Pixhawk causing motor sync issues.

Replies

- Justin Stiltner > Justin Stiltner August 10, 2014 at 4:15pm
  
  I did the motor test and all of the motors are plugged into the correct number and are in the correct position with proper spin direction.
Dan Houldsworth June 29, 2014 at 10:52am
I'd like to add my experiences to the list, and am very keen to see this resolved for all.

Frame : 3DR 2014 Y6

FC : PixHawk 3.1.4

Motors : 880kv (http://www.unmannedtechshop.co.uk/brushless-motor-2217q-880kv.html)

Props : 11x6 HQ

ESC : First built using 2 of these (http://www.unmannedtechshop.co.uk/quattro-4in1-30a-esc-quadcopter-e...) 4-in-1 30AMP ESCs, using 3 channels from each.

Performance was reasonable but sluggish with a 3s6200 LiPo, and then switched to a 4s5400 which was responsive.

Ran the Autotune which after 6-7 minutes returned some good results which I adopted, but noted that they were markedly higher Stab_P than default:

For a second build, I switched out the Quattro ESCs for 6x Afro 30AMP ESC. The result is a unflyable. Hence how I came across this (and some other) forum, and have registered to contribute and to find a solution.

Results:
- Early results were tendency to want to flip at take off (high Rate_I symptom?), and wildly jerky vertical bouncing 4-5feet when attempting to re-land or touch down. Previously this had flown like a large mothership, super smooth and stable.
- The AfroESC seem to be getting much more power out of the Battery/Motors as it makes the c.2kilo fully laden Y6 jump around like a fly.
- On a table test various motors would start before others, then with no further stick movement would accelerate up to very high speed while others were stationary. This turned out to be Stab_P and Rate_PIDs being now way too high.
- Better results were achieved with a table test in Acro mode but still acceleration was scary. Blipping to 30-50% throttle seemed more harsh aggressive power than I'd seen on full with previous ESC. I have no side-by-side data to back this up and may have seemed faster due to how jerky and fast the speed came on.
- I needed to drop Stab_P 3x, and Rate_P,I,D by 10x to be barely flyable. Any higher and wild (p-style) oscillations result. Also dropped Throt_Accel P & I 2x.
- Now flying - responds well in Stabilize mode however will accelerate harshly gaining considerable speed and altitude when held in full lean (default 45degs). And would rock and move all over the place during manoeuvres.
- Engaged Autotune which performed 2 twitches and then no more. (I need to see if logs has more info on why this stopped)
- I switched back to a 3s LiPo and it was now less punchy, but still seemed more aggressive than the 4s on the previous ESCs.
- Following some of the advice on this (and other) thread I changed RC_SPEED to 100, and this is where my experimentation ceased as it would uncontrollably want to flip during any slight throttle change and so when I attempted to bring it in to land it crashed.
Other data:

I have the Afro30Amp running successfully on OpenPilot-Revo boards (on different 1100kV motors)

I have the Arducopter 3.1.4 + Motor/Prop combo + SimonK running well (but using APM2.5 + flashed jDrones 30A ESCs)

Further tests I'll be doing and will feed back on:
- Reviewing the logs (I'm new to Pixhawk and so have yet to do this)
- Testing the Afro30A ESC with the 880kv/11x6 with a different FC (probably 3.1.4 on 8-bit AIOP Megapirates)
- Test Pixhawk 3.2beta, and flashing Afro with latest SimonK firmware as although these are new from HobbyKing I have no idea what version of afro_nfet.hex they have)
Points I'm very interested in hearing further clarification on :
- We now understand that Pixhawk PWM is 3.3v whereas APM PWM is at 5v - is this the only difference in waveform?
- Does anyone have data on Afro ESC results on any other Rx or FC at 3.3v PWM?
- I notice that MOTOR_ADVANCE=15 for Afro_nfet, but is =18 on the standard tgy.asm that I've previously used to flash
- Likewise, afro_nfet has DEAD_LOW_NS = DEAD_HIGH_NS = 300, whereas the standard is 450 and 8000 respectively.
- Again, afro_nfet has O_POWER & O_GROUND at 180 and 33, but standard are 10 and 1.
Happy to supply any wiring photo / diagrams. All ESCs are connected with their original connectors, ie PWM plug has signal, 5v and GND all attached for each of the motor connections to the PixHawk. PDB is this one (http://www.hobbyking.com/hobbyking/store/__38061__Hobby_King_Octoco...)
- Christian > Dan Houldsworth June 30, 2014 at 1:29pm
  
  Hi,
  
  so much text, don't be angry if I forget something.
  
