Debugging the tricopter mini racer

[Terje]

@lauka I’ve run some tests on my Arduino based test stand (still in early pre-alpha stage) so you can’t trust the values besides the timestamp, throttle input and thrust is relative. The load cell won’t work faster than 87ms so that is the resolution you have to work with.

The setup was:
Nano-tech 3850mAh 4S 65-130C Lipo.
LittleBee 30A ESCs as David is shipped them
RCExplorer DYS BE2208-10 1800Kv motor and the Emax 2205 2300Kv red bottom motor (I’m not sure I dare put a DYS BE2217 1300Kv on the stand, broke the 3D printed motor mount when testing the Emax motors…)
RCExplorer stock 6×4.5 HQ propeller

Attachments:

1. LOG012-1800Kv.csv
2. LOG014-2300Kv.csv

[lauka]

@Terje: thanks! That’s really interesting data.

With this sampling rate it seems that the acceleration and deceleration is the same. Need to dig some more.

What kind of setup did you use for this? I remember reading your post somewhere earlier.

Attachments:

1. 

   2016-04-14-18_32_49-LOG014-2300Kv.csv-Excel.png
   

[Terje]

@lauka my setup looks like this 🙂

[Rustypaint]

Don’t know how useful this could be, but the Applied Science guy is using a SimonK ESC to drive a spin coater and he has rpm feedback back via the I2C bus on the ESC.

https://www.bluerobotics.com/forums/topic/blueesc-firmware-compatible-with-afro-slim-20a-esc/

There are more links in the description of the video.

[lauka]

@Terje: that’s an awesome test stand. How’s the pressure sensor connected to arduino? Is it some piezo element? I found it quite funny when the Emax was making the littlebee fly:).

@Rustypaint: Thanks for sharing, very nice channel. Instant subscribe.

[pqueiros]

That is a really impressive setup :). and quite scary as well hehehe.

That is the protocol that the arduino speaks to the ESC, PWM or oneshot?

I am tuned to see those results :).

[Terje]

@lauka it is a load cell as used in common scales. The load cell needs a HX711 controller, and the correct Arduino library. The best I could find is this one: https://github.com/bogde/HX711 (Because it supports bi-directional reading: pusher and tractor prop)

I’m not sure if the load cell won’t break when using it for torque measurements, but I’ll give it a try once I have all the other sensors figured out 🙂 (That might take a while)

That Emax RS2205 2300Kv motor is just plain scary! I put the BE2217 1300Kv on the test stand after printing a new motor mount – the Emax motor broke the old one! Not quite as scary 🙂

[Terje]

@pqueiros I’m using the servo library to talk to the ESC. That is a PWM protocol which could influence the test results now I think about it. Problem is, I don’t know of any Arduino libraries that talk OneShot? On the other hand, OneShot communicates with the ESC in about 200us, PWM takes 1-2ms. That is less than 1% of the total time it takes the motors to go from 1100us to 2000us throttle input.

[Joost]

hmm, very interesting, but how well does the Emax 2205 swing a 6×4,5×3? 😀
Emax themselves only list 5×4,5 bullnose as biggest on 4s, pulling just under 30 amps.

[Terje]

@lauka I printed a beefier motor mount after the RS2205 tore it apart. Here is the log for the RCExplorer BE2217 1300Kv motor.

The logs of the 3 motors seem a bit too much alike – could be that the load cell / load cell controller is the limiting factor in way of latency. I’ll see if I can get some kind of tachometer running to verify this.

Attachments:

1. LOG018-1300Kv.csv

[pqueiros]

Hi @Terje,

thanks for the feedback. What I was actually wondering was whether when using PWM the deceleration can still be done using active breaking. Many times Oneshot is ´sold´ as enabling Active Braking (damping light).

I could imagine that be deceleration times would be significantly affected by the usage or not of damping light.

[Terje]

@pqueiros deceleration is indeed significantly affected by the usage of damping light. RCExplorer sold LittleBee 30A ESCs come with damping light enabled by default. Works fine with both the DYS BE2208 1800Kv and DYS BE2217 1300Kv as well as the awesome new Emax RS2205 2300Kv motors 😀

[hertz]

Oh boy these motors are awesome on my mini quad.
I wonder how they would run on the Mini Tri with DYS XM30A mini ESCs. Can’t wait to try, also very happy that we can now order RS2205 from David 🙂
Have any of you tried the XM30A? They look very promising with the OneShot42/MultiShot support, although BetaFlight currently required for that.
@lauka, any chance to backport MultiShot support from BF to TF?

[lauka]

I’ve been investigating the motor acceleration thingie. It’s very challenging to develop a generic logic that works also at min throttle. It does work well at higher motor speeds. But as there’s just no yaw authority left at min throttle, it doesn’t matter how big corrections are made, it won’t help :). Some kind of prediction needs to be made for the throttle drop so that we make the correction when there’s still some authority left. I don’t like adding special cases, but seems that there’s no options.

Another possibility is to increase the min throttle but that’s no good for acro flying. @PlanB: how high did you need to raise the min throttle to notice a difference?

This is how it currently works:
– I implemented a dynamic yaw gain so that the PID output is fed directly to tricopter mixer. The PID output is dynamically scaled based on the motor speed. This is like an advanced yaw TPA.
– a simple motor model that tracks the motor speed according to a configured value. Added a bit of “slowness” with a filter. It should estimate the real motor RPM accurately enough (based on Terje’s measurements).
– the motor acceleration is treated in PID as expected error.

Ideas and comments are welcome:)

@hertz: that should be possible. Is it just to get down to lower looptimes or are there other benefits?

[LeoTheHuman]

I’ve been thinking of that for a while and it looks like we need to disconnect tail throttle value from everything else. It makes sense to have less thrust in the tail when we’re WoT and lower angle and more angle and a bit more throttle when we’re almost at 0, but I can’t figure out how to calculate that yet. I’m still on 0.4 beta and it’s fine below 75% WoT, but then it makes me feel sick. I’ll update before my next flight and have a couple videos to compare.

[Author]

Posts