forums/showthread.php?p=3038396) to take before and after measurements of the magnets. I marked one edge of each magnet to ensure that I measured it in the same orientation each time.

Results:

N45
125 mV Before drop testing
124 mV After dropping on linoleum floor from 5 feet.
125 mV After dropping 5 more times. Did the magnet get stronger, or is this measurement error? (Somehow it seems unlikely that it got stronger. :) )

N50
157 mV Before drop testing
156 mV After dropping it three times on linoleum floor from 5 feet.
159 mV After tossing it three times against side of the fridge.

Looks to me like this little shock test didn't affect them, at least within the accuracy of the test. Maybe there's a way to make the test more accurate. I wonder if these neodymium magnets are less shock sensitive than the old alnico ones?

Temperature testing
For this test, immersion in boiling water seemed an easy thing to try. It's a pretty constant temperature close to 100 deg C (I'm only a couple hundred feet above sea level) and its self agitating so you get a fairly uniform temperature distribution.

N45
125 mV Before thermal testing
098 mV After removing pan with boiling water from stove and dropping magnet in for 5 minutes.
098 mV After repeat of above.
089 mV After leaving magnet in pan of gently boiling water for 5 minutes.
019 mV After repeat of above, but accidentally letting the pan boil dry for a minute or two. Don't know how hot it got, but the Teflon on the pan seemed fine.

N50
157 mV Before thermal testing
092 mV After leaving magnet in pan of gently boiling water for 5 minutes.
091 mV After repeat of above.

So what happened?
The bit of drop testing didn't phase them, but 100 deg C temperatures definitely reduced their strength. Since they are only rated to 80 deg C, that isn't too surprising, but I was surprised to see how much magnetism was left. I also thought it was interesting that both magnets came down to a similar level in boiling water, though the N50 started much higher.

It would be interesting to test over a range of temperatures, and to include some of the higher temp magnets as well. Hey, has anyone else built one of these flux meters yet? Feel free to jump in with your ideas and results! :)

Next time: Are these parboiled magnets now junk, or can they be restored? Stay tuned and learn the shocking truth!

Warske

empeabee

Dec 24, 2004, 09:25 PM

Flux capacitors are everywhere, once you start looking! :)

Now lets get to the fun stuff: Destructive testing.

Next time: Are these parboiled magnets now junk, or can they be restored? Stay tuned and learn the shocking truth!

AH so for once the ' recieved wisdom ' was half true and half false.

Thank you for the testing - now for the Lazarous ( dyslexia rules KO ) chapter.
Have a good Xmas

Mike

tolladay

Dec 29, 2004, 12:51 AM

Warske,

Me neighbor has a device for making magnets stronger. He uses it for his hobby which is custom winding electronic pick-ups for guitars. Is this what you are intending to use on the boiled magnets?

Warske

Dec 30, 2004, 12:53 AM

empeabee - Thanks for your support.

tolladay - I used your PDF of MetroGTi's magnet mounting method (http://www.rcgroups.com/forums/showthread.php?p=2222289) in the can picture I posted in #18 above. Thanks for making that available. I also added a link to your magnet testing method (post #16 above).

I don't know what is used for magnetizing guitar pickups, but it might be something like this (http://www.headwize.com/projects/showfile.php?file=provost_prj.htm). Apparently Neodymium magnets require considerably more magnetizing force than most other types of permanent magnets (about 10 x more than Alnico 2), so it might not work. But it might be fun to try, and if you do, please let the rest of us know how it goes.

---------------

After some experimentation, I was able to restore both the N45 and N50 magnets to within 96% of their original strength (using the hall sensor, or course, to take the readings). In the process I managed to flip the polarity on one of them, and then flip it back. I was surprised at how easy it was, and using parts on hand it took less than two hours, even including the failed attempts. Eventually I may try to optimize my setup and see if I can get the magnets up to or even higher than their original strength.

When I originally ran the experiments, I planned to post complete information on what I used. Since then I tripped across a thread where someone had admitted gluing a prop, and that person was roundly and almost universally condemned by a wagonload of people, even after stating that he had used it in a deserted area. I can sorta see the point. After all, I would rather not be around someone flying with a glued prop, so why encourage other people to do it? On the other hand, I expect that driving around to view Christmas lights (as I have done) poses an even greater risk to oneself and one's community (just look at the auto accident statistics!) and therefore merits at least comparable condemnation. Which is to say, I found most people's reaction to the prop gluing to be rather extreme, even scary.

This is all by way of explaining why I have decided not to post the construction details of the magnetizer, or the pictures I took of the blast area. Some people might consider it to be less than safe since hearing and eye protection are required, along with a knowledge of and respect for the hazards of electrocution.

Of course, much of that could be said for taking apart a toaster.

Warske

lazy-b

Dec 30, 2004, 03:39 AM

Warske: have you try using the Hall effect sensor, to make a DC clamp Amp-meter? Was thinking of using a donut shape Ferrite core, use in Switching Power supply.......can cut a slot on the Ferrite core....and insert the sensor in the slot....I think the current is proportional to the magnetic strenght.

Now to make a nice opennable a donut Ferite, might use 2 ferite donut, 1st donut can cut a "U" shape on the edge, and 2nd donut can cut a "I" shape, so that the "U" will mate with the "I" shape, to minimize the loss of magnetic field......I am not sure if this is necessary, I have not actually done it yet.....a Simple straight cut might do the job.

Your Graph shows, linear from 0 to 0.250 Volts, Its just right for 0.200 range for Digital Multimeter.

Base on your Solenoid experiment at 1AMPS, the Hall sensor produce 0.15 Volts......I just wonder how many turn is the Solenoid, cause a 1 amps current is multiply a no. of turn of wire to produce a magnetic field strong enough to be detected by hall sensor........image, if we need to measure just a single wire with a current of 1 AMPS only pass thru a Ferrite core, I guess we have to amplify that hall sensor to a same no. of times (No. of Turn of wire in Solenoid) to produce 0.15 Volts.

Might need a little less amplification due to the Ferite core help concentrate the magnetic field into smaller area compare to plain AIR only.

its a nice tool for converting a Digital Tester to DC Clamp Amp-meter.

empeabee

Dec 30, 2004, 06:48 AM

olmod

Dec 30, 2004, 08:24 AM

Im particularly intrerested if one can change the polarity of neos' so that the north south is from one edge to other in order to make a halbach array ;)

tolladay

Dec 30, 2004, 11:21 AM

empeabee

Dec 30, 2004, 12:03 PM

Warske,

You talking to a guy who is the proud owner of "the Anarchist Cookbook" (of course it's not something I'll be ready to show my 3.5 year-old boy just yet. ;) ) My point is, information is information. If someone wants to be stupid about what they learn that's a definate problem, but it is not the fault of the information.

Many guys have posted photos of their li-polys blowing up, but no one has condemmed them for being unsafe. In fact they have been lauded for doing a public service.

Glueing on a prop sound less like a safety issue to me, and more like a Darwin Award recipient. But blowing up things, while taking careful and clear measurements; now thats science! :D

I say post and be darned. Besides, with a title like "Inexpensive magnetic flux (gauss) meter" I don't think many teen-agers are going to be looking our wayin order to find something stupid to do. :D

I agree, but I think Warske may well be worried by what I call the Drunken Aeroplane syndrome - where a widow sued a plane manufacture, and won, when her husband got drunk, & flew down a narrow street below rooftop level - leaving the wings behind, but collecting his own private set and a harp.
She said the plane should have been designed to not allow that to happen.
That sort of thing is starting to happen here now. :confused:

Mike

Warske

Dec 31, 2004, 12:35 AM

Warske: have you try using the Hall effect sensor, to make a DC clamp Amp-meter? Was thinking of using a donut shape Ferrite core, use in Switching Power supply.......can cut a slot on the Ferrite core....and insert the sensor in the slot....
No I haven't tried it, but your idea should work as I've heard that's how they do it.

BTW, for other's reading this thread, most clamp-on ammeters only measure AC current. They work by clamping a "core" of magnetic material around the wire to be measured, and another coil of wire wrapped around the core material creates the secondary winding of a transformer. No extra battery or electronics are needed, which makes it less expensive to manufacture than the DC version.

Now to make a nice opennable a donut Ferite, might use 2 ferite donut, 1st donut can cut a "U" shape on the edge, and 2nd donut can cut a "I" shape, so that the "U" will mate with the "I" shape, to minimize the loss of magnetic field......I am not sure if this is necessary, I have not actually done it yet.....a Simple straight cut might do the job.
I think that keeping the air gaps small would be helpful. I would think even more important would be to keep them constant, since changes would affect your calibration. Your mating shapes might help with that as well by keeping them lined up.

I just wonder how many turn is the Solenoid
I don't know. Should be possible to figure it out with a physics book in hand. (Not wanting to unwind my solenoid.) I think I mentioned what the reading is for Earth's magnetic field, which is about 0.5 Gauss. The physics book should tell what sort of field you get from 1 amp turn.

its a nice tool for converting a Digital Tester to DC Clamp Amp-meter.
It sounds like an excellent project. If you make some progress with this, please let us know what you find out.

Warske

Warske

Dec 31, 2004, 12:36 AM

Im particularly intrerested if one can change the polarity of neos' so that the north south is from one edge to other in order to make a halbach array ;)
halbach array? I looked it up with Google and it does sound interesting. I read somewhere that the neo magnets may have an orientation and that they may be difficult to polarize against it. If I get things set up again I will test it, but I burned out my HV power supply trying to charge up my capacitor bank faster.

----------------

Ok folks, you twisted my arm. :) It sounds like there is enough interest here that we can keep the Propeller Police backed off. And I'll take the advice about the hazards list. So lets do science!

Let me emphasize that this is just a demonstration of physical principals. I do NOT recommend that you try this yourself. See the list of hazards involved at the end of this post.

Materials used
A length of 12 gauge house wiring was used to wind a small coil.

An electronic flash unit from a camera supplied 330 volts DC.

Four 750 uF 330 v capacitors wired in parallel. I put in strips of note card paper to keep the vaporized copper off the tops of the capacitors.

A blocking diode was used to prevent the capacitors from reverse charging (The inductor-capacitor circuit wants to oscillate, and this would reverse charge the capacitor).

A plastic pen was used to support the test magnet, which was taped to the end of the pen and inserted into the center of the coil.

Procedure
The capacitor bank was charged to 330 volts, then the coil wires were shorted across the capacitor bank wires. Note that the flash from the arc contains high intensity UV (harmful to look at) and the noise requires hearing protection.

I started out with one capacitor, with disappointing results. Each capacitor I added improved the strength of the test magnet. Magnetizing more than once with the same number of capacitors did not improve the strength. Unfortunately I only had four of these capacitors. As mentioned in a previous post, I was able to restore the N45 and N50 magnets to 96% of their former strength.

Links
After I ran the tests, I searched the WEB.
Here is some information about magnetizing permanent magnets (http://www.oersted.com/magnetics_articles.shtml) using a method similar to this.
A chart of Required Magnetizing Force (http://www.walkerscientific.com/Products/Product_Lines/Magnetizing_Stations/Magnetizers/MagProcArticle/magprocarticle.html)
More on magnetizing (http://hyperphysics.phy-astr.gsu.edu/hbase/solids/magperm.html)

Possible future enhancements
Apparently an SCR would work well as a switch instead of my crude contact switch. This would allow the device to be more easily activated from a distance. But it would also be possible to accomplish the same thing using a (home made) relay.

It would be better to use magnet wire with thinner insulation, to pack more turns into the area around the magnet.

My calculations suggest (I could be wrong) that more turns of wire in the coil would be better, up to the point where wire resistance takes over. Doubling the number of turns increases the coil resistance four times, assuming the coils are the same size (and thinner wire is used).

Potential hazards, or why you should not try this yourself
- Electrocution from the high voltage and current, potentially causing death.
- Hearing damage from the sound of the discharge.
- Eye damage from the UV flash.
- Hazard of breathing vaporized copper.
- Exploding capacitors (flying debris) if reverse charged.
- Damaging sensitive equipment (wrist watch) and erasing data due to intense magnetic field.
- Exploding Capacitors on first charge. If left uncharged, must be "formed". (contributed by empeabee)
- Ferrous items (including magnet) being ejected from the coil at high speed. (contributed by empeabee)
- Other hazards not listed.

Warske

Edit 2/7/2005: For an index into the Magnetizing Neodymium section of this thread, see post #7 (http://www.rcgroups.com/forums/showthread.php?p=2994828)

empeabee

Dec 31, 2004, 08:30 AM

Ok folks, you twisted my arm. :)
Let me emphasize that this is just a demonstration of physical principals. I do NOT recommend that you try this yourself. See the list of hazards involved at the end of this post.

Materials used
A length of 12 gauge house wiring was used to wind a small coil.

An electronic flash unit from a camera supplied 330 volts DC.