  1. Did you set your frame type right? X or + Some guys accidentially started wizard and skipped--> reset frametype--> flips during takeoff
  
  2. Motor sequence and directions right?
  
  3. ESCs throttle calibrated?
  
  4. TX in Missionplanner calibrated or reversed?
  
  5. Don't connect all BEC 5V pins to pixhawk, only one or none if you are using power sensor plug.
  
  6. Rate Pitch I is crazy, set it to Roll I value
  
  7. Stabilze values also crazy high, why not trying defaults? or only 3.0?
  
  8. I really dont want to learn simonk parameters, bcs I'm using BlHeli. but possible, something wrong with chicken brand afros. If you are so smart, change params.
  
  9. Buy other ESCs, perhaps they don't accept 3,3v or maybe 2,4v if you have clone pixhawk
  
  10. read the wiki http://copter.ardupilot.com/
  
  11. don't use autotune if you never have flew with default values before. try using rate_P 0.08
  
  12. set RC_speed higher than 100. (200 or 300) 400 is default.
  
  13. check your vibration levels and do some dampening.
  - Dan Houldsworth > Christian June 30, 2014 at 2:27pm
    
    Hi Christian
    
    Thanks for your reply. To be clear, I'm interested in feedback and information on the topic of the combination of Pixhawk + AfroESC giving unexpected behaviour, unexpected PID settings for normal flight, and / or motor synchronisation issues.
    
    Apologies if my earlier post was lengthy - just trying to describe the issues as best I could as didn't have a camera with me and I appreciate your points 1-13 although most of them do seem to be generic tips for how to start out in multicopters?
    
    To points 6&7 - it was Autotune that arrived at these PIDs with the Quattro ESC and they worked very well, I shared them to demonstrate the dynamics of the vehicle prior to the introduction of AfroESC and to make the point how much higher than default they were.
    
    To point 9 - I'm not looking for a work-around to this issue, I'm looking to understand it. I have a dozen other multicopters of various shapes, sizes and configurations to have fun with in the mean time...
    - Christian > Dan Houldsworth July 1, 2014 at 4:19am
      
      hi Dan,
      
      yes that points are some generic facts, causing strange behaviour for beginners. Sorry didn't know you are already above that level :-)
      
      6+7 ahh OK,
      
      9 and Emin: yes a few weeks ago I did some investigation on PWM output levels. I uploaded some pics of the osciloscope and asked for people to do some comparison. Although to do the test with a cheap multimeter(voltage=readings x2). But nobody was so kind to do that. So now in my opinion, and Lorenz told us, 3DR Pixhawk has at least 3,3V output. APM has 5V and my chicken brand pixhawk clone has only 2,4V. If so, the ESC should accept these voltages or may not work. Now these SimonK parameters makes sense. If O_POWER & O_GROUND is an input(signal) voltage divider you could trim the ratio to accept low voltages. The waveform is as stable as APM or KK board. The only thing I was scared, was the small zero crossing if frequence was set to 490hz. So I reduced to 400hz. But Lorenz said, they set that already to 400hz in hardware.
      
      12. So a good deal will be 200-300 with "no" or "less" adverse in response, but with a safe feeling that the ESCs and not shielded signal wires can handle the frequency.
      
      On the other hand, the big question is always, if motors and ESCs work together in peace. In some cases no firmware hack would help. In others it helps to change timings or demagnetization. I'm not sure if Simonk has that option.
      - Dan Houldsworth > Christian July 1, 2014 at 7:40am
        
        Christian - I take it back, it looks like your point 3 would have helped to isolate my problem after all......
        
        :-)
        
        Christian > Dan Houldsworth July 1, 2014 at 12:18pm
        
        My new best friend, never mind, your investigations shows me that there are still some soulmates on planet earth, and not only football zombies :-) This morning I ordered the DS201!
        
