New "TSDZ2 Torque Sensor Central Motor"

[skestans]

Calibrating the torque sensor is a bit of a pain but it's worth it, the values vary widely across motors and you're not getting the most out of the setup if you haven't calibrated. Just make sure to write your values down somewhere because they get wiped when you update.

I found that 1.1.1/1.1.0 has the powerboost setting inverted. Setting it to disabled enables it, and enabled disables it. So if you have no assist when starting, try disabling startup boost to enable it. That did it for me.

 

[skestans]

So I tried v1.0.0 and v0.6.5, always the same issue: enabling temp sensor disables pedal or walk assist, disabling the temp sensor restores it. If I set temperature to throttle instead, it does read a value and accelerates.

I guess I wired it wrong? I'm rather discouraged at that point because I can't see what's wrong. This is how I wired it, using the alternate GND as my controller didn't have a wire in the marked GND on the picture:

Did I get that wrong?

 

[beemac]

So I tried v1.0.0 and v0.6.5, always the same issue: enabling temp sensor disables pedal or walk assist, disabling the temp sensor restores it. If I set temperature to throttle instead, it does read a value and accelerates.

I guess I wired it wrong? I'm rather discouraged at that point because I can't see what's wrong. This is how I wired it, using the alternate GND as my controller didn't have a wire in the marked GND on the picture:

Did I get that wrong?

Maybe try another gnd? There's a ground on both the speed sensor and display controller cable - try that instead of the battery gnd which i think you're using (edit - correction looks like you're using the black from the pas sensor). Preferably the display gnd as that's the usual setup.


I know it shouldn't make a difference if all gnds are tied together... but it might!

https://github.com/hurzhurz/tsdz2/blob/master/pinout.md

 

[hetm4n]

what value is the sensor showing you? I had the same on 1 of the engines from the client. I could not install the sensor for him because he read wrong, so the entrance was damaged. the sensor showed a temperature above 100 °, which blocked the support. when I turned on the throthe, the engine started as if it had the throthe fully extended.

 

[skestans]

what value is the sensor showing you? I had the same on 1 of the engines from the client. I could not install the sensor for him because he read wrong, so the entrance was damaged. the sensor showed a temperature above 100 °, which blocked the support. when I turned on the throthe, the engine started as if it had the throthe fully extended.

The room temp is around 23 degrees, but the sensor reads 42C. It's still below the 75/85 limit though but the reading doesn't match the temperature.

 

[skestans]

None of it makes any sense. I had a look again, and I'm hooked up for GND onto the display's GND. 5V and throttle as per the wiki picture.

When I unhook the temp sensor from the motor and measure it with an external 5V and a multimeter, it reads 239mV which is pretty much the room temp (23C)

But when the display reads it (on 0.6.5), it shows 40 something degrees. Why?? I think it read it correctly on 1.1.0, I remember looking into the technical menu and seeing around 200mV.

 

[hetm4n]

here may be a hardware problem. and how does the throttle behave when you hook it up?

 

[skestans]

here may be a hardware problem. and how does the throttle behave when you hook it up?

The throttle works well (with the temp sensor physically disconnected.

Because of that and the fact that 850C + 1.1.{0,1} + temp sensor works well for everyone else but for me (also tested on 0.6.5), it HAD to be a bad connection somewhere and not a software bug.

Luckily, I had put the temp sensor on a connector so I unhooked it, connected it to external 5V and GND, and checked the Vout again: nominal, shows 250mV for 25C room.

Then I hooked 5V and GND back onto the motor and measured Vout, it was wonky. I checked voltage between 5V and GND, it was also weird. Then I checked GND at the temp sensor connection and at the end of the display cable where the actual display connects, and I didn't have good continuity.

So the culprit was a bad GND wire splice (it is a major PITA to do a 3 way splice with so little space). I redid it, and now everything works well.

Thanks everyone for all the help, turns out it was just a bad solder job on my part!

 

[electroriderIT]

Any IDEAS??



Thankx!!!

UPDATE! Tried the bike for around 20 km and seems to work...so strange but it's like the sensor had to settle... I will try it again and see if still gives me the problem of "no assist" at start but only when the bike is already moving.

The temp sensor works very good...I reached around 65° with a 3km climb.