Four 750 uF 330 v capacitors wired in parallel. I put in strips of note card paper to keep the vaporized copper off the tops of the capacitors.

A blocking diode was used to prevent the capacitors from reverse charging (The inductor-capacitor circuit wants to oscillate, and this would reverse charge the capacitor).

A plastic pen was used to support the test magnet, which was taped to the end of the pen and inserted into the center of the coil.

Procedure
The capacitor bank was charged to 330 volts, then the coil wires were shorted across the capacitor bank wires. Note that the flash from the arc contains high intensity UV (harmful to look at) and the noise requires hearing protection.

I started out with one capacitor, with disappointing results. Each capacitor I added improved the strength of the test magnet. Magnetizing more than once with the same number of capacitors did not improve the strength. Unfortunately I only had four of these capacitors. As mentioned in a previous post, I was able to restore the N45 and N50 magnets to 96% of their former strength.

Links
After I ran the tests, I searched the WEB.
Here is some information about magnetizing permanent magnets (http://www.oersted.com/magnetics_articles.shtml) using a method similar to this.
A chart of Required Magnetizing Force (http://www.walkerscientific.com/Products/Product_Lines/Magnetizing_Stations/Magnetizers/MagProcArticle/magprocarticle.html)
More on magnetizing (http://hyperphysics.phy-astr.gsu.edu/hbase/solids/magperm.html)

Possible future enhancements
Apparently an SCR would work well as a switch instead of my crude contact switch. This would allow the device to be more easily activated from a distance. But it would also be possible to accomplish the same thing using a (home made) relay.

It would be better to use magnet wire with thinner insulation, to pack more turns into the area around the magnet.

My calculations suggest (I could be wrong) that more turns of wire in the coil would be better, up to the point where wire resistance takes over. Doubling the number of turns increases the coil resistance four times, assuming the coils are the same size (and thinner wire is used).

Potential hazards, or why you should not try this yourself
- Electrocution from the high voltage and current, potentially causing death.
- Hearing damage from the sound of the discharge.
- Eye damage from the UV flash.
- Hazard of breathing vaporized copper.
- Exploding capacitors (flying debris) if reverse charged.
- Damaging sensitive equipment (wrist watch) and erasing data due to intense magnetic field.
- Other hazards not listed.

Warske

Thank you Waske.

May I add 2 more Hazards

1)
Exploding Capacitors on first charge if they are old/surplus items left uncharged for long time - recomend very slow charge first time.
I have seen one punch a hole in a room roof when it let go.

2) Ferrus items ( including magnet ) beging ejected from the coil at high speed.

A nice bit of Proof Of Principle Prototype Engineering.

I can see why you frightened your self and were reluctant to publish - there is a lot of high rate energy there.

A SCR will do - I have one rescued from an experimental pulse laser that switched 300+ amps at 1KV at 400 Hz - it is about 4 inches square and 1 inch thick.
I think it is rated at 2kv 400A - not exactly Radio Shack.
That would get over the Flash & Bang ( except when it went to that Silicon foundry in the sky its self).

Again - Thank You.

Mike

Warske

Dec 31, 2004, 09:54 AM

Thank you Waske.

May I add 2 more Hazards

1)
Exploding Capacitors on first charge if they are old/surplus items left uncharged for long time - recomend very slow charge first time.
I have seen one punch a hole in a room roof when it let go.

2) Ferrus items ( including magnet ) beging ejected from the coil at high speed.
Thanks Mike. I added those to the list (with credit to you). Eventually the list will be the longest part of the post. :)

I forgot to mention that I had formed the capacitors for several hours before starting. The electronic flash power supply was fairly low current, which helped.

For other people reading this: As Mike said, the electrolytic caps need to be charged at very low current the first time after not being used. Putting several thousand ohms of resistance in series is one way to do this. For more info, Google "forming capacitors" (http://www.google.com/search?hl=en&lr=&q=%22forming+capacitors%22&btnG=Search)

I can see why you frightened your self and were reluctant to publish - there is a lot of high rate energy there.
You were right about my concerns. My hope is that it is still possible to publish this sort of thing. I recently heard about "base jumping" (http://www.google.com/search?hl=en&lr=&q=%22base+jumping%22) where you jump off a cliff, etc. with a single parachute and you or your survivors tell people about it. Perhaps that is similar.

A SCR will do - I have one rescued from an experimental pulse laser that switched 300+ amps at 1KV at 400 Hz - it is about 4 inches square and 1 inch thick.
I think it is rated at 2kv 400A - not exactly Radio Shack.
That would get over the Flash & Bang ( except when it went to that Silicon foundry in the sky its self).
Sounds like just the thing. I did a quick search yesterday for surplus SCR's and it looks like they are available. I shouldn't need one quite as big, since I can let it cool between shots. I'll probably try to measure or calculate the current impulse before I decide what I need.

Warske

empeabee

Dec 31, 2004, 12:13 PM

The result of the Nanny society - people used to just jump off the cliff - now they take along the parachute incase they change their minds halfway down. :D ;)
Mike

empeabee

Dec 31, 2004, 12:18 PM

Sounds like just the thing. I did a quick search yesterday for surplus SCR's and it looks like they are available. I shouldn't need one quite as big, since I can let it cool between shots. I'll probably try to measure or calculate the current impulse before I decide what I need.

Warske

The initial inrush on that Fence Wire coil of yours will be pretty high - I'd say 100 amps at least - agreed no for long, but look what it did to your copper wires.... pity the poor silcon :)

Mike

olmod

Dec 31, 2004, 07:47 PM

Warske and others ,a very informative thread and the links were good also :) needless to say i wont be trying it :D i think ill just take a few magnets along to a pro' and have them done for me ;)
Happy new year to all. :D

Warske

Jan 01, 2005, 12:07 AM

i think ill just take a few magnets along to a pro' and have them done for me ;)
olmod, does that mean you know someone who has the equipment to magnetize these neo magnets? :)

Warske

Warske

Jan 01, 2005, 01:29 AM

The initial inrush on that Fence Wire coil of yours will be pretty high - I'd say 100 amps at least
I've been trying to get a handle on this, and I think you're right.

Trying to determine the peak current is a pretty problem. To help get there, I measured the coil:

Inductance: 3.825 uH
Resistance: 6.531 mohm
Turns: 20
Length: 0.946 inches
ID: 0..330 inches
OD: 1.375 inches

The equivalent problem to analyze is the transient response of a series LCR circuit to a step input voltage. We know V, L, & C, but we don't know R.

R is going to be the sum of the 6.531 mohm coil resistance plus the capacitor resistance (ESR) plus the contact resistance of my "switch." Unfortunately, I don't have much of a handle on the latter two components.

Just to take some quick shots at the peak current, what would it be if it were limited only by the coil resistance?
330 v / 0.006531 ohms
= 50,528 amps which is a lot.

And what if it were limited just by the inductance (assuming R is 0)?
For that calculation we can use conservation of energy and the knowledge that an LC circuit will trade its energy between the inductor and the capacitor as it oscillates. That is, when the capacitor voltage is zero, all the energy is in the magnetic field of the inductor, and when the inductor current is zero, all the energy is in the electric field of the capacitor.

The capacitor energy is
E = 1/2 * C * V^2
= 1/2 * (750e-6 * 4) * 330^2
= 163 joules

The inductor energy is
E = 1/2 * L * i^2

So the peak inductor current (assuming R = 0) is
i = sqr(2 * E / L)
= sqr(2 * 163 / 3.825e-6)
= 9,231 amps which is still a lot.

Lets approach it from a different angle. How much current did we need to magnetize the neo magnet?

From the "Required Magnetizing Force" link in post #30, we see that we need 20k to 45k Oersteds. Since we were apparently on the low side of full magnetization, lets assume we generated 20k Oersteds. How much current did that take?

I found a Magnetic Conversion Chart (http://www.walkerscientific.com/Products/FAQ/Conversions/conversions.html) which tells me to multiply Oersteds by 79.58 to get Ampere-turns/meter. Since we know the number of turns and the length of the coil, that should give us the amps required to get our 20k Oersteds.
20e3 Oersteds
79.58 fudge factor
20 turns
.946 inches coil length
39.3700787 inches per meter
20e3 * 79.58 / 20 * .946 / 39.3700787
= 1,912 amps

So maybe a couple thousand amps were required to magnetize the magnet, and we could have gotten up to about 9 thousand amps if our total resistance were zero. Its beginning to sound like we're in the ball park.

Just for fun, lets see what the series resistance would be if that were the only thing limiting us to 2,000 amps.
330 / 2000
= 0.165 ohms
That doesn't look too unreasonable. With a transient analysis program, we could factor in the inductance and get a real number.

I'm thinking that I may be able to get an actual current reading using the hall sensor (see, we're still on topic!), an oscilloscope, and a camera.

I bought 15 more capacitors from ebay, and they're 450 v 470 uF units. More voltage, less capacitance, about the same energy storage per device as my original four. Should arrive some time next week. I also bought a rat trap at the hardware store, which I'm thinking might be the basis for a high speed remotely activated switch. :)

Warske

olmod

Jan 01, 2005, 01:34 AM

He manufactures one of the leading insrument and hi fi speakers in the world and has a lot of equipment including magnetisers ,now i dont know how far up he can go ,but from one of the links you posted rare earth need more grunt to magnetise than the usual.he is also my scource for the special iron he uses for centre poles,he gives me the shorts from his CNC lathe :D

empeabee

Jan 01, 2005, 05:50 AM

I've been trying to get a handle on this, and I think you're right.

Trying to determine the peak current is a pretty problem. To help get there, I measured the coil:

Inductance: 3.825 uH
Resistance: 6.531 mohm
Turns: 20
Length: 0.946 inches
ID: 0..330 inches
OD: 1.375 inches

The equivalent problem to analyze is the transient response of a series LCR circuit to a step input voltage. We know V, L, & C, but we don't know R.

R is going to be the sum of the 6.531 mohm coil resistance plus the capacitor resistance (ESR) plus the contact resistance of my "switch." Unfortunately, I don't have much of a handle on the latter two components.

Just to take some quick shots at the peak current, what would it be if it were limited only by the coil resistance?
330 v / 0.006531 ohms
= 50,528 amps which is a lot.

And what if it were limited just by the inductance (assuming R is 0)?
For that calculation we can use conservation of energy and the knowledge that an LC circuit will trade its energy between the inductor and the capacitor as it oscillates. That is, when the capacitor voltage is zero, all the energy is in the magnetic field of the inductor, and when the inductor current is zero, all the energy is in the electric field of the capacitor.

The capacitor energy is
E = 1/2 * C * V^2
= 1/2 * (750e-6 * 4) * 330^2
= 163 joules

The inductor energy is
E = 1/2 * L * i^2

So the peak inductor current (assuming R = 0) is
i = sqr(2 * E / L)
= sqr(2 * 163 / 3.825e-6)
= 9,231 amps which is still a lot.

Lets approach it from a different angle. How much current did we need to magnetize the neo magnet?

From the "Required Magnetizing Force" link in post #30, we see that we need 20k to 45k Oersteds. Since we were apparently on the low side of full magnetization, lets assume we generated 20k Oersteds. How much current did that take?

I found a Magnetic Conversion Chart (http://www.walkerscientific.com/Products/FAQ/Conversions/conversions.html) which tells me to multiply Oersteds by 79.58 to get Ampere-turns/meter. Since we know the number of turns and the length of the coil, that should give us the amps required to get our 20k Oersteds.
20e3 Oersteds
79.58 fudge factor
20 turns
.946 inches coil length
39.3700787 inches per meter
20e3 * 79.58 / 20 * .946 / 39.3700787
= 1,912 amps

So maybe a couple thousand amps were required to magnetize the magnet, and we could have gotten up to about 9 thousand amps if our total resistance were zero. Its beginning to sound like we're in the ball park.

Just for fun, lets see what the series resistance would be if that were the only thing limiting us to 2,000 amps.
330 / 2000
= 0.165 ohms
That doesn't look too unreasonable. With a transient analysis program, we could factor in the inductance and get a real number.

I'm thinking that I may be able to get an actual current reading using the hall sensor (see, we're still on topic!), an oscilloscope, and a camera.

I bought 15 more capacitors from ebay, and they're 450 v 470 uF units. More voltage, less capacitance, about the same energy storage per device as my original four. Should arrive some time next week. I also bought a rat trap at the hardware store, which I'm thinking might be the basis for a high speed remotely activated switch. :)

Warske

Oh my giddy aunt - It's so long since I did that sort of maths - er about 1960.. so I was only an order of magnetude low, small beer :eek:
As for your High Speed Posistive Action System Energiser - I'm green with envy - it's a realy super lateral thaught, and it will work.
Mind you you still need ear protection :D :D :D

Keep going.
Mike

PDK

Jan 02, 2005, 07:32 AM

Hi,

I am planning on building a large LRK outrunner and have heaps of 1.5mm magnets from smaller projects. If two of these smaller magnets are stacked together will the magnetic strength of the combination be the same as a single 3mm magnet?

