High Speed Servo Setup

[fomoco_44]

@Lauka,

I’m after some rough tuning advice for a new/rebuilt high speed servo for the tail mechanism on my stock Mini Tri. I had an unscheduled crash the other day which bent the output shaft on the stock BMS servo, so while I wait for the new(stronger) SS replacement shaft I thought I would recommission a high performance servo I had purchased a while ago.
The servo is a BMS A206, 3.2Kg/0.05sec @ 6v:
https://hobbyking.com/en_us/bms-a206-mg-digital-mini-servo-3-2kg-0-05sec-22-5g.html

I managed to solder on the feedback wire to the center pin of the trimpot, and the benchtop and inflight tail tune have completed with success, so I’m pretty confident that the feedback system is working ok.
The problem I have is that the servo is ‘too’ responsive, and using the same settings as the 210 servo has previously damaged the A206 metal gearbox, probably from too much oscillation/overcorrection. I managed to source a replacement gearset from BlueBird directly, and so now that it’s rebuilt I’d like to get a better understanding of how/what I should change to compensate for it’s higher speed.

The tailtune results in a servo speed of 640 and thrust factor of 83, so you can see from this that the speed is roughly double that of the 210 servo.

My current plan of attack is to lower all the yaw PID’s(25,30,30), enable the yaw lowpass gyro filter @ 60Hz and enable the servo lowpass filter @ 30Hz. This seems to be working ok, but still needs some tweaking, but I would like to hear from you or anyone else as to the best engineering practice for this high speed servo….if you don’t mind 🙂

[lauka]

Wow, that’s some beefy servo you got there. I have had trouble with higher speed servos coupled with the unbalanced tail motor mount. The acceleration can screw up the yaw PID.

Your plan sounds good, but I wouldn’t lower the gyro filter at all. It hurts the performance of all axises. I would lower the yaw PIDs very low to start with. It’s a good idea to add the lowpass filter to the servo output. Slowly raise the PIDs until the copter is good and responsive and then start to raise lowpass filter of the servo, maybe even remove it in the end if all looks good.

Good luck with it! I burned 3 fast speed servos in a short time span when testing them out :). They were the Bluebird 22HV servos.

[Kevin_Erik]

From what I understand this servo is a good substitute for the stock one: https://hobbyking.com/en_us/bms-385dmax-digital-servo-metal-gear-4-2kg-15sec-16-5g.html

[MadPoet]

@Kevin thanks for the link. metal gears, no note on the shaft being SS, but the price is right.

Going to be following this thread regarding high speed servos with interest.

[Kevin_Erik]

Be aware that there is no feedback on this servo. Either you’ll need to add it or setup your copter without it.
Otherwise you should stay with the standard servo.

[fomoco_44]

@Lauka,
Thanks for the advice, that really helps with the tuning now….sometimes it’s hard to know which axis is causing the problem on a Tricopter!
I was actually going to get a 22HV, but after your experience I think I’ll stick with the A206….or the big brother A207 if I can find a good deal on one.

[fomoco_44]

@Kevin_Erik,
Thanks for the link. I can confirm that the A206 is a drop-in replacement once you add the feedback wire, whereas some other servos I have purchased from HK have had the wrong sized spline…..says 25t but it’s a smaller pitch than the standard Futaba 25t.

[Kevin_Erik]

@fomoco_44

I had similar issues with my latest build (tri-baby). What i found that worked was to set the Gyro LPF to 20hz, disable all notch filtering and go Dynamic. From there i cut the yaw PID rates in half and used that as the starting point.

High speed oscillation / jitter: Too much P-term.

Low Speed oscillation in forward flight: Too much I-term.

Keep in mind that you want to isolate flight controller as much as possible. Hence using silicon pads under the motors and FC anti-vibration mounts will also help a ton. Furthermore, make sure the servo and tail yaw assemble is mounted securely. If they are loose, this can create both slop in its movement and some funky oscillation.

Your power supply (BEC) is really important! Too many things running off the same BEC can cause the voltage to brown-out. Hence if your running something off the same 5-6V output as the Servo, consider getting a dedicated add-on BEC just for the servo. Be aware that when a servo experiences a brown out, it draws more current to compensate for the voltage loss. This in-turn can damage the servo, burning out the motor or driver circuits.

Lastly, look up the exact specs for your servo. Triflight defaults to 300Hz while Davids servo maxes out at 250Hz. Not 100% sure but if the aervo cant keep up, then it might be damaged by the excessive PWM frequency.

[fomoco_44]

@Kevin_Erik,

Great advice!
I’ve gone to dynamic filtering without a problem and cut the Yaw PID’s right down(34/40/40), so now I’ll just slowly bump them up and see how that goes. Based off lauka’s advice, I’ve removed the Gyro LPF and set the servo LPF to 42Hz…..though I’m thinking of lowering the servo LPF right down as well along with the PID’s.

I’m using the integrated F3 FC, so all I have done at the moment is use anti-vib spacers on the motors and added extra ties to the rear mechanism.
I saw your advice about the servo specs and Triflight, so I have set the PWM to 333hz and center pulse to 1520, and then redid the tailtune completely. Had a pretty good flight this morning, but it is a bit wobbly during sharp turns, so I think I can raise the P term a bit and try again.

I was planning to re-do the whole tune again, but I’m unsure if the standard method of raising P until you hear oscillations on each axis will work with the Tricopter…..I’m a bit worried that a bad yaw tune will induce a bad Roll/Pitch tune.

[Kevin_Erik]

If it’s wobbly and the tail isnt wagging, then this could be more of an I term issue. (IMO) Reason I say this is that the I-term is whats the “past” and tries to keep the servo from Drifting from its current position. However too much I-term results in a slow oscillation thats most noticeable when your attempting to fly in a strait line or hover.

The P term is more an “here and now” type of adjustment. Its determins how agressive and the imeadiate RC inputs will translate into a vector change. Too large of an P value results in rapid overshoot and over-correction. This is most noticeable as “tail-wag” as it ends up feeding off itself, keeping the highspeed oscillation going. (until something stops it)

The D term acts as a dampener for your P term / propwash, by reducing the amount of overshoot and oscillation. However too much D-term adds excessive noise and vibration into your system. This can cause your motors to overheat and also adds delay to the reaction time. It’s most noticeable when decending or at the end of an acrobatic manuver as your going through alot of propwash. (will feel as it the control is lagy or slow to respond)

[Author]

Posts