NOW...Probably a configuration and settings issue to solve for this behavior :

[youtube]y4ScIMUBOTk[/YouTube]

Probably, now that I serviced the motor, cleaned the torque sensor and springs from all dirt and grease, I have to lower the sensibility...but cannot make it not doing like in the video.... the motor start way to easily even without weight on pedals..

CAN YOU HELP ME?

thank you!!!
ps: i 've never calibrated the torque sensor..maybe I should do it....seems complicated though!

OK..Thank you...I made a video of the state of the art right now. I reassembled everything very carefully, routed the 5 wires not in contact with the 3 phase wires.

The problem that you see in the video is that if I power up the unit, wait, start pedaling ....the motor doesn't engage BUT if after start pedaling , I stop for a moment pushing on the pedals and then push them again (without letting the bike stop), the motor starts!!! and then everything seems to work good, even the sensibility of the torque itself. it's like the torque doesn't engage well .... I noticed also that if I "kick" the pedals, then the torque "unlocks" and it works....
[youtube]9ImPFxDXXXk[/youtube]
In the video you see that I start and nothing happens, then you can hear the motor just after I stop and restart pedaling without letting the bike stop...then when I stop completely (00:50 second of the video) and start pedaling..nothing happens until I release the force and put back on the pedals.....soooo strange!!!

The torque sensor from my close inspection, looks good so I don't know if it could be malfunctioning anyway and if there is a way to test it before buy a new one.....but the data from the "technical" menu are good?.....

thank you Casainho for helping me out with this…it’s frustrating….I got to understand the problem and solve it ….

In the “technical” menu I have those values…don’t know if they are correct or too low, but the motor doesn’t ingage
78112D99-4E83-499D-90CA-A4D562C1346F.jpeg

Hope you meant those ….

What do you think?
I also read that the 5 pin wires that comes from the motor must not be too close to the 3 phase wires…I will try to change their position…now are attached

Ok, cadence seems to be ok as also torque sensor, as long torque sensor ADC values increase when you press the pedals.

Maybe the issue is some wrong configuration??
Other way I can't understand why

 

[beemac]

ps: i 've never calibrated the torque sensor..maybe I should do it....seems complicated though!

calibrate your torque sensor. or at least check to make sure the weight values aren't way out of line with what your ADC is reading.

Check the values in setup - compare with the value of ADC torque sensor unloaded. If the values in setup are much lower - then it's going to think you're stamping hard on the pedals when you just touch them lightly.

calibrating it is a bit of a faff - i made a 1m long wooden arm to attach to the pedal with a vertical bolt on the end - then I can put dumbbell weights on it. The 1m length means you only need 1-10kg to measure over c. 10-100kg range... tape up the brakes and ideally get someone to help hold the bike.

the response isn't linear - otherwise you could just do the unloaded value and you standing on the pedals and interpolate... i find the value tends to rise quickly up to about 50kg then flatten off...

This is my last calibration - this is by far my most 'linear' sensor



although watching your video - i had a motor behave similarly and I replaced the controller board after much testing. The ADC torque sensor reading would go up under pressure to normal values but took ages to drop back down to unloaded value. So the motor just went to full speed and took ages to stop...

I'm still not sure whether it was the controller though (not tested the one I removed) - I wonder whether just removing and replacing moved or reseated something...

 

[electroriderIT]

Thank you! Is there any easy for dummies guide for calibration other than the one on the wiki page?

ps: i 've never calibrated the torque sensor..maybe I should do it....seems complicated though!

calibrate your torque sensor. or at least check to make sure the weight values aren't way out of line with what your ADC is reading.

Check the values in setup - compare with the value of ADC torque sensor unloaded. If the values in setup are much lower - then it's going to think you're stamping hard on the pedals when you just touch them lightly.

calibrating it is a bit of a faff - i made a 1m long wooden arm to attach to the pedal with a vertical bolt on the end - then I can put dumbbell weights on it. The 1m length means you only need 1-10kg to measure over c. 10-100kg range... tape up the brakes and ideally get someone to help hold the bike.

the response isn't linear - otherwise you could just do the unloaded value and you standing on the pedals and interpolate... i find the value tends to rise quickly up to about 50kg then flatten off...