Best regards
Phil

olmod

Jan 03, 2005, 08:35 AM

I had a chat today with the gentleman who knows about the magnatiser i previously mentioned,a handy thing to know is that it has removable focusing poles and they increase the power of gauss and it is a large and powerfull machine, he was inclined to say that neos' cant be remagnetised edge to edge as in the cooling down period when being moulded they are subjected to a field to help line up the structures,I was not inclined to argue that you had allready flipped one pole for pole ,now i wonder wether you can do one edge to edge. :)

thanhTran

Jan 04, 2005, 12:47 AM

Thank you all for this informative thread with a lot of knowledge. You all sound like scientists. The information here reminds me quite many things I've forgotten :). I feel like I'm in a class again. It's amazing that you can use the electronic flash circuit to be your high voltage source with very high amps.

When I was kid I used to charge a 1000 uF 63 Volts with a high voltage (using 6 volts - 110 volts adapter with 6 volts AC in the primary coil) and sneak up behind someone and short the two leads just to scare them :)

I also read a little book for kids about recharging a magnet. In there is says use copper wire like what you did and turn about 10 - 20 turns, but the power source is a 12-Volt car battery. I've never tried that as I'm afraid that I would burn the wire or burn my hand. In your test, the current will be gone when the capacity finish discharging, but in the battery set up it would not :)

Thanks again

Thanh

Warske

Jan 04, 2005, 12:53 AM

empeabee, I'm glad you liked the switch idea. :)

...have heaps of 1.5mm magnets from smaller projects. If two of these smaller magnets are stacked together will the magnetic strength of the combination be the same as a single 3mm magnet?
PDK, I don't know any reason why not, but that doesn't mean there isn't one. Obviously you loose a tad due to the protective coating between magnets, which is dead space, but I doubt that would be noticeable. You might try asking the LRK board on Yahoo (http://groups.yahoo.com/group/lrk-torquemax/), for a more definitive answer.

olmod, Thanks for the info on how the pros do this.:) I'll have to find out about focusing poles.

About changing the orientation, I finally found the reference I was searching for (http://www.oersted.com/about_magnets_3.shtml): "Many modern magnet materials are said to be ANISOTROPIC, meaning that they have a preferred axis of magnetic field, built-in during manufacture. The part may be magnetized in either direction along this axis, north-south or south-north. The part is extremely resistant to magnetizing in any other direction, however."

This echoes what your friend had said, but this says that flipping pole for pole isn't a problem. Based on this, I would think that if the neo was oriented for edge-to-edge during manufacture, it would magnetize OK edge to edge, but otherwise not.

I plan to try it when I get my fixture operating again in a few days. If it doesn't work, I think you could still take cube shaped neo magnets and use them to make your halbach array. Or you could make cube magnets out of, say, 5 of the 5 x 5 x 1's by stacking them together to get 5 x 5 x 5.

Perhaps the orientation could be removed by heating, but then they wouldn't be as strong. Which gives me an idea: maybe that's what happened with the boiled magnets. If true, it may be that creating a stronger magnetic field isn't going to help get them back to full strength. The magnetic field may be plenty strong already. I'll have to try a non-boiled magnet and see how it compares.

Talking about magnetic toys and curiosities, I found a very interesting web site that I think is worth a look: Science Toys (https://secure.avdns7.com/~scitoys/scitoys/scitoys/magnets/magnets.html)

-----------------------

I want to talk about the reason for the rat trap, but first I want to mention how I got the L and R values I reported last time on the coil:

Measuring the coil

To get the inductance, I connected a 0.68 uF capacitor (0.671 uF measured) in parallel with the coil and swept it with a signal generator. I found that the combination rings at 99.34 khz (the amplitude at this frequency was about 10 times that at nearby frequencies) and used the formula: L = 1 / C * (2 * pi * f)^2

To get the resistance, I passed 10 amps through the coil and measured its voltage drop.

Why a rat trap?

You no doubt noticed from the picture a while back that a lot of energy went into vaporizing copper instead of into the magnetic field. There was also a lot of light and sound that went with it. What we want is for as much of the 163 joules of stored electrical energy from the capacitor to become magnetic energy in the coil, and not be lost as heat and light.

How much energy is 163 joules anyway?
- A 100 watt light bulb turned on for 1.6 seconds.
- A 1 pound weight raised 120 feet.
- A 1 pound weight traveling at 60 MPH (which is the speed at which the above weight hits the ground if you drop it, neglecting air resistance).
- A 1 pound lead weight raised in temperature by 5 deg F.

As you can see, we don't have much energy here to waste in vaporizing copper (although that 60 MPH would hurt if you were in the way).

So how can we minimize the energy loss? Part of it is due to having two contact points, which gives us twice the loss we would get with just one. That's easy to fix buy using a switch.

And part of it is that before the contacts actually touch, an electric arc is formed because the air molecules ionize, and this arc has resistance and thus consumes power. This arc will last as long as the contacts are separated.

It turns out that air ionizes at about 40 kV per inch, depending on humidity, pressure, etc. So the 330 volts will break the air down when the contacts are about 8.25 thousandths of an inch apart. What we want to do is minimize the amount of power this arc consumes. We can do this by bringing the contacts together more quickly.

Because the inductance of the coil limits the inrush current, the current starts out low and builds to a peak. Based on the resonance frequency of the coil and capacitor, the current should peak after about 0.168 mS. It would be nice if we could close the contacts within a fraction of that time, say 10% (just to pull a number out of the air).

This gives us a goal for how fast we want the contacts to close. They should travel 8.25 thousandths of an inch in 10% of 0.168 mS, which is 0.0168 mS.
This is 8.25e-3 / 0.0168e-3 = 491 inches/sec
or 27.9 MPH

Will the rat trap do it for us? Who knows? But at least it will close the contacts a lot faster than I could by hand. Not only that, but it will close the contacts with a great deal of force. That should make the contact resistance really low (until they bounce, of course).

Yesterday I downloaded a couple of transient analysis programs based on the well known SPICE program. What I have in mind is to get a handle on the capacitor ESR and switch contact resistance, then try to model the circuit and get predicted output waveforms. This will help me to design the coil (what inductance and resistance to use). Finally, it should be possible to capture the actual output waveform using the hall sensor, and see how it compares with the prediction.:)

Warske

thanhTran

Jan 04, 2005, 12:54 AM

Warske

Jan 04, 2005, 01:43 AM

I remember my physics teacher told me when I was in 12th grade that a coil and a cap would create a resonate circuit. That may explain why you get that high current. I forgot all the calculations or the physic to come up with a formular though :)

Hi thanhTran,

Thanks for your encouragement. I've mostly been getting the formulas off the WEB or out of an old physics book. The resonant frequency of an LC circuit is

f = 1/(2 * pi) * sqr(1/(L * C))

and in our case we have
L = 3.825 uH
C = 750 * 4 = 3000 uF

So,
f = 1/(2 * 3.14159) * sqr(1/(3.825e-6 * 3000e-6))
= 1.485 khz

The period is
1 / 1485 = 0.673 mS

And the first current peak would be at 1/4 cycle (just the first peak of a sine wave).
0.673 / 4 = 0.168 mS

We don't really know what the current is, but there were some equations and speculation back in post #37. Hope to get a better handle on that soon.:)

By they way, if anyone spots a calculation or logic error, please let me know and I'll edit my original post with credit to you for the correction. I figure this info could conceivably stick around for a few years and be useful. More useful if it is actually correct. :) I know that I make lots of mistakes. Also let me know if you want to see the math behind a calculation, and I'll be happy to post it.

Warske

olmod

Jan 04, 2005, 07:30 AM

the gentleman also stated that when measuring with a gauss meter one should place either side of the magnet being tested pole pieces wich poke out in front and then place the probe between ,them not on the magnet face.
he then went into a lengthy explanation how the probes need to be treated with the utmost care as any pressure or bending will throw the calibration out.he also mentioned that you need big old oil paper capacators built for those sorts of applications.he then mentioned a pulse type magnetiser that does not use capacators mmmm he would not go into ant detail on that tho' ;) i thought when i saw the rat trap that was what you were going to do :D cheers.

empeabee

Jan 04, 2005, 08:31 AM

Warske
Most of the advances in most fields are the result of mistakes e.g. the transistor - the guy was trying to make 2 identical diodes for radar recievers - the clever bit is building on the results of the mistake :D

Your dismembered rat trap looks as though you could have got round the bounce problems by getting a cold weld on the 2 copper contact strips.

I just found 20 6800 uF 450 volt Caps from the old pulse laser - will start reforming some today - now where did I put those sandbags :D

Mike

Warske

Jan 05, 2005, 09:56 PM

he then mentioned a pulse type magnetiser that does not use capacators mmmm he would not go into ant detail on that tho' ;)
Thanks for all the info. The more the better! I have read that some magnetizers just used a half cycle of the mains voltage. If you had a 220 volt service, a half cycle of that would give you an 8 mS pulse (in the US) of 311 volts peak (root 2 times 220). That would work quite well. My capacitors were only charged to 330 v.

Many years ago, I worked in a lab at Texas Instruments in the motor controls div. The back room had special 440 VAC service piped in from the power grid with very high current capability. It was for testing their motor protector designs. They had SCRs set up to apply the full 440 to whatever they wanted for a few cycles. On slow days you could hear explosions coming from back there. The techs got bored, and were using it to blow up capacitors.

Your dismembered rat trap looks as though you could have got round the bounce problems by getting a cold weld on the 2 copper contact strips.
I have hopes!
I just found 20 6800 uF 450 volt Caps from the old pulse laser - will start reforming some today - now where did I put those sandbags :D
My puny capacitors pale by comparison! Each one of yours is more than 4 times the energy of my entire bank.! :) Happy magnetizing!

---------------------

Now that I've got it souped-up, lets see just how fast the rat trap will go!

Actually, since I silver soldered a stiffening bar to the business end, it won't be as fast as the rats experience, but it will be interesting to see how close it is to the desired 30 MPH.

I added another switch by attaching a bent paper clip to the side (see picture) which contacts the arm as it goes by. That signal goes to the 'scope trigger input. (Note that, instead of a scope, an inexpensive counter and frequency source could have been used to get the data.)

Then I captured a trace of the arm hitting the upper contact, and finally the upper and lower contacts coming together (see picture, and note the contact bounce later in the trace). The contacts are made out of flattened copper tubing.

I don't have a storage scope, but it worked pretty well to stick the camera on a tripod and point it at the scope. I set the camera for half second exposures, then clicked the shutter button and pulled the cord to release the rat trap.

With the scope set to 0.1 mS/division, you can see that the time for the contacts to close is 0.12 mS. With a contact gap of 0.048 inches, this works out to 0.048 / 0.12e-3 = 400 inches/sec. or 22.72 miles per hour. This is less than the 30 MPH I was hoping for, but it should still be a big improvement from before.

By the way, in the first picture you will notice some big C clamps holding the switch to the table. Without these, I found that my readings were somewhat random. The board would jump when I released the spring, and that affected the results.

Warske

new flyer

Jan 05, 2005, 10:18 PM

[QUOTE=Warske] Perhaps eventually we'll even get to the flux capacitor! :D
[QUOTE]

what is a flux capacitor

Warske

Jan 06, 2005, 01:39 AM

what is a flux capacitor
For my concept, see the picture below. I was spinning off of Andy W's post (http://www.rcgroups.com/forums/showthread.php?p=2996049):
Now, can you make a flux capacitor? :D
..a

You can also read more about it here (http://www.google.com/search?as_q=&num=10&hl=en&btnG=Google+Search&as_epq=flux+capacitor&as_oq=&as_eq=&lr=&as_ft=i&as_filetype=&as_qdr=all&as_nlo=&as_nhi=&as_occt=any&as_dt=i&as_sitesearch=rcgroups.com&safe=images), and there are other references which Google didn't pick up.

I think its a bit like the concept of a brass magnet. My dad used to tell the story like this (I think he actually saw this happen):

A little after the new guy Fred was hired, one of the foremen asked him to go down to supply and bring back the brass magnet, knowing of course that there was no such thing. When Fred asked supply for it, the fellow there told him to wait, and took a regular magnet and plated it with brass, which he gave to Fred without comment. Fred took it to the Forman, who was bewildered and exclaimed: "There really is a brass magnet!"

My picture is in reply to Andy's question, though I'm not sure he realized it! :)

Warske

Warske

Jan 06, 2005, 11:34 PM

I've measured the parameters of my capacitor, coil, and switch, and used those in a transient analysis program to model the behavior of the circuit.

The only parameter I didn't measure was the arc resistance. To get that I used the previous assumption of 2k amps peak, and adjusted the resistance until the simulation came out right. That gave me an arc resistance of 117 mOhms.

