TinyTwo Speed Controllers

[haz]

Ok no worries. I'll talk to you on the day about some pre orders of the next batch maybes?

[Rapidrory]

Yeah sure; as I've said I've got a few designs in mind, so it would be good to try and get an idea of the sort of specs people are after...

[Rapidrory]

@Duff, the one you're getting has also been conscripted into service for the AWS, so you may have to wait a bit while I extract it from one of my robots after the event

*EDIT* whoops, that was meant to be an edit not a double post!

[haz]

something that can handle the beetleweight sized motors and 4s might be good if you could. I'll give you a pair of motors at the aws if you want to see if you could do something.

antweight wise anything that can handle 3s and 2-4 hp motors would be nice

[Rapidrory]

I'll look into it :L What current rating are the beetle weight motors?

If anyone has any specs they'd like to see in a speed controller, feel free to suggest it.
Not saying i'd make one for every spec, but just want to try and get a better idea of what people are looking for so I can make something that suits the most people

[peterwaller]

They're still largely untested, and the tests I have done show waay too much noise being coupled onto the servo data wires, causing it to think you're waving the sticks round madly under heavy load, which isn't ideal... Gotta try sort that and a few other things it in software and hardware before Saturday..

I had similar problems which turned out to be the 5v regulated voltage dipping when a high power servo operated.
Below is the scope trace the red trace is the power supply and the blue trace is the 5v regulated supply.

By selecting a better regulator (1.5a 0.2V LDO) and a full 100uf decoupling capcitor things improved.
One of the more important things was to connect the servo to the controller board as close to the regulator as you can otherwise the servo is drawing it power down the same wires as the receiver causing bigger drops in the receiver supply voltage and interference with the receiver operation.
This unfortunately increases the size of the controller but some of these servos can take currents upto 1.5 amps for short periods.
Hope this is of some help.

[Rapidrory]

Thanks for the advice

It's not the servo though (yet..); this was just running it with motors and an Rx. I checked the regulator output and it was smooth as ever.. same with the outputs of the PIC to the driver chips. It seems to be capacitively or inductively coupling straight to the servo wires (although these tests were done at 2:00am last night just before I went to bed, so I'll re-run them later). My plan is just to connect them through a 1nF capacitor to ground, as it's all in the form of very short spikes rather than continuous noise so that should help tackle them... Also gonna try adjusting the software to ignore short pulses... Hopefully that should fix it at least for Saturday..

To tackle the voltage reg noise, I'm planning to run the PIC off a 3.3v zener diode in later versions (running off a 4.7v zener atm), so that it is stable even on a 1 cell. This should also cut out any noise from the aux power lines to the PIC. Though it'll still effect the Rx, so that is yet to be fully tested...

[BeligerAnt]

Surely your code and/or whatever you are using to capture the pulses ignores narrow pulses anyway? Anything less than about 800us is not a valid RC pulse (strictly 1ms, but I wouldn't bank on many systems adhering strictly to the limit ).

If you have noise on the supply, it could be causing the PIC to reset due to brownout detection. A zener regulator won't provide much protection against fast dips in supply voltage. An easy way to diagnose this would be to put a long delay in the startup code to make the effect more obvious. Once you can figure out the cause you can work on a fix. If you only have the effect to go on it's much more difficult

Ants are a surprisingly nasty electrical environment (everything is so close together) so don't be surprised if making the hardware and software robust in the "real world" takes far longer than getting it "working" on the bench in the first place... Such is the joy of development

[Rapidrory]

The positive side of the supply is fine; the zener's doing a fine job, and I've tried running servos off the 6V rail, and even when it's stalled on a 3 cell LIPO the voltage doesn't drop (does get very warm though!). You get a short 8 volt peak when one of my servos turns which isn't ideal, but it hasn't caused any noticeable instabilities... At least in comparison...

It's the ground tracks that are causing more trouble; These chips have an over current lock out if it senses that the ground pin has gone over 500mv, which is meant to be used with a low value resistor so that when the current limit is reached, 500mv is present on the ground pin. I've done without the resistor, but even so i'm getting enough noise on the ground track to cause a lock out when running on a 3 cell.. The ground track is only 1mm wide, which is small, but I had assumed it would be ok over such a short distance... obviously not! I'll try bulking it up with a layer of solder tomorrow; but that means removing the driver chips without braking them, which won't be fun -.-

My firmware ignores signals shorter than 128 program loops, which is about 160us... I'll add some more time to that, but that was the longest time I could do easily; It's all in assembler; nothing is easy! :L I've got motor braking to add as well.. fun times

On the plus side, I've at least identified where most of the instabilities are coming from now (I think)... I have 4 (maybe 5) hours in the lab tomorrow, so hopefully can correct them by Saturday!

[Rapidrory]

Well, despite it working in Lionel at the AWS, I don't think I'll be doing a commercial version of the High Power of the Nano Two, as those chips are just too picky to work under the conditions that they're put under on my boards.

Alex (Haywire) tried to fit one to Accelerant, but we found that the noise on the 0v line combined with running the chips too close to their lower operating voltage limit was causing the chips to run poorly or lock out when running high power pololu motors. I even tried running Lionel on a 3 cell after the event to see if that would help the low voltage problems, but this just added more noise and caused lock outs. I think part of the problem is that I don't have room for a proper ground plane on such a small board, so it's able to accumulate noise easily (and adding more capacitors didn't help). I may have another go with those chips at some point, but for now I'm moving on to other chips...

The chips from china have just arrived. These have 2v minimum voltage, and no lock outs other than thermal, so should be a lot more stable. As I think I've mentioned before, they only have a 1.2A continuous current rating, but they do have a 2.5A peak rating; I tried running stress tests on one of the chips using a high power motor, and it didn't complain even when the motor was stalled for a second or so... The new boards are now in print, so I'll have to try one in Accelerant once I've put them together. (Alex's robots are the ultimate stress test for any ESC )

As far as the onboard voltage regs are concerned, I think I'm gonna stick with the ones I have, as it managed to run Alex's 5.4kg/cm servo on stall at 6V without any noticeable voltage drop and without getting too warm, and they cost about 5 times less as the nearest equivalent on RS :L There was a fair bit of noise, but I do only have a 4uF capacitor on the output atm which I plan to increase by at least a factor of 10.