I posted this problem on RCGROUPS but on further investigation I a think my problem is in some setup I missed in mission planner.

I have a new X6 based on Pixhawk. The salient features are:
- Tarot 5008 motors
- 1855 CF props
- Pixhawk 
- Multi-pal 40A OPTO ESC

I was getting a Motor imbalance warning in flight that appeared to be getting worse.

       Test: Motor Balance = FAIL - Motor channel averages = [1380, 1586, 1353, 1580, 1371, 1598]
       Average motor output = 1478
       Difference between min and max motor averages = 245

Following a suggestion I tested the setup in the lab. No motors, ARM, spin up for a bit and then checked the logs.

I found an RCOUT mismatch. I re-calibrated the ESCs using all 3 methods in the Wicki and tested each time. I got different but not better results.

My channel 3 RCIN is well behaved and looks as it should but the 6 RCOUTS are all different by quite a bit.

Since this is "static" i.e. not flying am I right in thinking that the Pixahwk has no feedback from the ESC and motors and is open loop. Should it not just send RCOUT that equals RCIN?

ESC callibration spins the motor based on RCOUT and calibration but I am not measuring motor speed just Pixhawk output.

This bird flies OK but I think pixhawk is making all the adjustments

As usual, I assume I missed something fundamental. The attached graph is from a lab test with no props. 

Views: 140

Attachments:

Reply to This

Replies to This Discussion

On the bench the motor will all spin up differently because it is using the IMU to determine how the copter is positioned and will spin up based on level.

Looking at the output it seems that maybe the copter is not balanced because motors 2, 4, and 6 are higher than the reset indicating that the right lower side is heaver.

Mike

You are absolutely right. I realized a few minutes ago that I'm an idiot and that the Pixhawk is still trying to get level even on the bench so this test is worthless.

So I went back out for a test flight and I have improved the situation somewhat. I am now down to a warning.

Test: Motor Balance = WARN - Motor channel averages = [1217, 1324, 1226, 1346, 1247, 1340]
Average motor output = 1283
Difference between min and max motor averages = 129

I also use a couple of amps less in hover now. This imbalance is low enough to be down to weight balance - as much as I have tried to make it balanced. Thanks for the reply.

So I tried something else.

I have been working on this off and on since the last update (Christmas preparations do get in the way of progress).
On my last flight, 2 days ago, the motor mismatch was flagged as FAIL in Auto analysis again so I looked at the logs. The same motor sets showed up.
RCOUT confirmed this but gave me no more insight.
Then I remembered there was a suggestion to tray something else on the bench. 

The first test was hard to interpret so I tried setting the motors to about 30% and lifting the machine and circulating it like I was washing lettuce in a colander. 
I have attached the log file images of the result.

  • The waving about is characterized by the IMU curves X,Y,Z showing my movement was fairly symmetrical.
  • the RCOUT-ALL is what it sounds like but the less cluttered traces are easier to see.
  • RCOUT-C1C2 shows representatives of the 2 sets (high and low).
  • The other 2 images show the 2 sets of RCOUTs, C1C3C5 and C2C4C6


I would have expected symmetry. At least a more balanced set of curves. 
What's the bottoming out of the even numbers about? 

With no props and lift this is the Pixhawk sending different signals to different motors and nothing to do with weight distribution or ESC calibration (I think).

I don't know where to look in the parameters to see if something is wrong. Remember this thing flies so it's hardly a wiring problem I wouldn't have thought.
Any suggestions are very welcome. If you get bored over Christmas please read through my parameter file 

Attachments:

2 other images.

Attachments:

If you would post a data flash log or at least a t-log I could tell you exactly what is going on. It could be a couple of things, but there are no parameters that could cause a motor imbalance. If that were the case and the motor outputs were in error your copter would flip over. If you were in a stationary loiter then you are either holding position in a strong wind and the copter needs to lean into it. If its not that then your airframe is way out of balance. Your bench tests with the props off are meaningless as your control loops are not closed. I will give you a flight test plan.

1)  Fly in loiter mode in a fixed hover

2)  Yaw the copter to different headings like north, east, south, west and hold at these points to get some data.

3) Take the copter up wind from your position and let it drift with the wind in alt hold mode.

If the motor outputs are pretty equal wile in alt hold then there is no problem.