Attached is the schematic, the simulation output, and the circuit file I used.

For a simulator I downloaded and installed WinSpice3 v1.05.01 (http://www.ousetech.co.uk/winspice2/) which is fully functioning shareware. This one runs under Windows, but there are probably other versions of Spice3 for other systems.

To run the simulation, you would need to rename the circuit file to have a .cir file extension. Then in WinSpice3, you click File, Open, and open the circuit file. It should plot the data automatically. To modify the file, open it with Notepad, make the changes, and save it. The new plot should be generated automatically.

Looking at the attached plots of capacitor voltage and coil current, notice that the capacitor voltage never goes negative. Thus, my flyback diode wasn't actually needed, and I didn't model it in this simulation.

Warske

Warske

Jan 06, 2005, 11:35 PM

With the new rat trap switch, the resistance is about 1mOhm. Notice what this does in the simulation.

We get lots more magnetizing current, but the capacitor voltage goes negative, which in real life could cause it to explode. Its time to put in a flyback diode.

In case anyone asks, these curves are known as exponentially decaying sinusoids.

Warske

Warske

Jan 06, 2005, 11:35 PM

In this simulation, I added a flyback diode across the capacitor to keep it from reverse charging.

Notice that the peak current through the coil (red) is the same, but the diode current (green) gets up to nearly 4,000 amps. The Capacitor current is shown in blue.

Its a good thing I didn't actually test my hardware this way, since my diode is only rated at 3 amps continuous. But for short pulses of current like this, the diode will take more than three amps. With this pulse width, it might even take up to 400 times its continuous rated current, based on absorbed energy ratings. That would be 1,200 amps. Still not enough.

I can get a bigger diode, or I can add some resistance back into the circuit until it is critically damped so that the voltage doesn't try to go negative.

Next, I want to explore how the number of turns of wire in the coil affects the circuit behavior.

Warske

empeabee

Jan 07, 2005, 07:24 AM

simulation

Oh wow - Tha puts the Kybosh on my guestimated 100A inrush doesn't it :eek:

No wonder the early experimenters blew things up in comprehensive ways !

Thank you for your elightening work.

Will play with the sim b4 I try my 6800 uf caps - looks like I will need the sandbags! ( got a healthy spark from 9 volts so who knows what will happen on 250v from an old ' tube ' sig gen I found)
Time to work Very Slowly indeed.

Mike

Warske

Jan 08, 2005, 01:53 AM

empeabee,

Those monster caps of yours probably have quite a bit less resistance than what I've been working with. They should give you even more current! :)

I read that microwave oven transformers step up to around 4000 volts (http://www.dansworkshop.com/Homebuilt%20arc%20welder.shtml), and can be salvaged from dead ovens. If you need more than 250 v ...

---------------------

What happens if we keep the size of the coil the same, but change the number of turns?

If there were no resistance in the circuit, all of the energy in the capacitor would transfer to the magnetic field in the coil by the end of the first quarter cycle of oscillation. That is, the strength of the magnetic field would be determined only by the amount of energy in the capacitor, and the number of turns in the coil wouldn't matter.

For example, if we double the number of turns (from 20 to 40), the inductance of the coil goes up by a factor of 4. The oscillation frequency and the peak current are both cut in half. The number of amp-turns (and thus the strength of the magnetic field) remains the same.

As an aside, note that the oscillation frequency isn't really all that high. Its only about 1.3 kHz, so audio range test equipment is adequate for playing with much of this.

In reality, there is resistance in the circuit, and if we change the number of turns, we have to change the wire diameter so we can fit the new number of turns into the same space.

If we double the number of turns, the coil resistance (due to twice the length of wire with half the cross sectional area) goes up by a factor of 4, just like the inductance.

At first glance, this would seem to suggest that a larger number of turns is going in the wrong direction since the circuit resistance is increased. But I've neglected to deal with the resistance of the capacitor, which is large compared to the resistance of our coil.

Since the inductance goes up and the current goes down with more turns, the voltage lost across the capacitor resistance will also be reduced, and that's an improvement. The little bit of extra resistance added by the increased turns in the coil isn't enough to offset this improvement. Unless we go too far, of course. When the coil resistance is large compared to the capacitor resistance, increasing the number of turns will reduce the performance.

This simulation (see attachments) shows what happens when we double the number of turns in the coil (increasing both L and R by 4x). The peak current went down, but its still more than half. This means that the amp-turns in the coil went up. There are other advantages to the reduced current. The flyback diode and the switch don't have to be as rugged.

We can fine tune the number of turns and use the simulator to determine what the optimum is.
For my present circuit, I found that the amp-turns peaks when the coil has 38 turns (with correspondingly smaller wire) and its resistance is slightly larger than all the other circuit resistance combined (23.6 mOhms vs 21.0 mOhms).

Of course, the coil I used had thick insulation. Using magnet wire should work even better.

This is all assuming that we want to get the biggest magnetic field out of the charge in the capacitors. With more capacitors, it should be possible to use more turns (reducing the current) and still get a large enough field to magnetize the neo magnets.

Warske

empeabee

Jan 08, 2005, 09:17 AM

empeabee,

Those monster caps of yours probably have quite a bit less resistance than what I've been working with. They should give you even more current! :)

Using your Sim files & upping the C to 6800 uf resulted in only 500 Amps more in the coil - the pulse just lasted longer !

doubling to 13600 uf again only upped the amps a bit more!

huh

Smelt an old old smell today - the old sig gen used selenium rectifiers, and they went to that big selenium factory in the sky - with apropriate bad eggs smell - & that was starting the geni on 110vac ( its 230v nominal) to re-form it's capacitors. Back to the drawing board :( :mad:

( uWave psu - dead ones difficult to get here in UK due to lunatic re-cycling laws.)

Instead of more turns in same size, how about making a C core out of an old torroid transformer with the magnet in the (small) gap ?
= more gauss's to the magnet.
Maybe that could be coupled with the half wave multi pulse from the AC supply Idea. No caps to go bang.
Mike

Warske

Jan 08, 2005, 12:32 PM

Using your Sim files & upping the C to 6800 uf resulted in only 500 Amps more in the coil - the pulse just lasted longer !
That's interesting. I have to admit, it seems a lot easier to tweek a number and run the simulation than it is to heat up the soldering iron and reconnect 12 ga wire.

What will really shoot your amps up in the simulation is to lower the capacitor resistance. With about double the capacitance, I would guess you have about half the resistance. So you could set rCap in the simulation to a value of 10m. My guess comes from the idea that you can put two caps in parallel and get double the capacitance, but two resistors in parallel gives you half the resistance.

About capacitor resistance, the best reference I've seen so far is Bob Parker's ESR Meter Page (http://members.ozemail.com.au/~bobpar/esrmeter.htm). He did some research on what the highest resistance should be for a good electrolytic capacitor, and found that it didn't necessarily work like I suggested above. It probably depends on the way they are built. By the way, his ESR meter might not read low enough resistance to be useful here. It was designed more to detect capacitors that were out of spec.

Smelt an old old smell today - the old sig gen used selenium rectifiers, and they went to that big selenium factory in the sky - with apropriate bad eggs smell - & that was starting the geni on 110vac ( its 230v nominal) to re-form it's capacitors. Back to the drawing board
Now that is sad. :( But maybe its repairable with cheap silicon diodes? The signal generator I'm using is an old tube type Eico hobby model, and so far I've been lucky.

If you were wanting to use the signal generator to check your coil inductance, another way to get the answer would be to calculate it. On page 3 of this pdf file (http://www.radioelectronicschool.com/reading/reading14.pdf), I found a formula for the inductance of "single-layer air-core coils with a length approximately equal to the diameter"

L = r^2 * n^2 / ( 24 * r + 25 * l )
where
L is inductance in uH
r and l are radius and length in centimeters
n is number of turns

The formula gave me 5.449 uH, vs the 3.825 uH I measured. Not too bad considering I had an oddly shaped coil with 4 layers.

( uWave psu - dead ones difficult to get here in UK due to lunatic re-cycling laws.)
My original power supply was from the flash unit in a pocket camera that I had gotten for $5 from a thrift store. I've also heard that you can pull them out of disposable cameras that they throw away at photo shops. Or you could use a voltage doubler (http://hyperphysics.phy-astr.gsu.edu/hbase/electronic/voldoub.html) on your 230v, but that introduces yet another hazard to be careful of. There should be lots of other options, since you only need a few mA of current.

Instead of more turns in same size, how about making a C core out of an old torroid transformer with the magnet in the (small) gap ?
= more gauss's to the magnet.
A while back, Olmod reported: "I had a chat today with the gentleman who knows about the magnatiser i previously mentioned,a handy thing to know is that it has removable focusing poles and they increase the power of gauss ..."

Which suggests there is room for improvement over the simple air-core coil I used. It would be a good area to explore. Your C core idea should work well, the thing to watch being that your material doesn't saturate. You want maybe 40k gauss in the air gap, and I don't know if you can do that with iron.

It would be lovely to have a magnetic circuit simulator for this. Does anyone know of an inexpensive one? Or maybe a link to some designs that have already been worked out? Or more info on the focusing poles?

Maybe that could be coupled with the half wave multi pulse from the AC supply Idea. No caps to go bang.
It seems like that should work. An interesting design project... :)

Warske

empeabee

Jan 08, 2005, 05:45 PM

What will really shoot your amps up in the simulation is to lower the capacitor resistance. With about double the capacitance, I would guess you have about half the resistance.

could be that they are low IR, 'cos they were to pulse a xenon flash tube int a 400w (output) YAG laser - all 20 of them in //ell ( 3in dia 9in long).
Sig gen (EMCO NY) was ( & may be - SI diodes on order) to provide charging current from its HT line.
My original power supply was from the flash unit in a pocket camera that I had gotten for $5 from a thrift store. I've also heard that you can pull them out of disposable cameras that they throw away at photo shops.

humm must try local photo lab as well...
Though I've just now unearthed a little strobe flash that runs off 9v - will try first that as a charger tomorrow.
Don't like V doublers - more cap's to go BANG - I've had quite enough of those since 1950 !

the thing to watch being that your material doesn't saturate
AH gotchya - Oh Flip ;)
C core - An interesting design project...
Later - ;)

Mike

Warske

Jan 09, 2005, 05:20 PM

empeabee

Jan 09, 2005, 07:20 PM

For the next round of experiments, I decided I needed a more stable fixture for measuring the strength of the 5x5x1 mm magnets.

Its just like the one in post #15, but more ridged, and it uses an LM317 voltage regulator to keep the voltage at 1.50 volts.

Warske

How wonderfully crude - In the best tradition on the mad professor - where is the black board with E=MC2 on it ? :) :) ;) :D

Think is it must work or you'd not have published.
Great.
Mike

tolladay

Jan 10, 2005, 12:30 AM

Man oh man. Scale that thing up about 500% and you could make a living as a sculptor. Take the 4th photo down, where there is nothing in the frame to get a sense of scale, and go to any gallery in LA and you could sell your "piece" in a heart beat. The 3rd photo is even better, except the stuff in the background lets you know how small it is.

One of the coolest things I've seen on the internet in a long time. Do you have hi-rez versions of those photo's (numbers 3 and 4)? I like to send them to an artist I know.

Warske

Jan 10, 2005, 02:34 AM

How wonderfully crude - In the best tradition on the mad professor - where is the black board with E=MC2 on it ? :) :) ;) :D

Think is it must work or you'd not have published.
Great.
Mike
Crude?! And I was thinking modern art!:) But maybe there's not so much difference sometimes...

Oh, and yes it works great!:D But I'm not sure I've always used that as a criteria for posting things...

Man oh man. Scale that thing up about 500% and you could make a living as a sculptor.
Exactly my point!:D

Do you have hi-rez versions of those photo's (numbers 3 and 4)?
Sure, and thanks for the complements! They're about a meg each, though, so I can't post them here. You can send me a PM with info on how to get them to you. Email might work.

-----------------------------

I wound a new coil for the magnetizer (see pictures). In case you just tuned in and are wondering what I'm trying to do, rest assured that I often wonder myself! :) But in this case, I'm trying to re-magnetize some Neodymium magnets that got cooked (literally). In a previous attempt at restoring them, I got them up to 96% strength. At this point I don't know if I just need a better magnetizer, or if the heat damaged the magnets beyond recovery. Actually, I don't really care about the magnets. They're cheap. But I want to figure out how this stuff works.

So I've been trying to design a better magnetizer. Since the last attempt, I've optimized the switch to reduce its resistance, and I have (I hope) optimized the coil in several ways. Also, since I burned out my high voltage power supply, I found a 1000 v transformer that I had lying around in the garage.

