Autotune crash, lots of testing and can't seem to find why. It is repeatable

I have a hexa copter that has flown great for many months. The other day I changed the battery size and up dated the am32 firmware (tekko single esc’s) also updating to arducopter 4.6.3.

I thought that I would run a quick autotune and see what values it gave me and it crashed.

So the first autotune roll to the left it never recovered.

I put it back together and it keeps happening.

When it rolls to the left it sounds like the left motors are shuttering and the esc temps to that motor skyrockets.

If I flip it out of autotune to loiter the odd flying continues, but once I flip into stabilize it goes away.

I ran a quick tune and that worked fine.

I don’t really need to autotune this craft bc I can manual tune it but its got me curious as to why its acting this was.

I decreased the auto tune aggressive and it still did the same.

I Made other flights with some aggressive Pitch/Roll and it flew fine, it seems to be localized to autotune.

Here are my am32 settings…

I just reflashed my esc’s and did another test flight and did the same thing.

Yeah, been there. Done that. It sucks, but it is fixable.

Turn down the ramp rate. Try 10 or 12%

ESC-5 seems to have a lot of noise.

That same ESC is getting very hot. This may not be the cause of the trouble, but damage as a result. I’d consider replacing that ESC if I were you.

What filtering capacitors do you have? I’m not sure if those ESCs have caps right on them. (the ones I looked at on-line didn’t). If the caps are right on the ESC then you need at least 100-200uf per ESC. If you have to install the caps further away (on the power distribution board for example) you should consider more. If in doubt, add more or larger filtering caps.

@Allister

Thank you for the reply, the esc’s caps are direct 500mf. The esc only gets hot when it’s starts this issue with autotune. I can fly the copter for 15min and same temp as all the rest.

I will try turning down the ramp rate and see what happens.

If you have 500 on each ESC then you should be safe for caps.

Keep an eye on the heat of that ESC. You can see it was hotter than the rest even before the autotune. In my experience with this problem, once an ESC overheated it would start to overheat compared to the rest, even just on the work bench when the other ESCs were cool. I know there are a lot of variables to why that would happen, but I’d just keep tabs on that ESC for a while to make sure it doesn’t cause you trouble later on.

I think I’ll use your advice and replace it. I have more and if it doesn’t fix it then no biggie.

Again thank you for taking time to look over this

If it’s a bad esc then why does it only act up once autotune is flip and I can flip it to stabilize while it acting up and then it goes back to being/flying normal?

I’m just trying to wrap my head around all this.

1 Like

I went back and looked at logs from 6+ months ago and it appears that esc has ran 10+ degrees Celsius higher than other esc’s.

Then it might just be a cooling issue. Turning down the ramp rate will certainly help out.

I’m also looking at my notes, we’re running auto timing advance, and a variable PWM freq 16-32khz.

1 Like

Changed am32 settings below…

Autotune still causes the left side to fail.

I was changing one thing at a time to see if settings would fix it. planning on changing the esc here after while.

What was interesting, I used my infared temp gun and all esc’s are the same temp regardless of what they are reporting. Also then are all mounted directly under the propwash.

Also try the Auto timing advance, and 16-32 PWM freq.

I put a new esc and motor on and still doing the same thing.

This time I could reproduce with extreme stick movements outside of autotune.

I am going to try 16-32 PWM freq., I have it on auto timing in this log but not the 16-32 PWM freq..

I will change that now and see.

Oops deleted my post.
With the issue always on the same output that would seem to be an issue with the FC/firmware rather than an ESC setting that they all share. Try Double Loop instead of Triple perhaps. This may not be a problem with a quad and 4 outputs. Spitballing here :slight_smile:

If I think back it seems as though all this started when I upgraded to the 2.19 am32 firmware on my esc’s.

These 2 settings are what I think may be the problem but im not sure.

@Allister @dkemxr I believe its fixed.

I reduced the duty cycle down to 4% and I was able to run autotune with no issues,

Thanks again for looking into this.

I remembered reading the thread by @Allister when he was having trouble and @andyp1per said…blhelis was 8% slew rate so I figured i’d drop it low.

1 Like

Interestingly I increased it from 4% to 6% and it had the issue again. I’ve decreased it to 4% again but don’t have time to test tonight. I’ll be back at it tomorrow.

The copter seems very twitchy when I increased it.

You haven’t said what kv your motors are but generally the lower the kv the lower you need to set the slew rate. For 1000kv, for instance, don’t go above 10%. I run all my 5” quads on AM32 at 12% because this gives me about the same performance as blheli.

@dkemxr has a good point about loop rate. AM32 really needs a higher loop rate than BlHeli. You’ve got it triple now, but consider running SERVO_DSHOT_RATE,4

1 Like

They are 1750kv, 5in props.

What duty cycle do you run on 5in?

Here is the latest flight and still doing it. I increased my loop rate to 4. I changed a few settings in am32 as seen above.

1 Like

The timing advanced only saved on 1 ESC. (one green dot and 5 red ones) so you might want to make sure that takes evenly across the system.

Good catch, I’ll make that adjustment.

Oh since I picked “auto timing advance” its grayed out as it doesn’t matter

BTW I added the heavier payload/battery back to the copter and no change.