TxPID/EasyTune questions
« on: February 11, 2017, 09:58:29 pm »
Hello,

I use a Revo with latest LibrePilot software (16.09) on a Quad-X (Hammerhead Nano).
Today I flew the first time since about a year and tried to setup the PIDs using TxPID and Easytune.
I fly in Attitude mode for Roll and Pitch, AxisLock for Yaw, Attitude Estimation Algorithm: Basic (Complimentary), Thrust: Manual

  • In the Input tab, Accessory0 and Accessory1 change depending on the knob position on my FrSky Taranis, but the leftmost value is 172 (for both knobs) while the rightmost value is 1811 (which is also the maximum).
    I don't think it's a big problem that the leftmost value is not 0 because 172 is not even 10% as "minimum" reachable by the knob position.
  • I set EasyTuneRatePitch to Accessory0 with min 0,003 and max 0,022
    and EasyTuneRateRoll to Accessory1 with min 0,003 and max 0,022
In the Advanced Stabilization tab, I see when the knobs are at leftmost knob position a P-value of 0,0134 for rate and pitch. I would expect something around 0,003 here? (The rightmost knob position changes the P-values to 0,022 as expected).

Currently P-values of 0,016 (and the according auto-calculated I- and D-values) seem to be stable (no wobbling).
Sounds a little high for me, since I used P-values of about 0,010 with the old OpenPilot software but I think if the calculation or formulas for I and D are a bit different, the values are plausible. The reason why I'm asking is that the P- and I-values and bars in the Basic Stabilization tab are displayed in red with these high values.

Another question: I bought a Blade Inductrix to be able to train in my apartment.
When I switched back to the Hammerhead with Revo and LibrePilot I noticed a huge difference in behavior.
If I move in any direction (roll/pitch) and leave the stick go (center the roll/pitch sticks), the Blade Inductrix levels and stops at this position.
The Revo and LibrePilot only levels, but it keeps going in the "direction of force" (or how ever it is called/explained).

Is there a way to configure the Revo to also stop it's movement when centering the stick for roll/pitch, or is this not supported?

best regards
Yosuke

Re: TxPID/EasyTune questions
« Reply #1 on: February 14, 2017, 07:47:21 pm »
172 and 1811 sound like normal limits for SBus.  The low value is never zero.  It is thought of as a minimum pulse width or maximum.  There has to be a reasonable value to be detected as a pulse.

The overshoot problem you describe is simply that it isn't tuned as well as it could be.  :)

Try AutoTune.  It will fix that right up for you, and fix some related issues too.  Flying fast forward and loosing pitch control, descending in downwash goes crazy, etc.
https://librepilot.atlassian.net/wiki/display/LPDOC/AutoTune

Re: TxPID/EasyTune questions
« Reply #2 on: February 21, 2017, 09:18:38 pm »
Thank you for your hint regarding AutoTune - it's a gorgeous feature. I don't know why I overlooked it (I think I associated it with EasyTune).
It's really fast and easy tuning the Quad with AutoTune : 8)
Following values have been got (with SmoothQuick set to +100):
RollPitchYaw
P   0,004010,004340,01085
I   0,009570,010370,02073
D   0,000070   0,000076   0,000151   

Sadly it still keeps on moving in the direction I was flying when centering the Roll/Pitch stick and doesn't stop at the current position.
As far as I know PID-loops the "not existing / configurable" D-value of the attitude stabilization might be set to a quite high value to "overshoot" and help to stop the movement.
Maybe this is no good approach since it doesn't take the current speed of the quadcopter into account and it would also affect the behavior when "starting" to move (setting a roll/pitch angle).

Re: TxPID/EasyTune questions
« Reply #3 on: February 21, 2017, 09:28:41 pm »
I'll try to play with Attitude PI value next time. Maybe increasing these values already helps.