Improving the Coil

The original coil used 12 gauge house wiring with thick plastic insulation. This time I have actual magnet wire with much thinner insulation. That lets me put more turns in the same space or use thicker (lower resistance) wire. BTW, here is some interesting info on Producing wound components (http://www.ee.surrey.ac.uk/Workshop/advice/coils/index.html).

The old coil was about 0.9 inches long, and right now I'm only trying to magnetize a 1 mm thick magnet. So the coil was nearly 23 times longer than the magnet. This time I built a 1/2 inch long coil. Still overkill, but eventually I want to try olmod's idea and see if they will magnetize edge to edge. Reducing the length should nearly double the field in the center (since I'm working with the same amount of energy).

The old coil had much lower resistance than the capacitor bank. I have a theory (not proven yet) that I get the most energy to the coil when its resistance is nearly the same as the rest of the circuit. With resistance networks, it is easy to prove that the load should have exactly the same resistance for maximum power transfer. But this isn't a resistance network.

If my theory is correct, then once I choose a wire gauge to wind the coil, the length of wire needed is already determined by the resistance. If I use thinner wire, then I will need to use a shorter peice of it to get the required resistance. It turns out that using thinner wire will actually give me fewer turns.

Fewer turns might be good, but one thing it does is reduce the inductance. That increases the current, and that is hard on my flyback diode and switch. So I figure I want to go in the other direction.

With much thicker wire and more turns, the outside diameter of the coil gets large and that, from what I understand, reduces the flux coupling between turns which again reduces inductance. I really need a magnetics simulation program to nail this down.

What I finally did was take a guess. I went with 16 gauge wire, which gave me a nice four layer coil with twice the inductance that I had before.

Given the improvement in the switch, the reduced length of the coil, the reduced insulation thickness, and the optimized coil resistance, rough calculations suggest that the magnetizing force should be considerably larger than that of the first attempt.

The new coil is 6.57 uH with a resistance of 21.662 mOhms [Edit 1/10/2005 corrected typo per empeabee].
For the simulator:
lCoil 4 5 6.57u
rCoil 5 6 21.662m

The next question involves the flyback diode. As you can see from the simulation picture, the current pulse through the diode gets up to 2,000 amps. Where do I find a diode to handle that kind of current?

Warske

empeabee

Jan 10, 2005, 08:18 AM

Sure, and thanks for the complements! They're about a meg each, though, so I can't post them here. You can send me a PM with info on how to get them to you. Email might work

Put em in the Gallery

I wound a new coil for the magnetizer (see pictures).

OH - LeftBrain/RightBrain interaction - The Engineer is pushing the Artist aside ;) :D

2000Amps thru 16 g copper .. er maybe a quick calc on temp rise in the coil ?
or do it in a vacuum (spelling?) chamber and make more art with the flash deposit :eek:
Mike
<edit I * I * R = 2.4 Mega watts HUH?? only for 250 uSec - now I've run out of my maths....... /edit>

Warske

Jan 10, 2005, 11:19 AM

Put em in the Gallery
Excellent idea. I didn't know it would take large photos. Here it is:
http://rcgroups.com/gallery/showphoto.php?photo=13495

When I checked, I saw that photo 4 is just a crop of photo 3, so I just uploaded #3.
2000Amps thru 16 g copper .. er maybe a quick calc on temp rise in the coil ?
Looking at your calculations, you did them right but I see that the resistance number I posted was a typo. Should have been 21.662 mOhms instead of the 662 mOhms I posted. (I edited and corrected the previous post. Thanks!) With the new number, that makes the I squared R loss 2000^2 * 21.662e-3 = 86,648 or only 87 Kilowatts. :D

Multiplying by 250 uSec is 86,648 * 250e-6 = 21.662 watt seconds. Or a 22 watt light bulb turned on for 1 second. Maybe not so bad, but a good idea to check! With the original resistance I posted, it would have been 30 times worse.:)

Warske

<Edit 8:40 AM:

Oh, I don't think the coffee has soaked into the brain cells yet. Looking at the graph, its the diode getting 2 kA, but the coil is getting 4 kA for more like 500 uS. Let's see, thats 347 kilowatts and
4e3^2 * 21.662e-3 * 500e-6
or about 173 watt seconds. Starting to get slightly warm. :) >

tolladay

Jan 10, 2005, 12:28 PM

Excellent idea. I didn't know it would take large photos. Here it is:
http://rcgroups.com/gallery/showphoto.php?photo=13495

When I checked, I saw that photo 4 is just a crop of photo 3, so I just uploaded #3.

Thanks Warske. It's on the way to an artist I know. If you'd like to take a fairly bizarre detour from this discussion, check out his site. http://www.factory1019.com/ Click on the "Photographs" link. I should note that the content also has an adult component; not for the faint of heart or the little kiddies.

Warske

Jan 11, 2005, 04:13 PM

Back in post # 62 we saw that the new coil is going to send about 2,000 amps through the flyback diode, so its a good thing I didn't try this with the little plastic 3 A diode that I originally used. The question now is: which diode should I get?

If you go to Digi-Key (http://www.digikey.com) and search for "diode", then scroll down the page to Discrete Semiconductor Products, and select Diodes/Rectifiers
it tells you (well, it told me, anyway) "You have selected 9041 items, spanning 362 pages." Now what? Click "View page 1" then go over to "Current Rating" and click on the down arrow (sort descending).

Now you see the list sorted with the biggest, meanest, fattest diode at the top. Which one is it? Today when I checked it was a 400V, 9570A beast that sells for $506.66 in single quantities. It is 4.37 inches across and weighs 3.5 lbs. Fortunately we don't need something that big or that expensive. It handles 5 times more current than we're expecting, and although we would like some safety margin, we don't need that much.

We don't have to go very far down the list to get to the 2300 amp jobs. That doesn't seem like a lot of safety margin, but they are quite a bit less expensive than the 300 amp units just above them. The cheapest one in this category is a 1000V unit that sells for $103.87 (in quantities of 6). It is part # SD2000C10L-ND and is 2.3 inches in diameter and weighs a little more than half a pound.

But wait a second. Do we really need one this big for our application? The current ratings we have been looking at assume that the current is applied continuously. In our application, the current only lasts for a very short time.

Lets look at the datasheet. Click on the link at the left of the table, then scroll down to Technical/Catalog Information and click on the link there. Finally, click on the Datasheet (http://www.irf.com/product-info/datasheets/data/sd2000c-l.pdf) link (I also attached it below for when the link goes dead).

You should now be viewing a pdf file. Scroll down to the Forward Conduction table on page 2, and look at the section
I FSM Max. peak, one-cycle forward, non-repetitive surge current.

Note to the right that this is for a Sinusoidal halfwave.
Now look at the subsection of the table: No voltage reapplied
And the sub sub section: t = 10 mS
And we come up with a number: 23,900 A.

Now lets stop for a moment and try to translate that.

"Max." means maximum, which means that if you exceed this spec they don't want to hear about it and would claim they aren't responsible for what happens.

"Peak" means they are measuring the current by looking at the peak value on an oscilloscope, as opposed to the average or RMS value.

"one-cycle forward, non-repetitive surge current" means that we are talking about a single (one cycle) current pulse (surge current) in the forward direction (this is the Forward Conduction table, after all) and that we won't give it another pulse until the diode has cooled back down (non-repetitive). Since that's how we're planning to use the diode in the magnetizer, we seem to be in the right place.

In our application, after the capacitor discharges, it isn't going to charge back up immediately. This is the "No voltage reapplied" part. It really means "No voltage reapplied right away."

"Sinusoidal halfwave" describes the shape of the pulse. It means the first half cycle of a sine wave. The height and width of the pulse is described by the 23900 A for t = 10 mS. See the graph below and the attached .cir file

That is, for a "Sinusoidal halfwave" 10 mS current pulse, the diode will soak up 23,900 amps, even though it is rated only 2,300 amps continuous.

The current pulse in our application isn't a sinusoidal halfwave, but its a lot shorter than 10 mS. If you look back at the graph in post #62, the green line is our diode current pulse. It is only 1.5 mS at the most, and if you neglect the low current tail to the right, it is more like 0.5 mS. So our pulse is only about a twentieth as long as the pulse they tested this diode with.

So the diode we are looking at is at least 10 times more than we need (23,900 A vs 2,000 A), but before we go off searching for a smaller diode, lets stick with this for a bit and see if we can learn more about how these numbers work. For example, the spec sheet says the diode will take more current (25,000 A) if we give it a shorter pulse (t = 8.3 mS). How do we apply these numbers to our much shorter non-sinusoidal pulse?

We are about to get to the tricky bit. :)

Warske

olmod

Jan 12, 2005, 10:19 AM

That is so "war of the worlds" :D I gave up on the idea of remagnetising edge to edge neo mags as it was explained to me ,that in the mouding stage of manufacture the are held in a magnetic field to align the particles,then when cooled, and plated they are then remagnetised to their final power.but it will be interesting to say the least how much you can change it :) the magnetisers i talked about early on are from what i was told, just like a giant electro magnet with a lot of iron laminations ,and it is to those that are bolted the shaped pieces to intensify/focus the flux to the power and shape they want.
I am enjoying the thread :) lead on , cheers.

empeabee

Jan 12, 2005, 12:39 PM

yeah - I'm learning a lot here. :)
Mike

Warske

Jan 12, 2005, 12:53 PM

That is so "war of the worlds" :D
Yes, the movie version. Infrared beams if I remember correctly.:) I noticed that too after I had built it.

I gave up on the idea of remagnetising edge to edge neo mags ... but it will be interesting to say the least how much you can change it :) Yes. I'm planning to use a stack of 5 of the 5x5x1 N50 magnets to form a cube. Then I can measure the field at some fixed distance away from the N pole, attempt to re-magnetize at right angles, and then measure the field at the same distance from the (hopefully) new N pole. I think that will give a reasonable idea of how resistant they are to this. Getting there ever so slowly. Have to get the diode thing worked out first.:)

...the magnetisers i talked about early on are from what i was told, just like a giant electro magnet with a lot of iron laminations ,and it is to those that are bolted the shaped pieces to intensify/focus the flux to the power and shape they want.
If you get a chance to take a picture and post it, I would be very interested. Perhaps the few others who haven't given up on this thread would too.:) This stuf is like black magic and its hard to find information on it.

I am enjoying the thread lead on , cheers.
All encouragement is welcome!:)

yeah - I'm learning a lot here. :)
Mike
I am too! :D I never bothered to look closely at this before.

-------------------------

These hall sensors are so cheap, its nice to be able to mount them in fixtures and just leave them that way. But it was getting to be a pain to build up a way to get power to them and the signal back to the volt meter each time. So I hit on the idea of using a four wire connector and separating the sensor part from the rest of it (art sacrificed for utility). What a concept.:) The new sensor head is mounted at right angles to my original one, and will be useful for measuring the field inside coils and next to a straight wire. The hall sensor itself is very close to the top of the plastic surface mount package.

Warske

Warske

Jan 15, 2005, 12:02 AM

Back to the question of which diode. This gets into the tricky bit I mentioned earlier, and its a bit involved, so you have my apologies in advance.

To review, back in post # 66 we were looking at the specs for diode SD2000C10L-ND and noticed that this 2,300 amp part could actually handle a pulse of 23,900 amps for 10 mS, or 25,000 amps for 8.3 mS. This suggests we could actually use a diode in our application that is rated for 10 times less current than this one. That is, maybe a 230 amp diode would work instead.

In fact, our pulse (green curve on chart in post # 62) is an order of magnitude shorter than 8.3 mS, so maybe we can do even better than a factor of 10. Lets take a closer look.

According to the article Transient Peak Power Effect on Diodes (http://www.centralsemi.com/pdf/EC28semimonthly.pdf), the cause of diode failure under pulsed conditions is getting the silicon hotter than 200 C.

The thing to realize about a single short pulse is that putting a heat sink on the diode doesn't help. The silicon gets to its maximum temperature long before the heat has a chance to travel out to the diode case. What a heat sink will do is speed up the silicon's return to a lower temperature, so having a heat sink lets you safely fire off pulses at a faster rate. If you're willing to wait for a while before firing each pulse, you can just forget about the heat sink. It won't affect the maximum pulse current that the diode can handle.

To get an idea of how short the pulse has to be before the heat sink looses its effect, take a look at the chart (on p. 6 of the diode spec sheet) labeled "Fig. 10 - Thermal Impedance Z thJ-hs Characteristics"

You will see two curves, "Single Side Cooled" and "Double Side Cooled" (this is a hockey puck type diode, and it has two sides) that come together as one curve at a pulse width of about 30 mS. In other words, for a single pulse of 30 mS or shorter, it makes no difference whether you cool one or both sides of the diode, or whether you cool it at all. The article cited above explains what this curve means, but you don't need to understand it for what follows.

