Brake not adjustable if PWM goes below min throttle setting

[aaromalila]

In F5J-class a motor time limiter is used between RX and ESC. It forces throttle PWM to ~800µs (0.8 ms) after certain period of time. This "shut down" -setting causes ESC (Holybro Tekko Metal) to apply full brake force which damages the transmission gears.

Brake setting % has no effect on this phenomena. Brake setting works fine if lowest PWM is eg 1.0 ms, but the 0.8 ms causes this problem.

I suggest that there would be such an option, that brake force would not be dependent on PWM value applied; the brake value should be always the same, even PWM was less than minimum throttle. This would make the BlHeli32 yet better in our competition class.

[sskaug]

Ok, so this is probably because at around 800us, BLHeli_32 considers the pulse illegal and flags signal as lost. I can make you a test code where this limit is lowered. What is the exact name of the firmware that your ESC uses?

[aaromalila]

Hi, Thank you for a fast reply. The ESC is a Holybro Tekko F3 Metal, firmware seems to be 32.7 [image: image.png]

I made 2 short videos about the problem. In the first one the brake behaves ok as I manually control the PWM to 1.0 ms: https://youtu.be/BSSfQ0ArTUg

We like to use quite smooth brake setting in order to save the gears and motor; max power in some setups is almost 1 kW and motor can be spinning 70000 rpm. This is reducted to ~10000 with a transmission which really does not like very quick stops :)

In second video I let the F5J-logger do it's job; it shuts down the motor when 30 seconds is gone: https://youtu.be/RUmma4f5mQo

The logger inputs 800 microseconds pulse (0.8 ms) as the rules state that "motor must be reliably shut down". We can not affect on that.

Best regards, Aaro Malila, Finland

su 17. tammik. 2021 klo 18.22 sskaug (notifications@github.com) kirjoitti:

Ok, so this is probably because at around 800us, BLHeli_32 considers the pulse illegal and flags signal as lost. I acn make you a test code where this limit is lowered. What is the exact name of the firmware that your ESC uses?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bitdump/BLHeli/issues/507#issuecomment-761838192, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHK2LT572IL3SRI2UAY5CHTS2MFEXANCNFSM4WGFZXGQ .

[sskaug]

I now put a testcode here that accepts down to 750us pulses as valid signal: https://github.com/bitdump/BLHeli/tree/master/BLHeli_32%20ARM/Pwm%20down%20to%20750us%20testcode

[aaromalila]

Hi, Thank you. I will test in the evening. Is there some guide to install the se-version? Can I use BlHeliSuite?

Br Aaro

ma 18. tammik. 2021 klo 9.36 sskaug notifications@github.com kirjoitti:

I now put a testcode here that accepts down to 750us pulses as valid signal:

https://github.com/bitdump/BLHeli/tree/master/BLHeli_32%20ARM/Pwm%20down%20to%20750us%20testcode

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bitdump/BLHeli/issues/507#issuecomment-762048598, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHK2LT2HRETFKXCZAVNK34TS2PQJXANCNFSM4WGFZXGQ .

[aaromalila]

Hi Sorry to bother you with probably quite a trivial topic, but I have not managed to proceed with the flashing.

BlHeliSuite detects the ESC (and has all kinds of flashing options, including adding the file), but it tells the user to use BlHeliSuite32 instead, from where I'm not able to find a function to select a hex-file or anything like that.

Also, blHeliSuite asks for .eep -file, not hex.

Could you please give some quick hints how to proceed. I'm of course browsing all kinds of manuals and tutorials but without success...

Br. Aaro

ma 18. tammik. 2021 klo 9.36 sskaug (notifications@github.com) kirjoitti:

I now put a testcode here that accepts down to 750us pulses as valid signal:

https://github.com/bitdump/BLHeli/tree/master/BLHeli_32%20ARM/Pwm%20down%20to%20750us%20testcode

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bitdump/BLHeli/issues/507#issuecomment-762048598, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHK2LT2HRETFKXCZAVNK34TS2PQJXANCNFSM4WGFZXGQ .

