Homebrew Rondo!

[KreAture]

I saw the Rondo for the first time and decided it was nifty...

Then I got sick and couldn't go out and fly. I needed something to do, so I designed a Rondo Ripoff!


http://www.unixcore.com/~kreature/images/rr/rr300_6.jpg
Preliminary layout.

As you can see, it is round... As you can also see, it is in development. I have etched a prototype and it is looking promesing.

I will have to wait untill I get my hands on some real live speciments of the chip I intend to use for it to actually fly anything, but I have been testing the firmware in a equivialent chip, although to large to fit on the PCB.

Features are:
* Speed-controller with 8-bit PWM.
* Max 14A(theoretical), 10A max to be safe.
* 1A BEC
* arming-function
* auto-calibration

Status is:
- Working and running smoooooooth!

Current focus / todo:
- Add more decoupling to improve noise-immunity
- Get ahold of a LM2940 voltage regulator in SOT-223 case to compare against the lousy LM340 I have to use now.

And, before anyone asks: NO I will NOT be adding a breake, nor will I be adding current-limiting. The FET is fairly cheap and will burn if one short out motor. This can be your protection.
As for breake, use this ESC for fun stuff like flying wings or other placed u can stick a speed 280/300 motor...

Oh, and for those wondering: I will be using the Microchip PIC 12F629 in this ESC. It is only 8 pins, has flash-memory and costs less than $2 in singles!

[Trizza]

Nice work - do post pictures please!

[KreAture]

Will even post a vid of it running a speed 300 on my wing :)

[Sloper Mario]

Well done! Amazing what a little boredom can accomplish eh? :D

Mario

[KreAture]

New RR300 layout is done!

http://www.unixcore.com/~kreature/images/rr/rr300_9.jpg
I've upgraded my CAD program so that I can rotate parts in any angle! I think the new layout is better as it has better connections for the RX-wires. Also, I used a new function to create arched tracks along the outer-rim.

There is decoupling-capacitors on V+ near the connection to RX and the microcontroller. There is also one over the motor together with a emf-diode. There is also the usual 2 capacitors on the regulator. I did add another larger capacitor on the V+ rail close the the regulator. This should make the system a lot more immune to noise.

The previous version could run with a LM340 regulator all the way down to 5.1 volts with a speed 280. The new version should get lower and if you can get ahold of a LM2940 is SOT-223 package, this should reach noise-immunity down to 3.6v ! (Wich would mean I need to add a LVC in order not to kill your battery's if you are stupid.)

[KreAture]

Here's the new pcb almost done. Only thing missing is the actual microcontroller. I will be getting that tomorrow and then I will go flying!
http://www.unixcore.com/~kreature/images/rr/rr300_14.jpg

[Seared Ice]

Hey Kreature

Do you have that mounted on the motor? Good idea.

Ever since I started seeing you work on these electronics I have wanted to learn also. I'll be getting a soldering iron (i think this is a good first step...) soon just for regular hobby needs. My highschool offers no courses on working with electronics like this, the robotics club is more about building the robot (they buy pre-made robot control electronic boards), and the "tech" class is more about pneumatics and stuff....

maybe i can get a course at the local college over the summer or buy a few books or something

~John

[KreAture]

Books are cool.

[zagisrule!]

Hey Kreature, you might want to add a hole to strain-relieve the negative battery wire, it looks like it could pull the copper right off the board, especially if it is in a pusher, where a crash would pull that connection away from the board.

[Stickman]

Kreature,
Nice work. You are very close, but might I suggest one more detail that I feel will help the cause. I see you added as much copper fill as you could to your primary heat disipating components, namely the FET and the regulator. This is good. What concerns me is the height and mass of C4. The leads on these surface mount electrolytics won't take a lot of lateral G's, especially when mounted vertically...don't ask me how I know. It's one of those mechanical moment thinguses ME's know about...out of my range. If you would like to try some small tantalum brick types for C4 instead, PM me, I got many to spare.

