Dear Yag1z,
basically my initial manual ground tuning of PID, starting from standard values as explained in the tuturials (manual PID tuning) achieved good enough results for stable flying.
1st, Roll Axis: Increased D in 50% steps 'till oszillations at take off occur; decreased D until oscilations gone -20%; Same with P; I on guts feeling 50% of P;
2nd, Pitch Axis: same
3rd: Yaw Axis, similar but at the moment ~50% of P,D roll axis and I almost 0, I did the yaw axis on “ok yaw is stable in hovering and slow forward flight & reacts properly” - but this may be also the cause why I get in faster forward flight the “yaw unbalance XX” message (inproper tuned yaw axis )
Currently I dissassembled the aircraft and:
Change the battery and motor wiring to run farther away from the autopilot magnetometer & also to shorten the wire length ( with the new battery it safes me about 900g weight)
The CoG will be lowered too
Re-check and allign all push-pull rods from the actuators to gimbals to have the exact length
Re-adjust all actuators with the alligned push-pull rods
Replace the outer gimbal ring (“The Ring to Enslave”) with a new one, using a different 3-D printing method and material to stiffen the “ring”. Unfortunatelly this adds +100g
Prepare “aerodynamic” covers to avoid wind traps/inadvertent “rudder effects” in forward flight. (+300g)
To think about if it makes more sense to have CW prop fwd and CCW prop aft or visa versa. Or if that does not make any difference at all.
Added a nice mounting interface for the X-stabilization tubes (before only fixed with cable ties)
Added for the landing gear tubes bended ends.
When I am finished with the electrical/mechanical side my plan is:
to recheck behaviour in fast fwd flight
increase yaw authority a bit more
adjust YAW PID as required
test the same with the modified heli code instead of my home-brewed code
P.S: Don’t do Quick Tuning :-D; I’ll try auto tune with a low aggressivity after I have flown my configuration sufficiently to gain data.
Thanks for the update and the information provided. What do you mean by the oscillations? Do you check oscillations on the log or during the flight by your eyes? Looking forward to hearing from you.
I do it old school. If I see visual oscillations, it is sufficient for the beginning and the point where I reduce PID’s.Finetuning with logs is then a later step.
PID tuning works easy, if the mechanical/aerodynamical setup somehow reasonable. But I usually do it three times from scratch to see if I come to the same results.
When you need to beat it to dead with PID, there may be an outer factor limiting the success (with other words, find the outer factor, get rid of it, do tuning again).
Very nice, thank you for sharing this project. I would suggest folding blades. I think they are currently too rigid, and tuning could benefit from some freedom in flap and lead/lag. This principle does not only apply to swashplate mechanisms.
Dear woodoo,
great suggestion. I even have alread a new set of 24" folding props. When I’ve a bit of time, I check the clearance betwen the gimbals mount and props when tilted. The carbon props used are extremly light and are mounted with a a quick release, increasing a bit the height above motor. This allows me to use the 14° tilt range into all direction.
Any update
I have somehow similar design without swasplate but it lacks mechanical aspects and it is unstable while taking off. Did you face any similar issue? When I take off, it is unstable and it makes circle but stays in the air. However, I want it to be stay in the same spot while taking off. Do you have any recommendations?
Dear yag1z,
thanks for asking. I did some updates, which caused updates Some minor things on my to-do list, which should be doable in the next few evenings.
Regarding your problem: is it “toilet bowling” ? (which is often caused by a magnetometer, INS and the stabilization control loop disagreement).
Maybe you have already somewhere shared your design. It is very difficult to give an advise without seeing the physical set up.
It is something like toilet bowling, I’m trying to fix it by assembling again the design. I will share later the design. Thank you so much. You are right without seeing the design it is hard to give any recommendation. I hope your design also got improves.
Jujuuu! Updates and modifications in place. And it still flies
Added “aerodynamic” covers on left and right fork bottom because this was an aerodynamic sinkhole. Also, the wires to the motors and roll actuators are now guided.
Added “aerodynamic” cover above center (ardupilot) mount top plate. A bit for aerodynamic, but mainly to protect the wires against the airturbulances created by the propellers. Additionally it allwed me to move the GNSS receiver/ EX Magnetometer a bit higher away from current path
Moved all RC RX, telemetry and telemetry power supply to the bottom of the bottom plate of the autopilt.
Instead of 2 packs of (2x6s) on left and one right, I have now two big 12s batteries. Lowers a bit the CoG, and in this manner increases a bit the lever arm, but primarily allows me better to move the high current path away from the autopilots internal magnetometer. An weight saver of 600g.
The longitudinal landing gear skid is shortend ~50cm, but has now bended ends.
The lateral supporting x-landing tubes were shortend to about 40cm and have also now beautiful bended ends.
They are now too short and the 2-bees can now easily tilt over when landed a bit too dynamcially. I will extend them again.
Moreover I’ve added a mounting interface for the x-landing-tubes to get rid of the cable ties.
RID added
The two “rings to enslave” (my name for the outer gimbals ring) is now printed MDF. I hoped for a bit more stiffness, but unfortunatelly the stiffness is identical. However, now they have a beautiful color.
Decreased as good as possible all play and backlashes in the gimbals mechanic with shims.
Adjusted all actuator rods.
For the test today, the forward and aft gimbals angles were still not optimal equal (+/- 0.2°). It’s a real pain to adjust all servos to the same angle. Good for today, but to be improved in future. A must.
Software update to latest FW (stable 4.5.7) and added my code ex-coax
Changes roll:yaw to 75:25 (+/-12.5° roll and +/- 4° for yaw)
)
Some minor things on my to-do list, which should be doable in the next few evenings.