Another thing to point out is that the spec sheet says (back on page 2 where we got the 10 mS and 8.3 mS pulse info) that even before they applied the test pulses, the silicon was already at 180 C. Specifically, it says "Initial Tj = Tj max" where Tj is the junction temperature and Tj max is specified as 180 C. That gave them only 20 C to work with for the pulse. If we were to apply the pulse starting with the silicon at room temperature, it should be able to take more much current than indicated. I'm not using that information in what follows, but it should make us feel better about exceeding the max current specs.

So we have a pretty good idea that this diode can do even better than 25,000 amps for our pulse shape, but how much better? Its time for some number crunching.

What follows is unauthorized. I dreamed it up myself. If this design were going into hospital equipment or a Mars lander, I would need to be talking to the manufacture and get them to buy off. But hey, if this doesn't work, I'm only going to be out a few dollars, so why not try it?

The question is, how do we extrapolate the 10 mS and 8.3 mS pulse data so that we can apply it to the shorter, differently shaped pulse that we are interested in.

The approach I took is to reason that, if the silicon were a linear resistor, the power it absorbed, and thus its temperature rise, would go as the current squared multiplied by the time that the current is applied.

In fact, from fig 9 on the spec sheet, you can see that as the current goes up, the diode does look more like a resistor instead of a constant voltage drop.

Since the manufacturer gave us data on two different pulse widths, my assumption (about the current squared) can be checked. The data is for a "Sinusoidal halfwave", not a square pulse. To deal with that, I ran it in the simulator (post #66), and squared the current for each time interval of the simulation (1 uS), multiplied the value by the time interval, and summed them all together. This is the time integral of the current squared curve. I compared the results between the 10 mS and 8.3 mS curves, expecting them to be nearly the same, but they weren't. But with a bit more work, I found they were the same if I took the current to the 4.05 power. Its now possible to run the same calculations on the shorter pulse I'm interested in. The result is that 52,402 amps applied with my shorter pulse shape should stress the silicon the same amount as the maximum pulses specified in the datasheet. Note that with a bit of hand waving, I've managed to convince myself that this diode will take about 23 times the continuous current rating.

The above suggests I should be looking for a diode with a rating of around 115 amps.

As I mentioned before, with the high currents, the diode starts looking more like a resistor. One result of this is that multiple diodes can be wired in parallel and they will share the current in a friendly way.

You may know that, under ordinary conditions, paralleled diodes don't share current well. This is because they look like a constant voltage drop of around 0.7 volts, and the voltage drop decreases with increasing temperature. So if two diodes are wired in parallel and they aren't identical, one will have a slightly lower voltage and initially hog more of the current. Since it is hogging more of the current, its temperature will go up more than the other diode, and its voltage will drop even more, causing it to hog even more of the current. And so on until only one diode is carrying most of the current.

On the other hand, with the high current transient pulse, the diode looks more like a resistor and it will share the current. Also there isn't time for the feed back described above to take place.

The point is, if I find an inexpensive diode that handles less current than what is required, I can buy more of them and use them in parallel.

To cut to the chase, I went bleary eyed one evening paging through the Digikey diodes looking for something in stock that had the most bang for the buck. I came up with the 80EPS08-ND which is an 800V, 80A part in a plastic package for $5.85.

It will take 1,500 amps with a 10ms sine pulse. Applying the 4.05 power theory from above, it should take 5,851 amps with my pulse shape. This is about 73 times the continuous current rating (compared to 23 times for the previous one we looked at). It will more than handle the 2,000 amps expected.

One reason this diode seems to do so well with short pulses is that its housed in an inexpensive plastic package. If the same bit of silicon had been in a metal package, it would have been rated for higher continuous current, and then its pulse performance wouldn't have looked quite so impressive.

The spec sheet for this diode only has the 10 mS Sine pulse spec, so I couldn't double check the 4.05 number on it. I found another diode rated 85 amps (non stock and more expensive) that does have both numbers, and that one suggested amps to the 3.15 power, which is even better. However, I decided to use the more conservative 4.05 number for the calculations above.

I ordered a couple of the 80EPS08-ND and also two 70 amp SCRs (S6070W-ND). The SCRs may not hold up as well (they will be subject to 4 kA instead of 2 kA), but it seemed worth a try to replace the rat trap switch with something quieter and less aggressive.

Warske

empeabee

Jan 16, 2005, 03:03 PM

Back to the question of which diode.
.....
I ordered a couple of the 80EPS08-ND and also two 70 amp SCRs (S6070W-ND). The SCRs may not hold up as well (they will be subject to 4 kA instead of 2 kA), but it seemed worth a try to replace the rat trap switch with something quieter and less aggressive.

Warske

Rule of thumb is dead dead dead.
I look forward to the (hopefully lack of ) smoke test when the man on the pony canters up to your door. :D
Most enlightening & interesting.
Mike

Warske

Jan 16, 2005, 04:44 PM

I look forward to the (hopefully lack of ) smoke test when the man on the pony canters up to your door. :D

Well, I couldn't wait. After having done all that math, I was eager to try it out and impatient waiting for the pony to arrive. So I went to a local electronics shop that sells NTE parts and picked up a 70 amp diode and a 25 amp SCR (NTE6072 and NTE5556, see attached pdf files). They were about twice as expensive and not as beefy as the Digikey parts, but I'm into instant gratification when it doesn't cost too much. It didn't take long to wire everything together. To trigger the SCR, I used a 9v battery in series with a 100 ohm resistor (cathode (-), gate (+)).

I had an N50 magnet that had attached itself to the soldering iron and was substantially demagnetized. I tried magnetizing it, starting with a low voltage on my capacitor bank and working my way up (see attached photo and Excel file). Using the SCR wasn't nearly as dramatic as either the spark gap or the rat trap switch. In fact, there was just a tiny ping each time I fired it off. Somewhat anticlimactic.

At 135 volts the magnet appeared to be about fully magnetized, but my SCR failed shorted. I wasn't really surprised. It was only rated for 25 amps continuous, 300 amps surge, and I was planning to take it up to 4,000 amps. Even with the hocus pocus I did in the last post, I figured that might be asking a bit much. The SCR had survived the previous firing at 125 volts. If it was dropping 10 volts (just a guess), the peak current would have been 4000 * 115 / 330 = 1394 amps. Which was about 4.6 times its rated surge current.

I continued the effort with the rat trap, and got slightly higher readings out of the magnet, until at 193 volts it reached a peak. After that point, the magnet's strength actually appeared to go down slightly, but I kept on anyway. At 230 volts the rat trap contacts welded closed for the first time. I kept going to 249 volts but found no improvement in the magnet's strength. I tried it one more time at 193 volts, but never got back up to the peak reading. About that time my 3 amp capacitor charging diode gave up the ghost as well, so I took a break.

I figured something had changed. Maybe one of the capacitors failed, or the coil had arced over, or the rat trap contacts had degraded.

I started checking my parts. The NTE diode had survived through all of this. I unsoldered the capacitors and checked them individually, but they all appeared to be OK. I put the coil back on the signal generator, and it tested OK, but maybe it was arcing at a higher voltage.

This coil has 36 turns, and they should share the applied voltage between them. So at 250 volts, there would be about 7 volts between adjacent turns on one layer. But since this is a four layer coil, the first turn on the inner layer and the last turn on the next layer would have one half the applied voltage between them. In fact, there are three points on the coil where this is true.

Also, the wires could be rubbing against each other each time I fire the capacitor bank. Since parallel wires attract each other when carrying current in the same direction, but repel each other when the current is in the opposite direction, each pulse causes the coil to try to shorten its length and increase its diameter.

I hadn't used any extra insulation between layers and figured it could be arcing at one of the high-voltage-between-layers points.

Maybe its time to wind another coil. :)

Warske

empeabee

Jan 17, 2005, 12:26 PM

Well, I couldn't wait.
....
Maybe its time to wind another coil. :)

Warske
call to the Royal Society for the Prevention of Cruelty to Semiconductors
:D

?? add layers of mylar as added insulation tween layers
good work.
Mike

Warske

Jan 17, 2005, 03:18 PM

...add layers of mylar...
For some reason I'm partial to masking tape :), but mylar would be a better insulator...

--------------------

I decided to make the next coil just like the last one, except for insulating it between layers with two wraps of masking tape. To keep the wire tight while I fussed with the tape, I needed a jig (pictures).

To cut the masking tape down to the right width, I stuck the tape to a sheet of wax paper and trimmed it with a paper cutter to be slightly wider than my coil form.

When I was done, I put a final wrap of tape on the outside of the coil to hold it together. Then I removed the coil from the form and applied a layer of epoxy to the ends and to the inside. After that dried, I removed the outer layer of tape and epoxy coated the outside.

Specs on the new coil are nearly the same as the last one.

16 gauge wire
36 turns, 4 layers, 9 turns per layer
12.7 mm Length
8.03 mm ID
19.5 mm OD
6.6331 uH Inductance
23.208 mOhms Resistance

Warske

empeabee

Jan 17, 2005, 06:33 PM

For some reason I'm partial to masking tape
Warske

Whatever turns you on .. ;)
I use it as well for Q&D - was just being an engineer, not a Techie. :D
Nice jig.
Mike

Warske

Jan 17, 2005, 07:08 PM

Whatever turns you on .. ;)
I use it as well for Q&D - was just being an engineer, not a Techie. :D
Nice jig.
Mike
Its an interesting question, what to use for coil insulation. Actually, I had already built the coil before I got your post. I had been trying to think of something easy to work with that would be readily available. Having adhesive on one side helps keep it in place as you wrap. :)

Where would you suggest getting mylar for this type of thing? :)

Warske

empeabee

Jan 17, 2005, 09:23 PM

Its an interesting question, what to use for coil insulation. Actually, I had already built the coil before I got your post. I had been trying to think of something easy to work with that would be readily available. Having adhesive on one side helps keep it in place as you wrap. :)

Where would you suggest getting mylar for this type of thing? :)

Warske
Don't know how good its insulation is, but might be worth tryinjg laminating plastic - some I've seen has a lo tack adhesive then cure it with a light toasting.
Mike

olmod

Jan 19, 2005, 05:58 AM

serves me right isnt' polythene like whats used in plastic bags a good dialectric insulator? hey yer! I like your prototyping ;)

Warske

Jan 19, 2005, 10:27 AM

Thanks for the insulator suggestions.:) Maybe the next coil...

------------------

While waiting for parts to arrive, I want to explore a missing piece to this magnetizer puzzle.

The SPICE simulation tells me I will get 4,000 amps through the coil if I charge the caps to their max rated 330 volts. But it doesn't tell me how well that 4,000 amps will magnetize a neodymium magnet placed inside the coil.

To figure that out, I need to know how big a magnetic field the coil produces, and I need to know how big a field is needed to magnetize neodymium.

Lets look first at the coil. I want to know how many gauss I get in the center when I run an amp of current through the coil.

I haven't found a good way to calculate the value accurately, though I'm still looking. For example, there's a formula to use if the coil is long compared to its diameter, and that formula is supposed to be accurate to 2% if the coil is at least as long as 5 times its diameter. But my coil is much shorter than that.

Instead of calculating the value, I can just measure it. All I need is a ... flux meter! (Notice how deftly I circled back to the topic of this thread.)

The idea is to send a known current through the coil, say an amp, then stick the flux probe in the center of the coil and get a reading.

That seemed a pretty easy thing to do, and while I was at it, I also took a reading at the end of the coil. The results are:
With 1.00 amps flowing through the coil, I get
32.4 mv in the center of the coil
19.0 mv at the end of the coil

Now, though, I have a problem, because I haven't calibrated the flux meter. I don't know how many gauss it takes to give me a 32.4 mv signal.

Up to now, this hasn't been an issue, because I only used the flux meter for comparing magnets and magnetic fields.

At first all I could do was say which of two fields were bigger, and find out what direction a field was pointing.

Then (see post #13 (http://www.rcgroups.com/forums/showthread.php?p=3025907)) I found out that the readings were linear over a certain range, and I could say exactly how much bigger the second field was, without knowing the actual value of either one.

That means that if I know the value of one of the fields, I can figure out the value of the other. So I just need a magnetic field with a known value to compare my readings to.

Back in post #6 (http://www.rcgroups.com/forums/showthread.php?p=2994821), I worked a bit on calibration. For example, I looked at the possibility of using the Earth's field, but I didn't take it very far because I didn't have an accurate number for how big the earth's field is at the point on the planet where I live.

Since then I found an article Magnetic Field Calibration: Unwinding The Helmholtz Coil (http://www.conformity.com/0205iz.pdf) that would seem to be what I need, but frankly it seemed a bit involved.

I wanted something simpler. Really basic. Like a single turn coil.

The idea is, you make a circular loop of wire, pass a known current through it, and at the center of the loop will be a known magnetic field. Put the sensor there and compare its reading with the known value. It seemed like it would be easy to find the formula for the value of this field.