Another trick for heatsinking prototype stuff is to use a doublesided board and leave as much copper on this otherwise unused side as possible. Copper tape can make a double-sided board in a pinch as well. Then simply attach copper braid to the tabs, wrap over, and solder to the other side. Remember though to watch the polarity of your tabs and planes from one side to the other side--no shorts allowed. A +5 Vdc regulator tab inadvertently conected to the drain or source of the driver FET would get ugly. In that case, you would need to separate the backside copper to connect both.

FWIW, Very nice work!

Mark

[KreAture]

zagisrule! I have thought about that as well... I don't think a hole will do anything as I am using 0.8mm thick board.
The jolt would snap the board instead. M+ on the other hand is actually soldered to the motor tab so it should be safe :)

Stickman, I am testing different layouts and ideas. That capacitor was already replaced the same night as I etched that board.
I am now using 2 large tantalum at the regulator, 1 smaller cheramic at the V+/GND outputs to the tx (this also happens to be close to the microcontroller) and a 0.1 uF cheramic over the motor poles.

I have also split the return-path of the emf between the motor pins so that it now goes through the diode and cap and meets with the regulators large tantalum instead of joining up with the regulators feed pin before the caps. This should reduce the noise a bit too.

When it comes to heat, there is no real issue with the FET as it has sooooo low internal resistance it can easily handle 10-14 amps without a heatsink. The FET is a logical-level controllable HEX-FET with only 4.7 mOhm on reistance and a 0.6v threshold.
at 90% duty it does NOT even get luke warm running a speed 280 on 8 cells.

The regulator on the other hand leaves to be seen. I don't think there will be any issues but the regulator has both short circuit and temp protection and essentially can NOT be distroyed.

I have tested the circuit a good deal now and it can run a 8-cell battery down to about 5.1volts before the BOD kicks in and stops the controller. At this time there isn't really any power left anyway and if you didn't notice thrust dissappearing waaay earlier than that U deserve to crash.

The new layout isn't as pretty but looks like this:
http://www.unixcore.com/~kreature/images/rr/rr300_15.jpg

I do have room for some small 0805 or smaller type capacitors very close to the vreg pinns and will probably add that.

Also, I will add another cheramic up by gnd/V+ pads, simply because there is room :)

[vintage1]

Originally posted by KreAture


When it comes to heat, there is no real issue with the FET as it has sooooo low internal resistance it can easily handle 10-14 amps without a heatsink. The FET is a logical-level controllable HEX-FET with only 4.7 mOhm on reistance and a 0.6v threshold.
at 90% duty it does NOT even get luke warm running a speed 280 on 8 cells.

The regulator on the other hand leaves to be seen. I don't think there will be any issues but the regulator has both short circuit and temp protection and essentially can NOT be distroyed.



Hmm. I'd rather have a blown regulator than a crashed model :)

Mind you, if it goes, so does the model anyway :)

[KreAture]

It will only limit current... It won't CUT.

[zagisrule!]

OK,
I see, I have just had some experience with pads coming off and it really peeves me when it happens. I saw that the + lead is going straight to the motor tab, good strain relief and as little as possible resistance.

What MOSFET are you using? You could easily do a 400-480 controller with a TO-252 FET. Adding a brake is not all that hard either, it is just a P-chan MOSFET that pulls the negative motor wire to Vcc to short the motor. You just need to add a code segment driving inverting P-chan driver high on a no throttle
setting.

[Stickman]

KreAture,
Looking good....a good single sided layout--low noise, etc-- is often challenging to achieve. It really would be nice to have the luxury of a full ground plane on the other side wouldn't it--on the other hand: boring!
I'm breathing easier now, now that the electrolytic went bye bye.
Could always try reducing the 0.1 uF across the motor terminals to 47 nF with an additional 10 to 1 nF in parallel with it--might filter the higher frequency noise better.

Thanks for sharing your efforts.

