14 June, 2017 at 17:49 #38122bhuismParticipant
😀 great! am looking forward what the cause is of this.
0.7 beta2 flies great, first 2 packs flying fpv ever 🙂14 June, 2017 at 17:59 #38124
Intentional tricopter yaw spins… Feels like it is fighting itself.
I’ve tried numerous times, on several different firmwares, to do level yaw spins like I would on a quad by combining yaw and pitch to keep the camera level with the ground. It always feels like it is fighting itself and is difficult to get any real yaw speed going. 180 flip yaw spins seem ok. Thought I read somewhere a while back that this was due to the front motors I or D term winding up, but don’t remember the details. Is this a known issue being worked, Or should it be added to Github?
LitterBug14 June, 2017 at 18:13 #38127
Appreciating the testing help :).
@LitterBug: It does look better, but there’s still some waggle. Don’t know how much of it because of the wind. Next time you’re going flying, you could try with the yaw correction at 0.
There’s one more thing I want to implement. Now the control takes the motor feedback into account when calculating the servo angle for a PID sum. I.e. the servo deflection is smaller with higher motor speeds for a given PID sum. But it doesn’t take into account the pitch axis correction that will be added to the motor output once the servo reaches the angle. In theory this might cause a bit of overshooting for the servo, but I don’t think it’s a big problem in practice because the servo is anyways so slow. Taking this into account requires some resources, either it adds CPU overhead or uses up memory. Let’s see..
I will also add some debug information that will be helpful in the logs.
@bhuism: Hmm, couldn’t see from the logs what’s the problem..14 June, 2017 at 18:34 #38128mcstanParticipant
when I was doing the tune on my one I found adding too much I in gave it more tailwaggie
in the end, I think only got tail waggle when dropping the throttle.
my servo I’m using these days is bms-20HV running at 7.4 which is 4.3kg but slower than my old BMS-21HV
by the way, you should be allowed to look at my log.. I forgot to switch off permission on it.14 June, 2017 at 18:45 #38129
It is likely a combination of the wind and my running defaults with different motors/props. Wind has been really high here lately. I’m trying to only change one thing at a time for direct comparison of good/bad results. I did not post the previous video of this frame/prop combination before switching to the new yaw control. If I had, you would see just how much it helped. Biggest Improvement was in prop-wash handling. Really want to get some clean air flying in. Maybe early one day this weekend. Also need to remember to set the baud rate correctly on my OpenLager (defaults to 2,000,000 which *flight doesn’t support) so I can upload log files as well.
LitterBug14 June, 2017 at 19:36 #38130
Okay I implemented what I wrote in previous, and WOW, my yaw dropped. My baby is now flying better than it ever has. With zero tuning :). I noticed it from the first 5 seconds in LOS test that now it’s flying as it should.
It completely removed the oscillation when cutting throttle and I noticed no waggling at all in pretty fast freestyling around my yard.
So happy :)!
I will have to clean the code up and add the debug variables. I’ll try to make build tomorrow to see if hopefully others get similar results.15 June, 2017 at 19:55 #38143
Here’s the third build.
There’s now a new
debug_mode = TRI. Please use it when you’re taking logs.
I changed the default yaw PIDs to what I’m currently using.
This one flies awesome on my baby :).15 June, 2017 at 20:05 #38149
SWEET! Looking forward to testing….
Do you still want me to try yaw correction at 0 with this version? Or is that set by default?
LitterBug15 June, 2017 at 20:26 #38150
Forgot about the correction, I was flying with zero. You could also start with zero.16 June, 2017 at 02:24 #38153
Got two packs in on both Babies out at the field with virtually no wind. I set the correction to 0 on both after flashing the latest test. Did not fly good on either frame. Something really wrong with yaw. Roll and pitch by themselves were crisp. Tried to “just yaw” in a hover in acro and it flat would not move without large amounts of stick input, and then it would pitch hard. Punchouts did really well, complete throttle chops were a guess as to which way it would go. Yaw would sometimes wander rolling out of turns or suddenly make a correction on it’s own. Low speed flying it would oscillate slowly.
Ended up in the grass with Baby#1 so did some LOS flying in leveling and horizon modes. That did OK. Was able to do a yaw only spin and get a decent yaw rate.