Looking into this, I discovered there was more than one formula for the strength of the field, and it mattered which system of units (http://www.ee.surrey.ac.uk/Workshop/advice/coils/unit_systems/index.html) you are using (SI or CGS, for example), because one had factors of 4 pi and the other had the speed of light in it. It was a fascinating digression into the history of science.

To keep it simple, I will work in the SI (http://scienceworld.wolfram.com/physics/SI.html) system of units, also known as MKS (http://scienceworld.wolfram.com/physics/MKS.html). In that system, the formula for the B field in the center of a 1 turn coil (http://www.netdenizen.com/emagnet/solenoids/ilooponaxis.htm) is:
B = u0 * i / (2 * r)

where
uO = is the permeability of air (empty space, actually). 4 * pi * 1e-7 weber/amp-meter
r = the radius of the coil in meters
i = the current through the coil in amps.
B = the flux density at the center of the coil in Teslas.

To render the above equation into hardware, I found a plastic bottle cap and tied a loop of 14 gauge magnet wire around it (see pictures). Note the centering hole in the cap, and note how the lead wires exit at a right angle to the main loop and are twisted. The idea is that the magnetic fields caused by the little loops in the twisted wire will cancel each other (the magnetic fields of adjacent loops point in opposite directions). Also, since they are at right angles to the main loop, they won't affect the magnetic field in that axis of measurement.

I measured the diameter of the loop to be 41.65 mm. Actually this is the mean diameter, or the diameter half way between the inside diameter and the outside diameter. I also measured the hall sensor voltage at the center (7.25 mV) and the current in the wire (19.49 amps). The hall sensor was biased at 1.50 volts.

Using the equation, the result was
B = 4 * 3.14159 * 1e-7 * 19.49 / (41.65e-3)
= 5.880393e-4 tesla
which (converting the result (http://www.walkerscientific.com/Products/FAQ/Conversions/conversions.html) back to the CGS (http://scienceworld.wolfram.com/physics/cgs.html) system of units) is 5.880393 gauss
and my calibration constant comes out to
5.880393 / 7.25
= 0.811089 gauss/mV

If you don't have access to a 20 amp DC power supply, you can try something like a car battery with four 60 watt headlight bulbs hooked across it. Or you can use more turns of small gauge wire and use a lower current. For example, the same size coil of 20 turns of 28 gauge wire would require only 1 amp to get the same result. You would want to wind it so the wires are all bunched together and not spread out, because the formulas for long or thick coils are different from the one above.

I should mention at this point that I have no idea how variable these hall devices are. I would guess they might be within 10% or 20% of each other, otherwise they might not work well for sensing motor shaft position, which is what they were being used for. (Anyone found a data sheet for these yet???) Since each one is (probably) different, to get reasonably accurate readings, each one needs to be calibrated.

With the calibration factor for my sensor, now I can figure out how many gauss my coil puts out.
32.4 * 0.811089 = 26.279 gauss per amp.

So with 4 kA flowing through the coil, it will create a field of
4000 * 26.279 = 105,116 gauss
which is 10.5 teslas. That's actually a quite respectable magnetic field, although it doesn't cover a large area or last for very long.

Are we there yet? No.:) The next piece of the puzzle is to figure out what this 10.5 teslas means in terms of magnetizing the neo magnets.

Warske

empeabee

Jan 20, 2005, 06:24 AM

....
To figure that out, I need to know how big a magnetic field the coil produces, and I need to know how big a field is needed to magnetize neodymium.
...
Warske

This is heading for a PhD, or at least a publishable course in magnetics.

It does seem that most publised infromation you have found are the result of wind it a then try to explain the results, and then hit it with a bigger stick.

I realy do look forward to the posting advice email.

Keep going - please.

Mike

Warske

Jan 20, 2005, 01:01 PM

This is heading for a PhD, or at least a publishable course in magnetics.
I hope I didn't overdo it!:)

I was driving by the GoodWill thrift store yesterday, so I stopped in and picked up several flash cameras for $1.99 each to replace the one I burned out. One of them only requires a single AA battery to run. I like the idea of powering my capacitor bank from just one cell.

Hint: If you decide to buy these used, take a couple AA batteries along so you can test the flash.

These could also probably be made to work as model airplane strobe lights (using a smaller capacitor), or the indicator for a low voltage lipo alarm.

By the way, if you have any stories you'd like to share about working with magnetic fields, using flux meters, etc, I encourage you to post them here. I don't think it would be off topic (after all, this is partly about magnetic fields), and I think other folks besides myself would be interested. :)

Warske

empeabee

Jan 20, 2005, 06:12 PM

I hope I didn't overdo it!:)

Warske

No you didn't /arn't.

I enjoy your lucid style, & thaught it just might make something to help yougsters to get into magnetics - not the usual chalk/bone dry text book.

Apart from a Rolex ' anti magnetic ' analog watch attaching itself to the magnet of a Very Large permanent magnet, & Rolex saying that was NOT what we said it was proof agains..
( the strap snapped as it left my arm ) & it didn't tick anymore dunno Y
:eek: :D
Mike

Warske

Jan 21, 2005, 12:40 AM

a Rolex ' anti magnetic ' analog watch attaching itself to the magnet of a Very Large permanent magnet, & Rolex saying that was NOT what we said it was proof agains..
( the strap snapped as it left my arm ) & it didn't tick anymore ...
Good story! You would think Rolex could have made good on their claim w/o worrying too much that everyone would have the same thing happen. What was that magnet being used for???

-----------------------

Anyone else want to contribute a story?

-----------------------

Back to the math (sigh). I think this is about the last of it, though I don't guarantee I can restrain myself in the future.:)

I searched for more information on how big a field is needed to magnetize neodymium, and I wasn't able to come up with anything more exact than the 20k to 45k Oersteds listed in the "Required Magnetizing Force" link in post #30.

So how many Gauss is 20k Oersteds? To use an analogy, that is like asking what is the voltage of a battery, knowing only that the current flowing out of it is 20 amps. You need more information to answer the question, because like volts and amps, Gauss and Oersteds are different kinds of things.

To answer the question about the voltage of the battery, if the current is flowing through a resistor, you need to know the resistance value of the resistor, and you need to apply ohm's law. To answer the question about the Oersteds of the coil, you need to know the "resistance" of the magnetic circuit, and you need to apply the correct formula.

To get a sense of why this is like the voltage and current, think about the following experiment (See photos. The meter was flipping between digits in the second photo.):

As detailed in post #79, I sent an amp of current through my new coil and got a flux reading of 19.0 mv at the end of the coil. Multiplying by the calibration factor (19.0 * 0.811089), that was 15.41 Gauss. If I change nothing except to slide a steel bolt into the other end of the coil, my reading goes up to 65.6 mv, which 53.21 Gauss. That is, it went up by 3.5 times.

Going back to the battery analogy, this is like reducing the resistance in the circuit, and seeing the amps go up while the battery voltage doesn't change (this is a special low resistance battery:)).

Just like the battery, the energizing force of the coil (the amps, the number of turns of wire, its size and shape) didn't change. But the "resistance" of the magnetic circuit went down when I added the bolt, and the number of Gauss (like the current) went up.

If you aren't comfortable with volts, amps, and ohms, then that whole analogy probably didn't help. Instead you could think about the water pressure to your house and the water flowing through the tap in your kitchen sink. The water pressure entering your house is (essentially) constant and when you open up the tap in the sink a bit more (that is, reduce the resistance to the water flow) you get more water flow. So if you measure the flow and want to calculate the pressure from that, you also need to know the resistance (due to the faucet) and the correct formula.

So what is the formula relating Oersteds to Gauss? This time I'm going to give it to you in the CGS system of units. Here it comes...

B = u * H

Where B is in Gauss and is called "magnetic flux density" or "magnetic induction" or sometimes the B field.

And H is in Oersteds and is called "magnetic field strength" or sometimes the H field.

And u (actually the Greek small letter mu (http://www.ibiblio.org/koine/greek/lessons/alphabet.html)) is the reciprocal of the magnetic "resistance" I talked about. Its called Permeability.

(Notice I keep putting resistance in quotes. Its because in magnetic circuits its called by a different name.)

And what is the value of u? Well in space or in air, the value is just 1 (in the CGS system).

So now we are prepared to convert between Oersteds and Gauss, but before I continue let me backpedal just a bit.

If there are any physicists reading this, they may not be too happy with me about now. I haven't mentioned the non-linearity of the steel bolt, the concept of the magnetization field and how that is used with ferromagnetic materials, or how the usual analogy between Electric fields and magnetic fields pairs voltage up to something called MMF which is Oersteds multiplied by length. Also I have ignored the direction of the various field vectors. But while perhaps misleading in its simplicity, I don't think anything I said is actually untrue. If it is, please help me out by pointing out where I went wrong. OK, back to the calculations.

Remember back in post #79, I calculated that with 4,000 amps flowing through my "air core" coil I would have about 105,000 Gauss (105 K Gauss) of magnetic flux density in the center of it? Using the above formula and the fact the permeability of free space (or air) is 1, I can now say that the magnetic field strength would be 105 K Oersted.

Recalling that I need around 20k to 45k Oersteds to magnetize neodymium, it looks like I will have more than twice as much as I really need. When the parts come in I'll do another test and find out for sure.

Just a few more things about the above formula. Notice that when u goes up, so does B. Wheras (back to the battery analogy) when resistance goes up, the curent goes down. So u is actually more like conductance (the reciprocal of resistance).

Most materials, (air, wood, plastic, concrete) don't interact with magnets and have a u that is essentially 1. Of the materials that interact with magnets most have a u that is greater than 1 and they are attracted by magnets. Some, called diamagnetic, have a u of less than one and are repelled by magnets. Diamagnetic materials can be used for unpowered magnetic levitation (http://www.scitoys.com/scitoys/scitoys/magnets/pyrolytic_graphite.html) demonstrations.

Warske

empeabee

Jan 21, 2005, 08:33 AM

Good story! You would think Rolex could have made good on their claim w/o worrying too much that everyone would have the same thing happen. What was that magnet being used for???

It was a ' Fork Relay ' - 1920's technology ultra sensitive relay used for submarine cable telegraphy - way before electronics, in-cable repeaters and satilites. A quite gob-smacking solution to the problem of getting signals across the atlantic using a single wire and DC keying. ( 4 the teckies typically 3000 ohms long with a distributed capacitance of 300 microfarads. It could make a thread of its own - old pre electronic technology solutions)

Back to the math (sigh). I think this is about the last of it, though I don't guarantee I can restrain myself in the future.:)

When taken in small quantity math is fine - like ??nitroglycerine?? pills for weak hearts :)
keep going...

If I change nothing except to slide a steel bolt into the other end of the coil, my reading goes up to 65.6 mv, which 53.21 Gauss. That is, it went up by 3.5 times.

small nit pick - soft iron (bundle of garden wire) would be better.

If there are any physicists reading this,....

you just put them out of business - 3 years of collage explained in one post ;)

Warske
Thinks - adding your bolt to the pulse magnetizer could result in the rail gun from Quake :eek: :cool:
Mike
<edit dyslexia check>

tolladay

Jan 21, 2005, 10:06 AM

empeabee

Jan 21, 2005, 11:04 AM

Rail Gun! Did somebody say rail gun? And just to be picky, if I were going to start with the sources for a rail gun, Quake would be at the end. I think the US Navy would be near the start, and any of a dozen wonderfully written sci-fi books. Not that it is a topic I think we should discuss at depth in this thread; it's not exactly a safe toy I would think.
Well I thaught 90% of the computerLits might know Quake..& didn't the B.I.S. sugest it as a cheap orbital launcher?

And It definitly IS NOT SAFE - thats why I put the thinks in - remind viewers that that could be a side effect of boosting the field strength that way

- I remember trying it as a boy in the late '40s with dads car battery a nail and a soda straw - didn't work as I couldn't control the pulse lenght ( thank the lord ).
Mike

DanC

Jan 21, 2005, 11:48 AM

just a thought and I may be talking out of my posterior I think you are well on the way to provide the electrical energy to generate the magnetic feild but you actual coil looks like its not the best solution.

If you think of magnetic circuit similar to an electrical circuit I thought airgaps were considered as impedances so looking at this maybe this will help you generate higher magnetic fluxes for the same electrical energy

get a small transformer strip of the windings and then hacksaw through the laminations a gap so that it a very tight squeeze to get the magnet in .
the laminations should look like a "C"

then wrap around the remaining laminantions the thickest wire you can maybe make a single conductor out of three conductors twisted together to reduce internal resistance and impedance

this is all just a thought based around my own interest in electical motor design where the airgap in the magnetic circuit is said to reduce effiecency

empeabee like the avitar. Is that the plane that was designed in the UK with an all moving tail that and design was given to the US in order for them to have a go at breaking the sound barrier.

empeabee

Jan 21, 2005, 02:53 PM

empeabee like the avitar. Is that the plane that was designed in the UK with an all moving tail that and design was given to the US in order for them to have a go at breaking the sound barrier.