[aaromalila]

... and immediately after my previous post I found some guidance and managed to proceed with the hex-file. Following error was given: [image: image.png]

Maybe I did not elaborate, that the Tekko is the single ESC version "B" ? Anyway, I'm not able to proceed from this point. Sorry :)

Aaro

ma 18. tammik. 2021 klo 9.36 sskaug (notifications@github.com) kirjoitti:

I now put a testcode here that accepts down to 750us pulses as valid signal:

https://github.com/bitdump/BLHeli/tree/master/BLHeli_32%20ARM/Pwm%20down%20to%20750us%20testcode

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bitdump/BLHeli/issues/507#issuecomment-762048598, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHK2LT2HRETFKXCZAVNK34TS2PQJXANCNFSM4WGFZXGQ .

[sskaug]

The image above does not show. Again - what is the exact name of the firmware that your ESC uses?

[aaromalila]

Hi, Ok. The missing image shows the error information displayed when I loaded the HEX file in flash-dialog: "Current target ESC#1: TEKKO32_F3_Metal_B".

BLHeliSuite_32 tells as ESC information that current firmware version is "TEKKO32_F3_Metal_B - Rev. 32.7 - Multi"

Br. Aaro

ma 18. tammik. 2021 klo 20.52 sskaug (notifications@github.com) kirjoitti:

The image above does not show. Again - what is the exact name of the firmware that your ESC uses?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bitdump/BLHeli/issues/507#issuecomment-762419507, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHK2LT73YZ2NOX36LHTAH5DS2R7OHANCNFSM4WGFZXGQ .

[sskaug]

Ok, new testcode now posted.

[aaromalila]

Thank you again, I will test tomorrow evening. Br. Aaro

ma 18. tammik. 2021 klo 21.42 sskaug (notifications@github.com) kirjoitti:

Ok, new testcode now posted.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bitdump/BLHeli/issues/507#issuecomment-762439460, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHK2LT7EKWCV5NMQ2WMJRIDS2SFKBANCNFSM4WGFZXGQ .

[aaromalila]

Hi again, I tested it, no success; There seems to be some inconsistency between what is detected and what is expected.

This time I took screen captures from each phase and added those to a Dropbox-transfer; you can download the picture here: https://www.dropbox.com/t/YyYLhpQe6nLHyvGp (just to avoid typos when copying the error information).

I hope that helps :)

Aaro

ma 18. tammik. 2021 klo 21.42 sskaug (notifications@github.com) kirjoitti:

Ok, new testcode now posted.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bitdump/BLHeli/issues/507#issuecomment-762439460, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHK2LT7EKWCV5NMQ2WMJRIDS2SFKBANCNFSM4WGFZXGQ .

[sskaug]

Are you sure you select the right file? For me, the github file flashes fine:

[aaromalila]

Pretty sure - I deleted the old local file before downloading the new. Maybe I still try again. The error was about the ESC not being of the version to which the FW was ment, if I understood it correctly..

I'll go and test. I download the file from this location: https://github.com/bitdump/BLHeli/tree/master/BLHeli_32%20ARM/Pwm%20down%20to%20750us%20testcode

ti 19. tammik. 2021 klo 20.30 sskaug (notifications@github.com) kirjoitti:

Are you sure you select the right file? For me, the github file flashes fine: [image: image] https://user-images.githubusercontent.com/1584497/105077402-c4496e00-5a8c-11eb-95d0-0d27ea8b6de5.png

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bitdump/BLHeli/issues/507#issuecomment-763037050, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHK2LTY3NOTRN2KQ53OKBRTS2XFVDANCNFSM4WGFZXGQ .

[aaromalila]

Hi again Still same error after selecting the file:

ESC#1: Flash "TEKKO32_F3_Metal" failed in 0.00 s Found different ESC Layouts! Current Target ESC#1: TEKKO32_F3_Metal_B Hex File is for: TEKKO32_F3_Metal

I tried also wit another (identical) Tekko-ESC, same behaviour.

Br, Aaro

ti 19. tammik. 2021 klo 20.30 sskaug (notifications@github.com) kirjoitti:

Are you sure you select the right file? For me, the github file flashes fine: [image: image] https://user-images.githubusercontent.com/1584497/105077402-c4496e00-5a8c-11eb-95d0-0d27ea8b6de5.png

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bitdump/BLHeli/issues/507#issuecomment-763037050, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHK2LTY3NOTRN2KQ53OKBRTS2XFVDANCNFSM4WGFZXGQ .