Unfortunately didn’t have my laptop to try adjusting yaw correction back to 20, and forgot my logger.
Baby Tri #1 config (2100kv motors)
# diff # Triflight 0.7 Beta 2 - new control test / RCEXPLORERF3 3.1.7 Jun 15 2017 / 20:40:12 (3baeba5) servo 5 1155 1965 1547 40 40 100 -1 feature CURRENT_METER feature BLACKBOX map TAER1234 serial 0 1 115200 57600 0 115200 serial 1 64 115200 57600 0 115200 serial 2 128 115200 57600 0 250000 aux 0 0 1 1450 2050 aux 1 28 1 1450 2050 aux 2 1 0 1450 1550 aux 3 2 0 1950 2050 aux 4 13 2 1950 2050 aux 5 26 1 1950 2050 aux 6 31 2 1450 1550 set min_throttle = 1030 set disarm_kill_switch = OFF set serialrx_provider = SPEK2048 set ibat_scale = 720 set tri_tail_servo_speed = 306 set tri_motor_acc_yaw_correction = 0 set blackbox_rate_denom = 4
Baby Tri #2 (2450kv motors)
# diff # Triflight 0.7 Beta 2 - new control test / RCEXPLORERF3 3.1.7 Jun 15 2017 / 20:40:12 (3baeba5) servo 5 1085 1970 1513 40 40 100 -1 feature CURRENT_METER feature BLACKBOX map TAER1234 serial 0 1 115200 57600 0 115200 serial 1 64 115200 57600 0 115200 serial 2 128 115200 57600 0 250000 aux 0 0 1 1450 2050 aux 1 28 1 1450 2050 aux 2 1 0 1450 1550 aux 3 2 0 1950 2050 aux 4 13 2 1950 2050 aux 5 26 1 1950 2050 aux 6 31 2 1450 1550 set min_throttle = 1030 set disarm_kill_switch = OFF set serialrx_provider = SPEK2048 set ibat_scale = 720 set tri_tail_servo_speed = 307 set tri_motor_acc_yaw_correction = 0 set blackbox_rate_denom = 4
Video from last pack of the day on Baby#2 with some text comments added. Yaw tests near the end with different stick and flight modes.
LitterBug16 June, 2017 at 04:56 #38157
Is it possible that the wrong version got compiled? The Description is
Triflight 0.7 Beta 2 – new control test / RCEXPLORERF3 3.1.7 Jun 15 2017 / 20:40:12 (3baeba5
but there is no V3….
V2 had the following description:
Triflight 0.7 Beta 2 – new control test v2 / RCEXPLORERF3 3.1.7 Jun 12 2017 / 10:28:00 (3baeba5)
LitterBug16 June, 2017 at 07:37 #38160
@LitterBug: Your yaw is all over the place :(.
Is it possible that you didn’t have the air mode on? I see that you do have it mapped to a switch. There is btw a feature AIRMODE that sets it on all the time.
Do you have a logging capability now? Set the debug_mode = TRI if you do.
Quite surprised that yours flies that bad.16 June, 2017 at 08:15 #38161sboppudiParticipant
What is the preferred rotation direction of front motors for this release? Is it reverse to previous releases?
– Sree16 June, 2017 at 08:30 #38162
@sboppudi: I haven’t tried the reversed rotation yet, so I prefer the standard: left CCW and right CW.16 June, 2017 at 14:50 #38171
Neither of them wanted to yaw at all. It was like yaw was locked and it wanted to maintain current heading at any cost. If I flew aggressively and used roll/pitch/throttle, it would turn ok. Yaw was just there to waggle….. I have Airmode enabled the way I always do with BetaFlight. If it is armed, airmode is enabled.
Is the ground tailtune stored only in the following parameters:
servo 5 1085 1970 1513 40 40 100 -1 set tri_tail_servo_speed = 307
Or is it also in NVram? I did full chip erase when flashing firmware, copied the previous 3.1.7 diff from each build to CLI then saved, and then did the automatic servo speed calculation on both my builds. I did NOT do an in-air tailtune on either. Left tri_tail_motor_thrustfactor = 54 which is the default like I did previously when it was too windy to get an in-air tailtune.
I plugged my OpenLager into Baby2 so it will make it to the field tonight for BB logs. 🙂
- You must be logged in to reply to this topic.