This is my last calibration - this is by far my most 'linear' sensor

torque sensor.png

although watching your video - i had a motor behave similarly and I replaced the controller board after much testing. The ADC torque sensor reading would go up under pressure to normal values but took ages to drop back down to unloaded value. So the motor just went to full speed and took ages to stop...

I'm still not sure whether it was the controller though (not tested the one I removed) - I wonder whether just removing and replacing moved or reseated something...

 

[beemac]

Thank you! Is there any easy for dummies guide for calibration other than the one on the wiki page?

not really - the hard part is getting the weight on the pedals in decent increments and keeping everything upright. You add weight - you take a reading - repeat

If your start and end values are similar to mine - you could always guess and hope your curve is like mine - or you could just put linear values in between you standing on each pedal and unloaded - that will at least show you if the motor running away is torque sensor related.

edit: just to give you an idea of how much the readings can vary - here's the graph from my oldest motor..

 

[Akita]

Is it possible to change the Torque Value on original firmware 5.0 from VLCD5?
Will there be the same excellent program for the new engine?

 

[electroriderIT]

I tried to calibrate...not had all the weights needed but I don't know if a starting value of 144 with zero kilos..is fine....
here what I got with 7 out of 8 measurements:


a good thing is that right and left pedal are almost identical with a 3 to 5 points difference. but I feel from your values that my sensor doesn't read as much as yours ( you start from a higher value).

Whta do you think? should I put those values in my firmware to calibrate it?
thanks

Thank you! Is there any easy for dummies guide for calibration other than the one on the wiki page?

not really - the hard part is getting the weight on the pedals in decent increments and keeping everything upright. You add weight - you take a reading - repeat

If your start and end values are similar to mine - you could always guess and hope your curve is like mine - or you could just put linear values in between you standing on each pedal and unloaded - that will at least show you if the motor running away is torque sensor related.

edit: just to give you an idea of how much the readings can vary - here's the graph from my oldest motor..

torque sensor - older.png

 

[skestans]

It’s a big gap between 10 and 45 and 70. I’d suggest using weights in between these values and getting rid of the 5kg measurement as it’s too close to the 6kg.

Whatever value you get is what you put into the firmware. Everyone has widely different values and that’s why calibration is necessary. There isn’t much point in comparing values across units.

 

[electroriderIT]

here we go! I calibrated again in a more scientific way
here the results...what do you think?


But SAME BEHAVIOUR of the pedals!!!
I CANT GET RID OF THIS:
[youtube]y4ScIMUBOTk[/YouTube]

as Other informations that may help you or CAsahino himself to help me..is this...I don't have idea if the values recorrect when I make the pedals rotate without weight:
[youtube]09XU7NCepi8[/youtube]

HELPPPPPPPPP

It’s a big gap between 10 and 45 and 70. I’d suggest using weights in between these values and getting rid of the 5kg measurement as it’s too close to the 6kg.

Whatever value you get is what you put into the firmware. Everyone has widely different values and that’s why calibration is necessary. There isn’t much point in comparing values across units.

 

[skestans]

Yeah I have the same on 1.1.0/1.1.1. It's not as bad when you're riding and if you push just 1/8th of a turn the motor will spin less.

In practice, when you're riding, you notice that it spins a bit when pushing the pedals (for instance, approaching a red light and downshifting, have to turn the crank but not enough that it engages and transmits power to the wheel) but it's really not that bothersome. I wouldn't worry about it, personally.

I used 0.6.5 before and it didn't do that so that's an option if you can't live with that behaviour. If you change versions, be aware it will wipe your settings and you'll have to re-enter them (so keep them somewhere safe for the future unless you had too much fun calibrating the sensor and want to do it again in the future)

Also, I'd suggest switching motor control to power instead of torque. I find it a lot smoother personally and closer to how it felt like in 0.6.5.

Finally, don't feel like you calibrated "for nothing". Calibration really does make a difference as it matches your strength intent to the motor's reaction better.

 

[beemac]

here we go! I calibrated again in a more scientific way
here the results...what do you think?
Schermata 2021-05-25 alle 11.28.59.png

But SAME BEHAVIOUR of the pedals!!!
I CANT GET RID OF THIS:

as Other informations that may help you or CAsahino himself to help me..is this...I don't have idea if the values recorrect when I make the pedals rotate without weight:


HELPPPPPPPPP

Calibration values look ok... the ADC looks to be reading correctly - and the pedal weights seem good from the quick glance we get at the variables.

