RC Groups

RC Groups
    Micro EDF Jets and Power Systems
        Build Log 3'' Nano Squall (10mm EDF, Single Phase Sensorless Motor)

#1 Chris T Mar 03, 2012 01:50 PM

3'' Nano Squall (10mm EDF, Single Phase Sensorless Motor)
 
6 Attachment(s)
Hi guys,

here is my project for this year. After successful completion of the 6'' micro squall last year

http://www.rcgroups.com/forums/showthread.php?t=1397595

I decided to try pushing the limits a but further :) I can't promise if this will work out, but I have a complete plan in my head, so I have to go for it.

What you see here is the first attempt to assemble an edf unit with a 10mm printed fan (shapeways). Special thanks to David Nitrocharged for the great fan design!!! The magnet specs are D3mm x 1mm x 2.5mm, N50. The complete unit with fan weighs something between 0.3 and 0.4 g. Unfortunately, the bore is quite off center, actually I have not seen any symmetric magnet of any size yet, so I might need to redo a magnet with an off centered brass spacer as I have done before. I have to get used to machining at this scale first and gather some experience. I might need some new drills and reamers too. My first impression though is that I will be able to handle that size.

Meanwhile I have started to design a scaled down version of the 6'' squall electronics with much smaller rx chip (SI4311) and smaller output stages. But one step after the other...



Cheers

Chris

#2 pdawg Mar 03, 2012 02:02 PM

Insanity! If anyone can make work you will Chris!

#3 hanzie Mar 03, 2012 02:19 PM

NO...WAY...! Even smaller?!? omg! :eek::D:eek::D:eek::D (Pretty much switching between these 2) Nice! How do you even get electronics this small?

I'll be following this one for sure!
Hanzie.

#4 NitroCharged Mar 03, 2012 03:39 PM

He's back in the game! :D

If this works, we'll have our first true "Pocket Rocket"! :cool:

:popcorn:

#5 Chris T Mar 03, 2012 05:00 PM

Ok, the motor is running on the existing controller with the FDC6327 bridge. It's terribly out of balance as expected due to the off centered magnet. Anyway, the first try and a decent air stream I can feel behind the fan. It's looking promising :D

Chris

P.S.: I forgot to mention a very important point: The fan did NOT disassemble! Something we weren't sure about printed fans was their strength. It seems to be fine. And rpm seem to be 100k+. Very astonishing, I was expecting it to be lower. I have to confirm with the scope. The voltage with my big test battery must stay up about 4.1V with this small motor, which might explain a higher rpm than in a flight setup. This is starting out very exciting!

#6 Kevin Greene Mar 03, 2012 07:52 PM

All I have to say----WOW!!!...You go, dude!!!...Awesome project and paving the way for others...Keep us posted!!!...Again, WOW!!!

Kevin

#7 toxic-trip Mar 06, 2012 10:18 AM

HOLY @#$%, if you were to ever market this, kids would be able to put them in their lunch boxes and fly at school anytime, and adults could hide them in their empty coffee cups to fly at work.

#8 Chris T Mar 06, 2012 03:59 PM

Quote:

Originally Posted by hanzie (Post 20926282)
How do you even get electronics this small?

I'll be following this one for sure!
Hanzie.


Hey,

actually, the electronics is the one thing that won't scale down as much as the rest, but there is no need, since the weight of the 6'' squall electronics was almost negligible. I am trying to get the size down by building servos without extra driver, running the coil directly from the atmel and using the former servo bridge A3903 for the esc this time. Also, as mentioned the SI 4311 should bring the size down. I am also checking an alternative, we fooled around with some rf detector chips years ago, they might be ok for this size. Don't remember their name but I should be able find the threads. The SI4311 is the proper solution though.

So I managed to build a centered magnet and the motor was running very silently this time. The winding is very hot, seems like I get reasonable rpm on half throttle. Can't believe I got so far on the most critical unit within such short time. Building the motor half size doesn't seem to be harder from the mechanical point of view. Actually, building goes faster, since less machining is required. So it looks like it is worth finishing the board designs :)

Chris

#9 Joe Malinchak Mar 07, 2012 12:33 AM

Chris,

Your 10mm EDF fan unit is looking great! I can't wait to see how much thrust you are getting out of the little motor.

Joe.

#10 devmonkey Mar 09, 2012 02:43 PM

Sensorless ESC design
 
Hi Chris,

Your work on the 6" was amazing, this is taking it to a new level!

I'm about to build 4 single phase micro motors for a project, I'd like to control them all with a custom controller but I cannot for the life of me work out how you sense the back emf across the coil without floating at least one end which you can't do without floating each half of the bridge which requires each of the 4 fets to be switched independently (i.e. 4 gate lines to the uc).

I know you had similar questions in the mg motor thread, did you ever get to the bottom of it? To me it just looks like the coil will be shorted when trying to measure the back emf yet obviously it seems to work.