If there is am imbalance while loitering and the high and low outputs are moving around its just the copter leaning into the wind. Does your copter have a very low center of gravity? That would make it harder for it to tilt resulting in a greater motor imbalance to hold position.

My RCOUT values are mismatched in hover. That's my issue. I have tried very hard to balance the machine but had little luck changing anything. In no wind is loiters accurately but the RCOUT is still off. Always the same channels high and low.

The static test was an attempt to see if there was an open loop output difference to each motor. I have 3 motors C1, C3, C5 which apparently get or need a different average RCOUT to the other 3. The static test should show no correlation to these groupings if the problem is weight imbalance,  ESC callibration, motor differences  or propellers. I agree it shows little else but the groupings of results in the static test cannot be influenced by these external factors. Yet still there are 2 groups of motors showing different curves.

I have attached the last, meaningful log file which was the flight where I ran Autotune. The tuning completed properly (i.e. with a success message) but the motor balance is plain to see in the RCOUT section. I have hovered in absolute zero wind and still the same numbers show up. 

When the weather cooperates I will attempt to get the log file you have suggested but I am very certain it will look like the one I have posted here. 

I appreciate your help and input but surely the static test shows asymmetry. That's why I moved it around to allow the Pixhawk to send "random" throttle values to each motor but there are still 2 groups of 3. Is that not strange?

It rejected my log file text so here's the binary.

Attachments:

I just flew again in zero wind. All in loiter (a few seconds in ALT-Hold) and changed yaw and moved a little roll and pitch. This time I removed the gimbal and camera and I swear the machine balances along all 3 rotor arm axis. 

A couple of other observations:

  1. Motors 2, 4, and 6 are warm. Not hot, but noticeably warmer than the other 3 which are cold.
  2. Loiter mode is stable laterally but it often drifts up or down. Sometimes it sits but that's the alt curve. I was not intentionally changing altitude in the middle of the flight.
  3. The reason I mentioned parameters is that the RC3 trim was 1903 and I re-calibrated the transmitter which put it close to 1500. After this flight it has gone down to 982 which is the minimum. Does this have anything to do with MOT_HOVER_LEARN?

I am considering wiping the Pixhawk (loading plane then copter again) and re-doing set up but I will wait and see if anuyone has anything else to suggest or try.

Attachments:

Bob,   None of your logs have shown up here, only the jpg screen shots. They maybe too big. You should use a dropbox or google docs ect. Once again your "static tests" looking at the motor outputs are meaningless. Don't waste your time trying to find something there. The copter must be flying. Either your copter is imbalanced, which I doubt is the case because it would have to be huge, or you have a motor problem or no problem. When hovering in ALT HOLD mode the copter should be level as long as the sticks are neutral. This is the only test to see if there is some kind of mechanical imbalance. It could be a bad ESC calibration or a bad motor. One thing that come to mind, although unlikely, could they have sent a wrong motor, like same size, but different kv? Not something you would normally look for, but check the part numbers on your motors.

Good thing is that it's flying well. From your equipment list it sounds like a very nice machine. It should handle a huge battery and get 35 to 40 min flight time if your payload is't huge.

Good luck and Merry Christmas

I posted 2 zip files with bin files in them. Aren't they visible? 

I could get the bin zipped into 7MB.

cheers

Bob,  Don't do anything to wipe the fc. I really don't think this is a software or param issue. I have been looking at the 2 bin files you posted. It looks like there is something wrong with motor # 6. Since every thing is new I would check the ESC cal of that one or any other thing that might cause it to not put out the thrust that it should. If all else fails you could check each motor individually with a servo tester controlling the ESC or a direct feed from the receiver. Tie it down and run one motor at a time directly, bypassing the fc. Look at the logs and you can see the amp draw for each motor and see if they are drawing the same current for the same input. I know its a pain, but I don't know what else you could do and this would confirm a motor/esc issue with out a doubt.

Thank you. It's a pain but I'll take the whole thing apart and re-build it one piece at a time if I have to. I will isolate motor 6 and measure it against the others. This gives me something to work on and hopefully a lead to follow.

I had gone back and calibrated all the ESCs individually straight from the receiver but I'll also check and see if this motor behaves like the others.

I am happy now that I have something to try. I'll let you know what I find,

Thanks again.

Reply to Discussion

RSS

© 2018   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service