This post is about motor tuning for a drone on VESC tool. The hardware is a VESC 6, and this is related to a post I made more than a year ago about the KDE 2814XF-515. I had trouble tuning that one and still don’t have a firm grasp on these parameters, the tuning process and how to optimize this better. Here is what I’ve been using for the KDE 2814XF-515 for the last year to reasonable effect.
R = 62.0 mOhm (for a quoted 130 mOhm motor this seems about half and correct)
L = 6.41 uH
lambda = 1.459 mWb
Kp = 0.0107
KI = 102.95
OG = 496.77
These tunings were done on the vesc_tool_0.95. Those tunings work reasonably well, but it’s often difficult to tell if it can’t be tuned better or if there is some blunder in the settings. It cogs on start-up, but I think that can be solved with the openloop ERPM and time. To simplify our drone thrust model, we control RPM of the motor. 14 polls, 7:1 ERPM to RPM ratio. We use 12.5’ KDE propellers in dual and triple blade varieties. It’s set up coaxially if that matters. In any case, we’ve recently been aiming to try the same motor, but with a different kV – KDE 2814XF-775, mainly because we want to test a hypothesis where these motors afford us a higher maximum thrust capability. I am currently trying to tune the 775 kV motor and I’ve upgraded to vesc_tool_2.05. There are new features on 2.05 that I don’t how to tune and I don’t know the impact of, and when I try to copy over my existing 515 tunings and running a 515, it seems to behave differently despite the same major parameters (R, L, lambda, gain, Ki, Kp, PID settings, maximum and minimums and cutoffs…). The only change is that our 32k switching frequency has been capped to the new 30k maximum, but when I originally tuned the 515, that value was not so sensitive to create such behavior changes.
For reference, these are the two motors in question
In any case, my goal here is to get an optimal tuning for the 515 and 775. When running the detection, I get erratic results that change between consecutive detections. On the 775, my results span from:
R = 25-50 mOhm (The specs have a 70 mOhm listed, so it should be around 35, so this is in the ballpark)
L = 1.00-5.00 uH (No idea how this works or how it’s derived)
Lambda = 0.00 – 25.00 mWb (Using the sample formula to estimate this, I get 1.00 mWb and the detection has often been close to exactly this value, so I think I know when it is right and wrong).
Many of these setups don’t work at all and cog violently. Some work but badly. Some work but it’s unclear if they’re optimal or if they will need retuning with the propeller load.
Note that the setup wizard asks for a wheel diameter for which I enter 0.00 since there’s no propeller on this bench setup and there will never be a wheel on it. Some sample detections I’ve arrived at:
- Current = 18.41 A, R = 29.50 mOhm, L = 2.61 uH, lambda = 0.00 mWb
This tuning seems to work ok on duty and rpm control modes, despite the lambda value being quite wrong. This was done through the wizard on 200 g outrunner (the motors are about 95 g)
- R = 35.35 mOhm, L = 3.45 uH, lambda = 22.593 mWb and with settings T=500 us, I = 6.97 A, D = 0.30, omega = 2000 ERPM/s
This was determined through the FOC step by step detection where you can pick the I, T, omega
- Current = 19.1 A, R = 27.40 mOhm, L = 2.58 mOhm, lambda = 1.00 mWb
Also using the wizard on 75 g outrunner (the motors are about 95 g)
I don’t’ think I’m getting anywhere conclusive without asking for some help about this. I know this is a common post, but any help would be appreciated. I’m using a 6SHV battery of 4500 mAh, im on windows looking to use vesc_tool_2.05. Let me know if you need more clarity or info to help me out here