Would you be willing to provide a sketch of the switching/sense side of your esc schematic?

Cheers, Joe

#11 Chris T Mar 10, 2012 05:38 PM

2 Attachment(s)
Hi,

I did some work. I built a new motor, since the first one kind of melted at that crazy power level. The CA that held the coil together became soft and the coil lost shape finally. That was also the reason the motor was running silently I believe. The was no solid connection to the stand in the end :D I bit of balancing the fan is still to do it seems...

However, the motor was wound way too hot, that was obvious, so I used .1mm instead of .15mm wire this time. That's all material I have right now in this range, so I need to get supplied as soon as I know the direction I am heading. I also figured that building the coil is not that easy. The thinner the wire gets, the harder it is to get it of the mandrel without being glued already. This is how the 6'' version was done. Winding, taking off the mandrel and then appliying CA to the flying coil. It doesn't work for this small version, so I CA'ed the coil on the aluminum mandrel this time and had a hard time getting it off. I finally found a way to knock out the center piece, but I haven't checked if the coil survived it. I am quite sure that technique will be successful in the end. I need to find something like wax to coat the mandrel before winding so the coil doesn't stick too much. I need to make a photo of the great mandrel design, but I am not home right now, so I will do this later. Learning how to manufacture the bushings too. Good material I got from Dan a while ago :)

The best thing I have done lately is the new design of the rx/esc board. It came out at 9.3 mm x 4 mm double sided. One side is rx the other is esc, so it can be separated too. I am really happy how it came out. It is about as amall as it gets with these components. Hope that all bugs are sorted out. No experience with that SI4311 rx chip yet, but it looks fine to me. This board is considerably smaller than the 6'' servo board, which was already tiny. I need to order those new components quickly and get a prototype assembled. It's important to have a bridge with driving power in the right order of magnitude for these small motors too. ;)

Can't wait :D

Cheers,
Chris


P.S.: DAmn, I forgot the sense lines ;) I will get back to that topic and explain... Need to fix the board first. Size will increase :(

P.P.S.: Fixed. Same Size :D Later on the train I will have time to explain the sensing. I am close to open sourcing everything ;)

#12 forsakenrider Mar 11, 2012 07:30 AM

Very amazing Chris! I wanted to try building some of these little motors but it was too hard to get the right bearings, bushing material, and motor shaft up here. Maybe I didn't look far enough out of my igloo though. Either way, I'm following along with great interest!

#13 Kevin Greene Mar 11, 2012 11:32 AM

Chris---I've never built one of these motors so maybe what I'm about to suggest may seem crazy to you...Have you thought about using for the mandrel some very hard/stiff Nylon rod material???....You can easily cut this material down to whatever diameter you like with a lathe...Being Nylon, the coil should easily slide off...What do you think???

Kevin

#14 Chris T Mar 11, 2012 04:45 PM

1 Attachment(s)
Hi devmonkey and others interested in the sensing,

the idea is actually very simple. The trick is to switch both outputs of the driver to high impedance during sensing, which is when the back emf is doing the zero crossing. I am using four signals to be able to drive the H-bridge this way. With the A3903, this should be possible with two signals, since it is doing some pre-processing. The motor is driven with one positive and one negative pulse per revolution. The pattern has a high impedance phase after every pulse while the zero crossing is happening. The sensing is activated not right after switching off the driver but after a certain dead time to wait until the ringing is over, otherwise, you will detect a bunch of wrong zero crossings right switching off the driver. That time is quite short, just a fraction of the revolution. Best duty cycles I found in terms of efficiency are around 70% (30% high impedance time). Above that, the timing becomes critical and the efficiency goes down while rpm doesn't rise any more...

Am I getting this clear? Feel free to ask me questions...

Cheers

#15 devmonkey Mar 13, 2012 01:56 PM

Quote:

Originally Posted by Chris T (Post 21004846)
Hi devmonkey and others interested in the sensing,

the idea is actually very simple. The trick is to switch both outputs of the driver to high impedance during sensing, which is when the back emf is doing the zero crossing. I am using four signals to be able to drive the H-bridge this way. With the A3903, this should be possible with two signals, since it is doing some pre-processing. The motor is driven with one positive and one negative pulse per revolution. The pattern has a high impedance phase after every pulse while the zero crossing is happening. The sensing is activated not right after switching off the driver but after a certain dead time to wait until the ringing is over, otherwise, you will detect a bunch of wrong zero crossings right switching off the driver. That time is quite short, just a fraction of the revolution. Best duty cycles I found in terms of efficiency are around 70% (30% high impedance time). Above that, the timing becomes critical and the efficiency goes down while rpm doesn't rise any more...

Am I getting this clear? Feel free to ask me questions...

Cheers

Thanks Chris. Do you think this would work with a 2 line bridge switching the gate lines to input during sensing?


All times are GMT -5. The time now is 01:05 PM.