TSDZ8 OSF (open source firmware)

I was not satisfied to assist and I ended up to change new 860C display because existing display cables had broken insulation. I installed v20.1C.5-860C version to display and just tested it with default settings. Assist was not very good. Then I set torque calbiration enabled and there was default values for Torque adc offset 150 and Torque adc max 300. Assist was still not good. Then I set really measured values Torque adc offset 173 and Torque adc max 420. No good assist. Then I reduced the values gradually ended to Torque adc offset 15 and Torque adc max 30.

Assist get better all the time when reduced values.
No I have very small values at torque mode and assist is very good and flexible.

I wonder why Torque calibration is functioning like that.

View attachment 370653View attachment 370652

Also today I uploaded OSF to my TSDZ8/860C (thank you very much @prozyc for your help) and for now your settings are also good for me, the settings from engine from Prozyc worked poorly on my engine, noticeable fluctuation under pedaling and overall low smoothness of power delivery.

Only on your crazy settings that it started to work quite nicely, still not as super smooth as on the original hex software but I can go for a ride now :)

When and where did you buy the engine? Maybe we have some older first versions? I bought mine on the first of October at Greenergy on Aliexpress, maybe we have the same "strange" revision TSDZ8 :)
 
Last edited:
Have you tested the C5 firmware with this OSF? We did a quick test, and it seems to be working. Both the throttle and walk assist work correctly, and the configuration menu also works correctly. We haven't tested the assistance because the motor was not installed in a bike frame.
You can see the TSDZ8 motor running with the SW102 display in the video below:
Thank you for confirming, the video is private though.
 
... Only on your crazy settings that it started to work quite nicely, still not as super smooth as on the original hex software...
This is a very interesting observation that deserves attention.
The values should not be different for cases when calibration is enabled and when calibration is disabled (I thing) .
I had to leave calibration disabled for the assistance to be smooth.
 
Hi all,

Sorry to disturb the thread.....

I have fitted the replacement transistor ( Q7 on underside of controller near the torque sensor cables part number 2SJ168 ) to my controller and it now runs without fault, I have not tested the torque sensing side of things but the manual throttle works and so I must assume all to be good.

If anyone thinks that they might have damaged their controller by flashing with main battery power still applied then this information might help, although it is a lengthy job to remove all of the potting compound, but it's quite easy.

Best regards,

Dorro
 
Hi all,

Sorry to disturb the thread.....

I have fitted the replacement transistor ( Q7 on underside of controller near the torque sensor cables part number 2SJ168 ) to my controller and it now runs without fault, I have not tested the torque sensing side of things but the manual throttle works and so I must assume all to be good.

If anyone thinks that they might have damaged their controller by flashing with main battery power still applied then this information might help, although it is a lengthy job to remove all of the potting compound, but it's quite easy.

Best regards,

Dorro
Thanks for sharing this. This will probably help me to repair my controller.
To test torque sensor you can look at the ADC values provided by the torque sensor. Those values can be displayed on VLCD5 and 860C display. With no load the value should be about 160 and it increases up to about 450 when load is applied.
 
Also today I uploaded OSF to my TSDZ8/860C (thank you very much @prozyc for your help) and for now your settings are also good for me, the settings from engine from Prozyc worked poorly on my engine, noticeable fluctuation under pedaling and overall low smoothness of power delivery.

Only on your crazy settings that it started to work quite nicely, still not as super smooth as on the original hex software but I can go for a ride now :)

When and where did you buy the engine? Maybe we have some older first versions? I bought mine on the first of October at Greenergy on Aliexpress, maybe we have the same "strange" revision TSDZ8 :)
Motor game to me at 14 February 2025. I bought it from PswPower and it was shipped from Germany.
 
Thanks for sharing this. This will probably help me to repair my controller.
To test torque sensor you can look at the ADC values provided by the torque sensor. Those values can be displayed on VLCD5 and 860C display. With no load the value should be about 160 and it increases up to about 450 when load is applied.
Hi, Thank you for that..

I will put my scope on the torque sensor plug later on and compare waveforms against my new controller, if they look similar then I think it's ok... :) ....my motor assembly is on the workbench I don't have it re-fitted to my bike yet....

Here is Q7 at the end of my finger!....de-potting is quite easy, there are a couple of screws holding the heatsink in place, when most of the potting compound has been CAREULLY picked out the pcb can be teased / levered out and access to the reverse of the pcb, is gained, the old transistor can be removed by heating each leg with a soldering iron and lifting bit by bit...unless you have an smd hot air tool....

Cheers,
 

Attachments

  • transistor.jpg
    transistor.jpg
    1,014.9 KB · Views: 15
Last edited:
I want to confirm what @ebikestuff.eu wrote that sw102 works with the version for 860C 15.hex osf.
It didn't work for me before and there was an error like in the picture (e6 fatal) because I had version v20.1C.5-update-1 installed, 2 other versions work fine (I think they are the same): v20.1C.5-beta- and version
v20.1C.5
(I tested and working quite fine without jerking Foc 23, unfortunately on sw102 max is 20A, pwr mode, acc 5 dec 7, tq max 450, calibration disable, if there is jerking you can change to lower dec values maybe it will help as @Bubbels noticed)
 

Attachments

  • 20250525_201822.jpg
    20250525_201822.jpg
    1.2 MB · Views: 6
Back
Top