[sim-]

Hello! :) Just a note, I used 768µs as the minimum pulse length because I saw at least one thing outputting 800µs for idle. It has been a few years, though, and I don't remember what. It may have just been a servo tester.

[aaromalila]

Hi Ok. In my case this is not possible as the f5f-logger between rx and esc requires 1.2 ms pulse to detect "motor start" event. If min pw is set much lower, there will be a big domain where the logger does not work correctly. Aaro

ti 19. tammik. 2021 klo 22.56 Simon Kirby notifications@github.com kirjoitti:

Hello! :) Just a note, I used 768µs as the minimum pulse length because I saw at least one thing outputting 800µs for idle. It has been a few years, though, and I don't remember what. It may have just been a servo tester.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bitdump/BLHeli/issues/507#issuecomment-763134840, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHK2LT6WJHQHV22SSZBNMZLS2XWWTANCNFSM4WGFZXGQ .

[sskaug]

Hi Simon! Thank you for the info, good to hear from you again.

[sskaug]

I just confirmed that this file can be flashed on my TEKKO32_F3_Metal_B ESC. https://raw.githubusercontent.com/bitdump/BLHeli/master/BLHeli_32%20ARM/Pwm%20down%20to%20750us%20testcode/TEKKO32_F3_Metal_B_Multi_32_74.Hex Can you do "save as" and please try flashing again?

[aaromalila]

Hi again, Now I managed to flash the ESC with the test version. The original way to flash ended up with same errors as earlier, but after some inspection I tried a workaround and it worked. Here is a screen capture of the success : https://www.dropbox.com/t/jnBxWCYCC1Bez7ee .

Anyway, the result was not yet ok. Despite of whatever brake setting (starting from 0%) the behaviour is always a "full stop" when PWM goes to 1 ms or whatever is set as low limit. Brake setting has no effect on this, it seems that the brake function is actually not working at all.

Please tell if you want me to do some specific tests. I have a protocol analyser / oscilloscope and I already verified from signal wire that PWM is what I wanted it to be, 1000 - 2000 µs, and there are no anomalies.

Br, Aaro

ke 20. tammik. 2021 klo 9.09 sskaug (notifications@github.com) kirjoitti:

https://raw.githubusercontent.com/bitdump/BLHeli/master/BLHeli_32%20ARM/Pwm%20down%20to%20750us%20testcode/TEKKO32_F3_Metal_B_Multi_32_74.Hex

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bitdump/BLHeli/issues/507#issuecomment-763389138, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHK2LT6XNGMLZDVIIVRH3TLS2Z6TPANCNFSM4WGFZXGQ .

[sskaug]

Can you share the ESC settings that you use (e.g. screenshot of BLHeliSuite32? Do you run nondamped?

[aaromalila]

Yes, I will send you a picture tomorrow. I remember the settings that differ from default: No damping, auto timing, slowest possible acceleration, min thr 1.2ms, max thr 2.0ms.

I tested today min thr 1.0 and checked that my radio also provides that. Originally 5..15% brake, but now it has no effect at all.

Pulse train repetition interval is 17ms.

I can test also with another Esc-unit tomorrow in case the one used until now is somehow bricked. I have 3 identical units.

Basically, as more or less pseudo code, could the issue be solved with eg. a quite early filtering block like

If (pwm_in < 1e-3) { pwm_in = 1e-3; }

Br Aaro

ke 20. tammik. 2021 klo 21.17 sskaug notifications@github.com kirjoitti:

Can you share the ESC settings that you use (e.g. screenshot of BLHeliSuite32? Do you run nondamped?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bitdump/BLHeli/issues/507#issuecomment-763872932, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHK2LT5LYJQJ43IZVREGY4TS24T5RANCNFSM4WGFZXGQ .

[aaromalila]

Hi, Here are rest of the settings: https://www.dropbox.com/t/0Ag1r29HZcu4LMhH I tested with another ESC, same behaviour.

