TBS DiscoPro + MultiWii (Baseflight on Naze32) PID regulation

[Asurendra]

Hello,

I recently posted a poll to gather indication of alternative flight controllers in use by the community: http://fpvlab.com/forums/showthread....ecommendations

I took the plunge and stripped the Naza v2 and friends out of my Disco Pro. I have a couple more questions regarding MultiWii/Baseflight, which are more general purpose, but specifically regarding the Disco Pro airframe.

If you're flying MultiWii with Discovery Pro, I'd like to chat PIDs; could you share how you learned to govern PID regulation, how you determined if you were moving toward critical dampening, and where one would begin to learn?

Initial flight tests are going well however I'm generating some oscillation which is negatively affecting the gimbal stabilization as well as the flight envelope, acceleration and top speed compared to the Naza V2 (which was critically dampened, trivial to do so). It's visible through pilot cam but I don't have DVR here yet. I can share the gimbal footage if it's useful, but I don't think it will be.

In the documentation for Naza V2 gains in the TBS DISCO PRO manual, Magical Naza Rates of 120%/120% are mentioned for Pitch and Roll. Should I be looking for similar rate (120 pitch/roll vs. 100 yaw) increases for Pitch and Roll, in the MultiWii, based on characteristics of Asymmetric Spider QuadX Layout?

Thanks in advance!

System:
* TBS Discovery Pro
* TBS Bulletproof 30A ESCs
* TBS 900kV^2 Motor set
* TBS 4500mAh 4S packs
* Graupner 9x5 e-props
* abuseMarK Naze32 (baro/mag). Default PIDs.
* 2G4RC & 5G8V

[Varis]

Can you describe the oscillation more closely? When does it occur?

I've noticed some oscillation on my FPV camera, not really anywhere else - I fly stock PIDs, they seemed to work great, I do get some low frequency oscillation at least time to time on the feed, but I haven't paid attention to it. I've been attributing it to my 3 x zip tie mounting of the Runcam, which is not a stiff way to mount the camera. But it works surprisingly well!

[Asurendra]

@Varis the oscillation occurs constantly above more than say, 10km/h, along the pitch and roll axis'. I'll get some footage from the gimbal online, although it may be hard to judge from that. I was thinking of just attempting the "scientific" method -- bump values for each of P/I/D along a single axis, observe and record results.

This exact setup had no oscillation with my critically dampened Naza v2 (#justsayin); I know it's not air-frame/setup specific, FC is the only thing that changed on the build. Both the pilot camera and gimbal were perfectly stabilized before (pilot cam attached by 3 solder welds and clips to frame, gimbal bracket/arm/support).

[Asurendra]

An example: Forward pitch attitude, heavy throttle, full fast forward flight / minimal climbs; previously was able to attain 90km/h+, now wobbles around @ 80km/h and struggles to stay locked in, feel distinct forward and backward rocking motion with a locked in collective position.

[Asurendra]

I found http://www.multiwii.com/wiki/index.php?title=PID which answers most of my questions. Sorry for the noise.

[Asurendra]

Down the rabbit hole I go.

Custom "Deadcat" TBS Discovery Pro mix:

mixer CUSTOM
cmix 1 1.000 -1.270 1.000 -1.000
cmix 2 1.000 -1.540 -1.000 1.120
cmix 3 1.000 1.270 1.000 1.000
cmix 4 1.000 1.540 -1.000 -1.120

Scavenged from all over. Run out of sunlight today. More testing tomorrow.

Sources:
http://www.multiwii.com/forum/viewto...scovery#p53428
http://www.multiwii.com/forum/viewto...scovery#p49586
http://polakiumengineering.org/?p=1612

http://www.multiwii.com/forum/viewtopic.php?f=12&t=3809

[Dawagner]

Does this apply to all the multiwii boards or just the one (baseflight) that you have. I just bought the Pro 3.0 (again) as I am tired of my naza and really liked the features on the MW boards.

[Asurendra]

AFAIUI, Any MultiWii board running Baseflight, on a Deadcat asymmetric spider quadX, will need a custom motor mix (due to the difference in triangles between front motors/rear motors).

Try the mix first (it is EXACTLY measured for a TBS Discovery Pro's arm distance) without tuning PIDs. I ran it yesterday on Stock PIDs with the Mixer I discovered and it flew beautifully, no/minimal oscillation, I'm sure it could be tuned to be even better.

Edit: If you don't have a TBS Discovery Pro, exact measurements need to be taken to perform the calculations in order to determine the correct mix offset values. If you follow the threads I linked, it shows all of the working.

[Uav lab repair]

Down the rabbit hole I go.

Custom "Deadcat" TBS Discovery Pro mix:

mixer CUSTOM
cmix 1 1.000 -1.270 1.000 -1.000
cmix 2 1.000 -1.540 -1.000 1.120
cmix 3 1.000 1.270 1.000 1.000
cmix 4 1.000 1.540 -1.000 -1.120

Scavenged from all over. Run out of sunlight today. More testing tomorrow.

Sources:
http://www.multiwii.com/forum/viewto...scovery#p53428
http://www.multiwii.com/forum/viewto...scovery#p49586
http://polakiumengineering.org/?p=1612

http://www.multiwii.com/forum/viewtopic.php?f=12&t=3809

I have been looking into motor mixing alot recently because my frame desperately needs one, I fly a mxp230 with arms for 6inch props (this thing is so far from an x that ts not even funny) I watched the video you linked about and read all the web pages you provided. When my friend helped me figure out the numbers the mix for motor one was as follows: cmix 1: 1 -.75 1.25 1 (this is the mix for motor one he loaded on to my quad and its currently the one it flies.) My question is, was the mix calculated correctly? Im just trying to figure out how you could get a value greater than one if your scale is supposed to be half of one of your motor to motor distances? The numbers I came up with when I did the math are as follows:
cmix 1: 1 1 -.59 1 If anyone can shed some light on my confusion I would appreciate it. The frame dimensions are as follows: 250mm width (front motor to motor) 127mm length (front motor to rear motor) and distance from one motor from center of flight controler is 147.49mm All motors are equal distance from the flight controler

[kompuntu]

How is MXP230 not an X? I mean, yes, its a H-frame, but according to all the pictures all 4 arms are identical and are attached with identical angles to the frame, aren't they?
Looks like a perfect X motor config to me... regardless of the prop size you use.
It wouldn't be an X config if two of the arms were shorter or longer or were angled to the frame using different angles for front and rear arms (like the Disco in OP).
Am i wrong?