Mark

[vintage1]

Check out self resonance of suppression capacitors from manufactureres data sheets. Often bigger capacitor is worse at RF freq's.

[KreAture]

I will be experimenting a lot with noise suppression. Thanks to my 100 MHz scope I can see how bad it is and what change occurs with my different counermeasures. Makes it very much easier to achieve good suppression.

It is already rather good actually. No noise issues on a speed 300 even running off 8 cells. It won't be especially usefull on 6 cells for now though as I can't get ahold of the LM2940 5v version on SOT-223 package. (Nothing larger will fit)

I am also getting some ATtiny-15 MCU's to test with. Will write a new program for those. They are same size as the PIC12F629 but have one advantage: Hardware PWM!
This will remove all jitter from PWM and NOT cause jitter in the sampling routine. Thus making a far better controller.
There is NO 8 pin MCU from Microchip that has hardware PWM, so I have to try the Atmel.
Only downside is that it costs 4 times as much :( The PIC12F629 is less than $2!

[vintage1]

All sounds rather good. If you have enough code space, do do the nice things like auto endstop election on switchon, and a nice (subjectively) linear throttle response. I'd also like to NOT see teh 'feature' applied to some controllers whereby if they lose RX pulses for more than half a second they reset to power off. Make it more like ten seconds. !!!

One think that is relevant to throw in the pot is the one and only 'rondo' style controller I had, went mysterious on me after a crash. It was a multplex unit. It *seemed* to fire up at exactly hjalf teh clock frequency it had before it crashed, leastways the 'singing' was an octave lower, and it only went to half throttlle ...:D...these things get a pounding when strapped to the motor rear.

If teh parts are easy to come by, and the chips are programmeable with just a PC, serial lead and a bit of free softare, I'd be interested in supplying these to UK modellers..

What's the current parts cost? My guess is under $10?

[Nobert]

It looks like you are using a N channel TrenchFET. Suggust a SI4442DY. Has .005 on resistance and can handle 22A max. Looks like a real nice design.
Neal

[KreAture]

I am using a N-channel HEX-fet. It is better than you think.
It has no heatsink solution and yet can handle 14 amps.
(Spec sheet say 14A max continous.) The resistance is 4.7 mOhm (wich I think I've mentioned before) and the threshold voltage is 0.6-2v so it is extremely low.

[Stickman]

Kreature,
You may want to double check that data sheet. Often they are spec'd out at max current cont. when mounted on a 1" square, 1 to 2 oz copper plane integrated with the footprint. Beware of "specmanship." Moreover, the transistor arrays may only use a portion of the available die area insted of a "full die" component. The FET's Rjtheta (junction to ambient) will provide a clue as to if it is full die or partial. Full is better on minimum footprint. The greatest power dissipation actually occurs on the switching edges of the FET, not when fully enhanced at DC. Themal rise at worst case will dictate.

Just sharing some thoughts.

Mark

[Stickman]

Oh yeah, I really like the atmel stuff also!

Mark

[KreAture]

I know Stickman. That is why I did the obvious a looong time ago... I tested it using no extra cooling, 12 amps and different duty-cycles. I never got the FET to more than luke-warm, not even proper body-temp. I have NO concern at all for the FET and I actually wish people would shut up about it. What IS of importance is the voltage regulator as it will get hot.

I am redoing some of the layout again for the ATtiny15 as it will allow for hardware pwm. I have now discovered that it has other features too, that may make this a kickass ESC. Some of the things it could do without extra components is measure voltage accross FET. By doing that when FET is ON, one can implement a over-current solution. In addition the ADC on the chip will be able to give me an idea of battery-levels to allow for LVC with half-throttle setting and full cut. (Would be nice eh ?)

I just hope I can cram in the extra traces without making em too thin, or it will he harder to make for people at home.

Still trying to KISS but it may bloat a bit. Will focus on main functions first as I know only amateurs need LVC and OCP.

[vintage1]

Software is fun.

