Thanks, will let you know how I get on.
Yes, this is F427 pixracer. ESC is LittleBee BLHeli-s SPRING 30A. DJI 2312E motors + 9" DJI props.
I’m ready to test custom firmware to spot the issue. But mainly during the weekends 
This hardware setup is working well for me since 3years and I have been following arducopter versions for that time.
@rmackay9,
Thank you for explanation, looking on it everything seems clear. Only one doubt arises. I believe many pilots use QGroundControl so the issue should be spotted earlier. Maybe it is something specific to my pixracer setup still and usage of ESP8266 wifi for mavlink. I have reset copter parameters just a 2 weeks ago for the very first time since initial setup 3 years ago. I did reset when upgraded from 4.0.7 to 4.1.5 and issue appeared. On version 4.0.7 I had SERIALx_PROTOCOL=Mavlink1 and after upgrade SERIALx_PROTOCOL=Mavlink2. Fortunately I did backup of parameters and that is how I spotted the issue (diff on parameters backups).
I did not understand the bitwidth options, so I just tested it against the valid permutations. Firstly I set MOT_PWM_TYPE to 6 for Dshot600. I then set SERVO_DSHOT_RATE to 3 initially and obtained the following results for each of SERVO_DSHOT_OPTS:
OPT
0 No tones 2/4 motors spun on arming
1 No tones No motors spun on multiple arm attempts
2 Triple tone All motors spun on arm
3 Triple tone All motors spun on arm
4 No tones None on first arm, 2/4 on second arm attempt
5 No tones No motors spun on multiple arm attempts
8 Triple tone All motors spun on arm
9 Double tone All motors spun on arm
SERVO_DSHOT_RATE set to 4:
OPT
0 No tones None on first arm, 3/4 on second arm attempt
1 No tones No motors spun on multiple arm attempts
2 Double tone All motors spun on arm
3 Double tone All motors spun on arm
4 No tones None on first arm, 1/4 on second arm attempt
5 No tones No motors spun on multiple arm attempts
8 Double tone All motors spun on arm
9 Triple tone All motors spun on arm
@andyp1per So it looks like minor differences between DSHOT_RATEs 3 and 4, but SERVO_DSHOT_OPTS 2, 3, 8 and 9 seemed to work.
I have not flown this build - is it as safe and valid as the 4.2.0-rc4? Presumably I should look to keep the MOT_PWM_TYPE to 6, SERVO_DSHOT_RATE to 4 and then choose a SERVO_DSHOT_OPT - 9 on the basis that I seemed to get most tones from the ESCs? Or set it to 2 for 4.1 behaviour?
Sorry, when I said dshot bitrate I was referring to MOT_PWM_TYPE
So I am most interested in option 8 at this point. Do you get good behaviour with this and dshot 150, 300, 600 and 1200
@andyp1per
SERVO_DSHOT_RATE = 4
SERVO_DSHOT_OPT = 8
MOT_PWM_TYPE
4 No tones No motors spun on multiple arm attempts
5 Double tone All motors spun on arm
6 Double tone All motors spun on arm
7 No tones No motors spun on multiple arm attempts
So good for 300 and 600, but not for 150 and 1200. Who uses 150 these days? I have no idea if my ESCs work at 1200 - never tried it.
dshot 150 is recommended for longer cable lengths
Can you try dshot 150 with some of the other options to see if any work? e.g.
SERVO_DSHOT_OPT=0,1,2,3,9
there’s a possibility that the prescaler fix affects dshot 150
Hi @andyp1per
No joy I am afraid…
SERVO_DSHOT_RATE set to 4:
MOT_PWM_TYPE set to 4
OPT
0 No tones No motors spun on multiple arm attempts
1 No tones No motors spun on multiple arm attempts
2 No tones No motors spun on multiple arm attempts
3 No tones No motors spun on multiple arm attempts
9 No tones No motors spun on multiple arm attempts
SERVO_DSHOT_RATE set to 3:
MOT_PWM_TYPE set to 4
OPT
0 No tones No motors spun on multiple arm attempts
1 No tones No motors spun on multiple arm attempts
2 No tones No motors spun on multiple arm attempts
3 No tones No motors spun on multiple arm attempts
9 No tones No motors spun on multiple arm attempts
Same result for regular and bdshot?
Thanks - pretty sure I have that ESC, I will set up a test at the weekend
Only tested it on the .apj you sent me.
Can you try this version?
1 Like
Will try and make some time over the next day or so (wife is back and her aunt have turned up so I have to be attentive). Same tests for Dshot150 I presume…
Yeah option 8 is looking good for me - works well on BLHeli32 on Pixracer and MatekH743 and also appears to work well on a littlebee spring with BlueJay
New version works with DShot150! I haven’t notice any issue related to motors now.
I tested new firmware with 15min flight on small backyard only. But still I have tried to push motors as much as possible in that conditions, mainly with up/down rapid moves and short back/forward shakes.
Before firmware update I reverted settings from PWM to DShot and then tested the issue. It still existed. After update motors armed w/o any errors so I went for testing.
Great thanks.
Good stuff, thanks for testing
1 Like
@bjowitt @jtkacz there is a problem with the solution in that it does not work well for F745 boards, I have cooked up another bit slicing that seems to work ok for me. Please could you try it?
MatekH743: Dropbox - arducopter.apj - Simplify your life
MatekH743-bdshot: Dropbox - arducopter.apj - Simplify your life
Pixracer: Dropbox - arducopter.apj - Simplify your life
Hello.
So using Matek 405-se the option for Flow_type is not available. Unless I am doing something wrong there is no option to select the optical flow.
@andyp1per
Version for pixracer works very well. 19min flight with some sharp moves up/down and left/right. Wind has added extra pulses. No issues found.
1 Like
Flow could be feature limited out with that board as is the case with some 1Mb flash boards. You could try the Custom Firmware Builder and check Flow. Just for kicks I ran it and the firmware is sitting there if you want to try it. Or just build it yourself.
Go here and then select Go To Build Directory:
F405 Firmware builder