Yes - the Miles supersonic design.
The 1/4 scale rocket powered model exceeded Mach 1 just two weeks after the project was cancelled ( full size 90% complete, afterburner jet engine fully tested).
- some time B4 X1 was built.
Oh BTW - the all flying tail was Chuk Y's own design - acordning to Discovery Channel. ;)

Mike

Warske

Jan 21, 2005, 05:28 PM

It was a ' Fork Relay '
I found a Fork Relay (http://www.porthcurno.org.uk/html/object12.html) link which describes it. I told my wife your story and she thought it was great!

soft iron (bundle of garden wire) would be better.
I agree. And the separate wires would keep down the eddy currents. The bolt was handy, what can I say?:) Glad you pointed it out.

3 years of collage explained in one post
Slight exaggeration. Only one year, I think :)

rail gun from Quake
Whenever I search the web for info on magnetics, I keep tripping across sites describing rail and coil guns (http://www.google.com/search?hl=en&q=coil+gun+capacitor&btnG=Google+Search), can crushers (http://www.google.com/search?hl=en&lr=&q=can+crusher+coil&btnG=Search), and quarter shrinkers (http://www.google.com/search?hl=en&lr=&q=Shrinking+quarters+joules&btnG=Search). Some of them are a bit weird. Two of the better ones (lots of useful info) seem to be The Magnetic Gun Club (http://mgc314.home.comcast.net/index.htm) and Barry's Coilgun Design Site (http://www.oz.net/~coilgun/home.htm). Both mention FEMM, which is the sort of magnetics simulator I had been looking for. I just haven't had time to test it yet.

Rail Gun! ...not exactly a safe toy I would think.
This from the fellow who said "You talking to a guy who is the proud owner of 'the Anarchist Cookbook'":D

get a small transformer strip of the windings and then hacksaw through the laminations a gap so that it a very tight squeeze to get the magnet in .
the laminations should look like a "C"
I agree 100%, and Mike (empeabee) mentioned that also. My concern is that the iron will saturate before we get enough flux in the air gap. But I don't know that will happen and I don't want to stifle this line of thought. Olmod mentioned focusing poles (I think) which is probably along similar lines. The question is how to design such a thing without a lot of trial and error. Maybe the FEMM simulator (mentioned above) would help.

Do you know if its possible to get 45k Gauss in the air gap w/o saturating the core material?

-------------------

Speaking of things "not exactly safe," I took apart the camera that used a single AA battery, and attached some pictures of the process.

Disassembly was straight forward:
(1) Remove all visible screws.
(2) Remove parts
(3) Back to step (1) or stop if done.

The resulting flash unit comes with a nice on/off switch and an indicator light that blinks at 277 volts and stays on at 285 volts. It regulates the capacitor voltage at 315 volts.

It takes 7.9 seconds to charge its 120 uF capacitor, but a little over 4 minutes to charge my 3,000 uF capacitor bank.

Warske

empeabee

Jan 21, 2005, 06:07 PM

I found a Fork Relay (http://www.porthcurno.org.uk/html/object12.html) link which describes it. I told my wife your story and she thought it was great!

Good Grief - I did my training there in 1954 - Heap much strange feelings. BTW the Fork Relay is about 2 ft high, & a lighning strike in the sea 3 miles from the station could wreak the coil suspension ( .002 inch Phosphor Bronze wire ) - Sigh how things have changed ( 4 the better - I think).
Thanks a lot.

I agree 100%, and Mike (empeabee) mentioned that also. My concern is that the iron will saturate before we get enough flux in the air gap.

Hey your current method works - if it aint broke dont fix it :)
( unless you ARE going 4 a PhD :) :) )

The resulting flash unit comes with a nice on/off switch and an indicator light that blinks at 277 volts and stays on at 285 volts. It regulates the capacitor voltage at 315 volts.

umm with several of those, a couple of mirrors ( one 95% other 100%)
and a small Yag rod, you have the makings of a pulse laser - the flash tubes will pump the Yag rod & voici coherent IR light...

Warske

Warske

Jan 25, 2005, 09:44 AM

I took my own advice and stopped in at a One Hour Photo processing place to see if they would give me some disposable flash cameras. The answer was yes. I left with 11 of them and an offer to come back any time for more.

The two that I took apart didn't have voltage or capacitance labels on the capacitors, but they both charged their caps up to 330 v.

Warske

tolladay

Jan 25, 2005, 10:27 AM

I like the sound of free

empeabee

Jan 25, 2005, 11:01 AM

I like the sound of free
That puts a whole new twist on the term voltage tripler ..
r u sure u don't have famly in translavania :) . :D
Mike
<edit woops that was supposed to be 4 Warske - slip of the mouse >

sawicki

Jan 25, 2005, 11:08 AM

The two that I took apart didn't have voltage or capacitance labels on the capacitors, but they both charged their caps up to 330 v.

Warske

I'm sure their will be differences. But here is a schematic of one that was reverse engineered several years ago. These circuits are often used for low cost CD firing boxes for fireworks. Pyros tend to string a lot of electric matches together in series and fire them over long distances of wire.

Warske

Jan 25, 2005, 12:18 PM

...here is a schematic of one that was reverse engineered several years ago.
Thanks for posting the schematic. Very nice! Most of the flash units seem to use a neon indicator light as shown. Two of my freebees used an LED. There aren't many parts in these things, but I'm still amazed they can afford to just give them away.

...electric matches together in series and fire them over long distances of wire.I hadn't heard of electric matches. It looks like that might be a bit of resistance wire wound around a match head. Is that about right?

---------------
Measuring capacitance

I've been assuming that the WinSpice simulation of the magnetizer is accurate, but its only as accurate as the data I put into it. For the capacitance value of the electrolytic capacitor bank, I had used the value marked on the capacitors: 750 uF times 4 capacitors equals 3,000 uF.

These electrolytics are usually manufactured to fairly wide tolerances, with +50% to -20% being a common range. Used as power supply filter capacitors, more is better so most folks don't care about the +50% on the high side. The tolerance isn't marked on mine, but if I use the above numbers, I could have anywhere between 2,400 uF and 4,500 uF. That's been bothering me for a while because it will affect the results of the simulation. So I finally decided to measure the value.

There are lots of ways to measure capacitance. I opted for an inexpensive method, although its a bit time consuming.

With a large value of capacitance like this, its possible to get a good reading by charging the capacitor up, then using a stop watch to time how long it takes to discharge to a ceratin voltage through a known resistance.

The math is simple because a capacitor has the interesting property that, starting from any voltage, when you discharge it through a resistor it will reach 36.79% of that voltage after one time constant. The time constant (in seconds) is just the resistance value (in ohms) multiplied by the capacitance value (in Farads). That is
time = R * C

For example, if I really have 3,000 uF and I discharge it from 300 volts to 110.37 volts through a 54 k ohm resistor, it should take 162 seconds (54e3 * 3000e-6).

Or I can measure the time and solve for C in the above equation:
C = time / R

In making these measurements, there are a couple other things to be aware of:

Leakage resistance

Electrolytic capacitors, like batteries, will eventually loose their charge if left to sit long enough. You can model this effect with an ideal capacitor in parallel with a resistor, and its value can be called the leakage resistance, since the charge leaks off through it.

For accurate time constant measurements, you want the leakage resistance to be high compared to your test resistor. That way the leakage resistance can be ignored because if a high resistance is in parallel with a low resistance, the combination of the two is very close to the low resistance. Leakage resistance changes with temperature and with how long the capacitor has been charged up. Capacitors that have been discharged for a long time have more leakage, and to minimize it, you can run them at their operating voltage for a few hours. Camera flash power supplies work well for this. :)

I did this with my capacitor bank, after making sure it was disconnected from the rest of the circuit. Then I let it sit for a little over 12 hours. It self discharged from 313 volts down to 209 volts. This works out to about 29 Meg ohms of leakage. Note that if I had left my (10 meg input resistance) digital volt meter hooked up the whole time, it would have discharged much faster. Note also that even after 12 hours there was still a potentially lethal charge on the caps. With some capacitors, this charge can last for days or weeks, and is the reason you want to discharge them before handling them.

With a test resistance of 54k ohms, the combination of this and the 29 meg leakage resistance is 1/(1/54e3 + 1/29e6) = 53.9 k ohms, so we can see that the leakage resistance isn't going to affect our readings by much.

Resistor power dissipation

We want our test resistor to stay cool so that we don't have to worry about its value changing as it heats up. At 300 volts, the power is V^2/R or
300^2/54e3 = 1.7 watts.

For the resistor to stay cool, its rated power should be much more than this. I used two 10 watt 27 k ohm resistors in series for a total rated power of 20 watts.

Results

With the test resistors hooked across my (10 meg ohm) volt meter, I measured the total test resistance and got 53.45 k ohms. Then I charged the capacitors up to 330 volts and attached the test resistance. I started the stop watch when the voltage hit 300, and stopped it when it got to 110.4, and came up with 210.75 seconds.

C = 210.75 / 53.45e3
= 3.942937e-3

which is 3,943 uF

More equations

By the way, if you want to know where the 36.79% number comes from, it is just 1/e, where e is the base of the natural logarithm (http://en.wikipedia.org/wiki/Natural_logarithm) and is approximately 2.7182818. The full equation for the voltage Vc on a capacitor after time t, starting at voltage Vinit is

Vc = Vinit * e^(-t/(R*C))

So when the time is equal to R*C (that is, after one RC time constant), then (-t/(R*C) is equal to -1 and e^-1 is the same as 1/e

You probably knew I couldn't stay away from the math.:)

Warske

tolladay

Jan 26, 2005, 01:22 AM

olmod

Jan 26, 2005, 02:06 AM

that made me laugh out loud :D im sure many will get a chuckle out of that :D keep up the good work, cheers.

Warske

Jan 26, 2005, 07:22 AM

A little asomething I cooked up for an article on DIY brushless motors soon to be seen in Quiet Flyer... Or, maybe this is Warske's alter ego?
Ahhh... Looks like a Wimshurst Machine (http://physics.kenyon.edu/EarlyApparatus/Static_Electricity/Wimshurst_Machine/Wimshurst_Machine.html)... I had one of those when I was a kid! :D But this one has a subtle brushless motor motif. Nice picture editing job! :)

Warske

empeabee

Jan 26, 2005, 07:26 AM

Here's an example of what NOT to do with your brushless motor. :D

A little asomething I cooked up for an article on DIY brushless motors soon to be seen in Quiet Flyer...

Or, maybe this is Warske's alter ego?

Oh dear - I need to see a physiotherapist to get me back working after seeing your pix - fantastic made my day.
Mike

Warske

Jan 26, 2005, 10:07 AM

The Digikey diodes and SCRs arrived, and I ran some more tests. I replaced the NTE diode with the two new 80 amp 80EPS08 diodes wired in parallel (see attached data sheet), and replaced the rat trap switch with one of the 70 amp S6070W SCRs (the datasheet wouldn't fit in the 100k attachment limit, but can be found here (http://rocky.digikey.com/WebLib/Teccor/Web%20Data/SCR%20(%201%20A%20to%2070%20A).pdf)).

Flyback Diodes
I used both diodes because I was starting to get a little worried about the reverse voltage on the caps. Information on the web seemed to suggest that the maxim allowable reverse voltage was around 1 to 1.5 volts, but it seemed fairly vague. Someone playing with coil guns reported seeing 100 volts of reverse voltage, apparently without any damage to his caps.

With one diode, the reverse voltage is about 5 volts peak (based on the simulation). Two diodes in parallel limit the reverse voltage to about 3 volts.

Since then, I found more detailed info (http://www.cornell-dubilier.com/tech/appguide.pdf) from Cornell Dubilier:
"Application of reverse voltage much beyond 1.5 V causes high leakage current quite like the forward conduction of a diode. Neither of these operating modes can be maintained for long because hydrogen gas is produced by the capaci-tor, and the pressure build up will cause failure."

In my application, the reverse voltage lasts for only a millisecond, and the associated energy is very low. So now I'm back to thinking that one diode would work, but I haven't bothered to remove the extra one.

Rat Trap Switch vs SCR
The rat trap switch has lower resistance and lower voltage drop than the SCR, but it tends to get bent out of shape (literally) and is somewhat obnoxious to use. The SCR works well enough and makes it possible to run tests without bothering the cats, so I'm using the SCR when I can. To trigger the SCR, I used a 100 ohm resistor, a 9v battery, and a micro switch.

Revised simulation
I updated the simulation (see picture and attached .cir file) to include the measured value of the capacitor bank, the new coil inductance and resistance, and the flyback diode resistance.

The max coil current went up a tad to 4,088 amps. The max flyback diode current went down to 1,249 amps.

Next: test results.

Warske