Glad you checked out teh FET for temp rise.

14A refers to its max current. There is a max junct temp as well remember - and even 4.7 milliohms at 14A is 57mW...well thats not too bad, but what about toggling it at part throttle when peak currents will be higher, and its spending some time in switchover...

I think you are onto a great little project, but try to think of everything!!! Design consists in thinking it all through first, then building it and working out what has happened that you didn't think of :D

[KreAture]

This isn't my first ESC you know...

It's my third. The other two are working great, except I am not satisfied with the software thus I have never released them.

With the Atmel chip I may get this issue fixed and it may lead to me releasing the other two as well.

[Stickman]

Kreature,
Sorry about moaning on about FET issues. You know how it goes, sometimes it's hard to determine how much knowledge and experience someone may have in an area just from written correspondance alone. The important thing, however, is if the exchange of information helps others to learn.

Now where did I set that book covering the Tiny15.

Mark

[KreAture]

Hehe, I only use online docs for such stuff :)
Haven't bought a spec book in my life...

In case you wonder Stickman, I have 2 years of EE education followed by 2 more years at university-level. Paralell to those and ever since I was 11 I have been building electronics.

I do have a tendency to ballpark stuff, but usually I overshoot so that I avoid problems. I'd rather use a FET that can pass a lot more than needed then to have it break and let out the magic smoke.

I am however rather new to flying :) Started this summer and have only owned a cheap Nikko model car that ran on 6 AA's.
Now I have a Eclipse 7 and am playing around with both brushed and brushless motors.
I got my Model Motors MiniAC yesterday and I love it! (I had calculated wich one of the brushless motors I would benefit the most from, based on my weight and size limits. I ended up with a 1215/16 and it performs right on the money. If I gear it a bit I can get my flying wing to do vertical...

Oh, and I haven't gotten around to the new software yet as I need to make myself a better battery-charger first... The one I made first can't do more than 1C without massive cooling so I now made a switch-mode charger (http://www.unixcore.com/~kreature/images/max713/pwm_4.jpg) instead capable of 3-4 amps charge-rate :) It may need some fine-tuning and I ballparked the PWM system bigtime, but atleast the charger is nerly half buildt already so I get my packs charged and can start testing code...

[Stickman]

Never purchased a data book either...must have over a couple hundred. Vendors give that stuff away. Older eyes I guess....just easier to read hardcopy sometimes. And if the MFG takes the time to print it, I don't have to.

Mark

[Robbie d]

my experience with fets is when they burn from excessive heat, the motor hits full throttle!.
looks like a great design though. i'm studying surface mount at the moment, any plans for a kit? or just the boards?

[Trizza]

Kreature, sounds like its going great! I have two ATTiny15L in SOIC form sitting in a tube here waiting for projects, and a few 400s lying around without speed controllers. My PCB etching gear is also lying dormant, so keep me posted on progress!

[Foxy]

Any chance of some info on the charger? I have a 713 sitting there, just need to get around to it!!

[KreAture]

Well, my charger works fine now and it terminates reliably on peaks. I need to go over the temp circuit though, just to make sure. Will be posting full schematic and layout for pcb.
The coil I use can be scavenged from a dead ATX PSU. One should re-thread it though so it can have a higher impedance. (Higher the better.)

[Kevin Murray]

Any updates ? And can you repost the pictures ? They are not loading.

[KreAture]

Here's something.

I kinda left this one dead when I got tired of wearing out the brushes on sp300 motors and went brushless.

I'll probably end up posting some code when I get a use for some brushed ESC's again.

[Hovertime]

Looks really nice... :cool:
The usual questions-any left-over boards, code, part numbers? :o

[KreAture]

I etch em myself.
Will probably get around to cleaning it up and releasing something someday.

[Hovertime]

Will probably get around to cleaning it up and releasing something someday.

Hurry up before brushed motors become collectors items :D
What are the latest specifications/features of this gem?