Br. Aaro

ke 20. tammik. 2021 klo 21.17 sskaug (notifications@github.com) kirjoitti:

Can you share the ESC settings that you use (e.g. screenshot of BLHeliSuite32? Do you run nondamped?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bitdump/BLHeli/issues/507#issuecomment-763872932, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHK2LT5LYJQJ43IZVREGY4TS24T5RANCNFSM4WGFZXGQ .

[sskaug]

In these settings you have nondamped mode off. That will give you quite hard active braking.

[aaromalila]

Hi, Ah. Ok The logic has to be read and thought carefully; disabling non-damped mode activates the damped mode :)

Sorry if I wasted your time with the non-damped-setting, it is possible that it has been wrong during these last tests.

Now I tested with test-sw and non-damped mode on. ESC behaves ok, no sudden stop on 1.0 ms or 0.8 ms, but brake adjustment feels somehow binary. Till ~80% brake there is no effect at all, and with 80% or more, eg 100%, there is some.

Aaro

to 21. tammik. 2021 klo 18.44 sskaug (notifications@github.com) kirjoitti:

In these settings you have nondamped mode off. That will give you quite hard active braking.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bitdump/BLHeli/issues/507#issuecomment-764781804, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHK2LT2YUWAY3AM2DLMEGB3S3BKYPANCNFSM4WGFZXGQ .

[sskaug]

Yeah, it's a double negation - awkward, but with some history.

So the new test code is ok, but the Rev32.7 code misbehaves?

Makes sense that the braking effect is quite nonlinear.

[aaromalila]

I will test and cross test with few different setups and units. Brake behaved earlier OK and adjustment was linear, but it might be also possible that the esc-unit used has somehow damaged - I noticed yesterday that the system has some hick ups when applying more power. These did not occur earlier.

Br Aaro

pe 22. tammik. 2021 klo 9.13 sskaug notifications@github.com kirjoitti:

Yeah, it's a double negation - awkward, but with some history.

So the new test code is ok, but the Rev32.7 code misbehaves?

Makes sense that the braking effect is quite nonlinear.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bitdump/BLHeli/issues/507#issuecomment-765186376, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHK2LT3R7KPXFHGZRBU4IH3S3EQPXANCNFSM4WGFZXGQ .

[aaromalila]

Hi Now I tested the 32.74 with another setup and I guess it works quite ok. There is no sudden stop, and in the end, brake works well enough for our purpose. This is currently based on ground testing only as flying season will start no earlier than in March...

For some reason the brake behaviour vs. adjustment value has changed, with fw32.7 (and legal input pulse) the adjustment was rather linear, but now with 32.74 it brakes only if set to 100%, after a ~0.5...1 seconds free spinning. But anyway, this is fine as there is the braking effect in the end, and it is strong enough to stop the prop and let the blades fold back.

(without brake the blades keep spinning due to airflow, and causes aerodynamic brake effect which ruins the glide...)

Can I share the test version with a few friends, or should they wait for possible "official" release? And is it ok to discuss this topic / testing on some forums related to our FAI F5J class?

Thank you very much for your efforts.

Best regards Aaro

pe 22. tammik. 2021 klo 9.13 sskaug (notifications@github.com) kirjoitti:

Yeah, it's a double negation - awkward, but with some history.

So the new test code is ok, but the Rev32.7 code misbehaves?

Makes sense that the braking effect is quite nonlinear.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bitdump/BLHeli/issues/507#issuecomment-765186376, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHK2LT3R7KPXFHGZRBU4IH3S3EQPXANCNFSM4WGFZXGQ .

[sskaug]

Thanks for the feedback, I will then implement the minimum limit change from 800us to 750us on future codes.

Sure, please go ahead and distribute code and discuss!

[aaromalila]

Ok, thank you. Aaro

On Sat, Jan 23, 2021, 09:46 sskaug notifications@github.com wrote:

Thanks for the feedback, I will then implement the minimum limit change from 800us to 750us on future codes.

Sure, please go ahead and distribute code and discuss!

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bitdump/BLHeli/issues/507#issuecomment-765884268, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHK2LT6F6EQNEU6L76KDTNLS3J5G7ANCNFSM4WGFZXGQ .