Would be useful to see the ADC and pedal weight result when you put a decent weight on the pedals e.g. stand on one (make sure brakes are pressed!) - and then let go.. should return to the unloaded value pretty quickly - when I had a system with a similar problem - the ADC value took a long time to drop back. I was similarly puzzled and could not get things sorted - so I swapped out the controller in the end.

The software for whatever reason is setting the motor PWM very high - you can see it goes to 96-100% straight away. What level of assist are you using? What values are set for each assist level - have you changed those?

Out of interest - why did you remove and clean the torque sensor in the first place - was there a problem with it?

 

[electroriderIT]

Ok let’s wait for Casainho to put light on this behaviour!

Yes I’ve changed to only 5 level if assistance setting 1000 to level 5 and kind of dividing the values to increase from 0 to 5 ( actually level 6 is 1500)

I serviced the motor to put the temp sensor and thermal pads inside, it works good and I cleaned the too much grease mess that went from the gear to the 3 little springs and torque sensor

here we go! I calibrated again in a more scientific way
here the results...what do you think?
Schermata 2021-05-25 alle 11.28.59.png

But SAME BEHAVIOUR of the pedals!!!
I CANT GET RID OF THIS:

as Other informations that may help you or CAsahino himself to help me..is this...I don't have idea if the values recorrect when I make the pedals rotate without weight:


HELPPPPPPPPP

Calibration values look ok... the ADC looks to be reading correctly - and the pedal weights seem good from the quick glance we get at the variables.

Would be useful to see the ADC and pedal weight result when you put a decent weight on the pedals e.g. stand on one (make sure brakes are pressed!) - and then let go.. should return to the unloaded value pretty quickly - when I had a system with a similar problem - the ADC value took a long time to drop back. I was similarly puzzled and could not get things sorted - so I swapped out the controller in the end.

The software for whatever reason is setting the motor PWM very high - you can see it goes to 96-100% straight away. What level of assist are you using? What values are set for each assist level - have you changed those?

Out of interest - why did you remove and clean the torque sensor in the first place - was there a problem with it?

 

[beemac]

Ok let’s wait for Casainho to put light on this behaviour!

Yes I’ve changed to only 5 level if assistance setting 1000 to level 5 and kind of dividing the values to increase from 0 to 5 ( actually level 6 is 1500)

https://github.com/OpenSourceEBike/TSDZ2_wiki/wiki/Features-and-configurations-on-display#Assist_level__Assist

1000? Isn't that your problem - assist level 1 is usually 0.005...

 

[electroriderIT]

I will try to restore to the normal assist level scale, but I remember that even the 20 levels has 1000 ( or 1500) at level 20...

I hope Casaihno can see my video showing the motor engaging even without pressure on pedals and suggest what can it be and hopefully how to reduce that! :wink:

Ok let’s wait for Casainho to put light on this behaviour!

Yes I’ve changed to only 5 level if assistance setting 1000 to level 5 and kind of dividing the values to increase from 0 to 5 ( actually level 6 is 1500)

https://github.com/OpenSourceEBike/TSDZ2_wiki/wiki/Features-and-configurations-on-display#Assist_level__Assist

1000? Isn't that your problem - assist level 1 is usually 0.005...

 

[hetm4n]

has someone uploaded OSF to the factory sets with the 850c display? it works?
https://a.aliexpress.com/_uxOVZj

 

[skestans]

has someone uploaded OSF to the factory sets with the 850c display? it works?
https://a.aliexpress.com/_uxOVZj

I dont understand what you’re asking.

If you’re asking if you can convert the 850C and TSDZ2 that have factory software over to OSF, then yes it’s possible. Just note you can’t ever put the 850C back to factory software because it can’t be dumped and isn’t publicly available.

 

[mctubster]

has someone uploaded OSF to the factory sets with the 850c display? it works?
https://a.aliexpress.com/_uxOVZj

If its a 2021 850c likely won't work until a patch is applied to the existing code

 

[mbrusa]

If anyone is interested, in the signature link there is also the file 850C_2021_v20.1C.2-bootloader.bin