        Now I'm a wee bit jealous on your DS203..
        
        Thank you so much.
        
        So 3DR pixhawk rocks with solid PWM output levels. But I also noticed that "full high" and 0V symptom and some ESCs don't like that. Have a look here under Timeout section and of course the other interesting things...
        
        http://pixhawk.org/users/actuators/pwm_escs_and_servos
        
        So my next task is to disable the safety switch and check again. (alter setting in "full parameter list")
      - Dan Houldsworth > Christian July 1, 2014 at 7:30am
        
        Well here's some interesting results.
        
        I set out to prove the hypothesis that the PixHawk PWM voltage was sagging dangerously low under load due to the PixHawk having underpowered PWM drivers.
        
        I ended up showing that the PixHawk actually has a better voltage than a Spektrum Rx, with a clean waveform, and it was actually my favourite brand of ESC (Afro) that were shipping with differing firmwares / calibration settings.
        
        Anyway, to the findings:
        
        Spektrum AR6210 - sending an idle 900us PWM @ 250Hz at 2.82v while attached to Afro30Amp..
        
        PixHawk + QuattroESC. (Safety Button pressed). PWM = 3.02v 1050us @ 400Hz...
        
        PixHawk + QuattroESC. (Full Throttle). PWM = 3.02v 1780us @ 400Hz. ESC drawing 17Amps...
        
        PixHawk + Afro30amp (Armed) : 3.02v 1120us @ 400Hz. Motors spinning...
        
        PixHawk + Afro (Full Throttle) : 3:02v 1780us @ 400Hz. Afro drawing 20Amps...
        
        So far, all I've shown is that the PixHawk works correctly with both Quattro ESC and Afro30 albeit :
        
        i) Full throttle isn't being reached as the PWM is 1780. The two ESC's must have different throttle curves as the Quattro draw 17A, whereas the Afro draws 20A. Both of these I previously tested to draw 25Amps at full using a servo tester. Interesting, but this is clearly an RC calibration issue and consistent across both ESCs and so not significant to the experiment.
        
        ii) I noted that on boot up PixHawk PWM is held high at 3.0v, whereas after Arming, Disarming and renengaging the saftey, the PWM is at 0v. Again, interesting but not significant to this experiment.
        
        At this point, just to test my own sanity, I took off one of the Afro30amp from the Y6 and tested that (so far for the tests I had used a brand new ESC). And here we have it :
        
        PixHawk + Afro30amp (identical but from different delivery) : 25Amps below 50% throttle !
        
        (Note: ignore the deltaT=2ms, I didn't have the nerve to let go of the board and move the T1, T2 points on the scope. Looking at the grid where each box is 500us you can see the PWM is at <1500us which is typically mid stick)
        
        My conclusions:
        
        The AfroESC I was flying the Y6 with were reaching 25Amp before even half stick. Thats +47% power vs the previous ESC setup at full throttle.
        
        This would have thrown my PID requirements way out. The combination leading to the very unpredictable flying I experienced.
        
        Going forward I will always reflash AfroESC with consistent firmware and calibrations just to be sure.
        
        Ultimately, following these tests I can't find any fault with the PixHawk PWM.
        
        For a further test I will reflash both AfroESCs and retest.
        
        All above tests were with the same 880kv + 11x6 props used in the original Y6...
        
        Dan Houldsworth > Dan Houldsworth July 3, 2014 at 5:58am
        
        To reassure myself that I hadn't somehow calibrated the throttle of the Afros without realising it, there are other reports of late 2013 shipments hitting max throttle at <1500us PWM here.
        
        It really does seem that a batch of Afro30amps were shipped with dodgy factory settings for throttle calibrations.
        
        RC Groups - View Single Post - AfroESC series
        
        RC Groups - the most active Radio Control model community: electric and fuel rc airplanes,rc helis,rc boats and rc cars. Features discussion forums,…
        
        Christian > Dan Houldsworth July 3, 2014 at 12:26pm
        
        :-)
        
        chicken brand

of 37

Discussions

Pixhawk causing motor sync issues.

Replies

Activity