TSDZ2 mid drive with 860C, 850C or SW102 displays only -- Flexible OpenSource firmware (Casainho code only)

buba said:

I have added so many functions and optimized so much space it is insane. But there is a limit. Trying to make the firmware react so good that we do not need Startup Boost and get some space back.

Click to expand...

flufferty said:

I may be completely out of the loop here, but why do we need the human power to be accurate?
It seems that all the TSDZ2-units have differently calibrated sensors, and perhaps the stock firmware is compensating for that (undocumented eeprom-data?), perhaps it isn't.

Couldn't we include a calibration routine that takes the resting value, and then records the highest torque value read, and uses that for dynamically applying the power-factor. (given we don't need real-watt human power, we only need a deterministic torque-sensor range)

The firmware could continuously save the highest torque recorded, and use that. So the harder you press your bike (torque sensor range allowing) the higher you'll raise the torque-range (until its saturated, or you cant push it more).
So given a "well-calibrated" physical torque sensor, you will get a larger/finer range, but someone with a badly calibrated physical torque sensor, will have the same range, but with worse resolution.

If needed, perhaps a new "weaker" person wants to use THAT motor, you reset the calibrated value in a menu-option.
With the above option of "dynamic" high-point, a child could ride the motor and enjoy it's full benefits, even without reaching the 40kg or so where the torque sensors seems to max out.

Click to expand...

jimmyfergus said:

buba said:

I have added so many functions and optimized so much space it is insane. But there is a limit. Trying to make the firmware react so good that we do not need Startup Boost and get some space back.

Click to expand...

I definitely think Boost seems like a band-aid that can be discarded if you improve the responsiveness. Personally I would be very happy to see it dropped. It seems to be a frequent pattern with control systems that they accumulate band-aids to try to fix behavior where if you get back to the fundamentals you can avoid them altogether.

Click to expand...

jimmyfergus said:

I'm assuming that you're also improving the slow-down responsiveness. In 0.19 the run-on is pretty strong, feels like ~1s after stopping pedaling, and to me at high assist even feels like a ramp up and then sudden stop. This is a problem for me with an internal gear hub, where gear shifts are very abrupt - unlike a derailleur when the chain ramps between cogs - I'm pretty sure if my shift comes through under power I'm really stressing that poor blue gear, but avoiding it means a very long, slow gearshift.

Click to expand...

jimmyfergus said:

I'm also hoping whatever causes set-off to lag sometimes will go away. I'm not running a brake cut-off at present, and I'm not sure if that contributes, but a lot of times I get nothing much for the first revolution or two on set-off, and then a slow ramp-up. I think it's related to having a foot on the pedals when stationary.

I'm looking forward to your new beta - though I'll probably be on vacation.

What recommended sdcc version works for the codebase? I tried a few and got various SEGVs and other issues during compilation.

Thanks for your efforts!

Click to expand...

buba said:

The torque sensor will be calibrated during power on and establish a resting value. After this, the more you press on the pedals the more power you will get. Simple as that. And if you have some user having problems with applying enough force you simply adjust the assist level multipliers so it suits that user.

Click to expand...

casainho said:

buba said:

The torque sensor will be calibrated during power on and establish a resting value. After this, the more you press on the pedals the more power you will get. Simple as that. And if you have some user having problems with applying enough force you simply adjust the assist level multipliers so it suits that user.

Click to expand...

There is an issue here. My torque sensor could measure only 10 kgs and after that was very not linear.

So, the first 5 kgs was my legs resting than just another 5 kgs for measure user actual leg force, which is almost nothing, so user will simple not get any assistance by putting more force on the pedals because the sensor saturates very early -- it is like if sensor is faulty.

Click to expand...

buba said:

casainho said:

buba said:

The torque sensor will be calibrated during power on and establish a resting value. After this, the more you press on the pedals the more power you will get. Simple as that. And if you have some user having problems with applying enough force you simply adjust the assist level multipliers so it suits that user.

Click to expand...

There is an issue here. My torque sensor could measure only 10 kgs and after that was very not linear.

So, the first 5 kgs was my legs resting than just another 5 kgs for measure user actual leg force, which is almost nothing, so user will simple not get any assistance by putting more force on the pedals because the sensor saturates very early -- it is like if sensor is faulty.

Click to expand...

Correct, for that case I have added an error code warning users that something might be wrong with the torque resolution. The new wiki explains that the user should:

1. Not press the pedals during calibration

2. Follow your tutorial and adjust the torque sensor as it needs to be calibrated

So if the resting torque is higher than 64 (value can be changed, just chose 64 for now) it is due to the user applying force on the pedals or the torque sensor simply needs calibration. I think we should advise users in the wiki to check their sensor if they feel that the response is lacking. And if it is above a certain level (probably higher than 64) the sensor should be calibrated.

What do you think?

Click to expand...

fi7ippo said:

dameri said:

fi7ippo said:

I am very happy that you like the idea, but I was thinking, and I believe that if it is not too much trouble, and there is enough memory space, the ideal would be to have two parameters that can be changed by the user:

1- User calibration weight available with an approximation of 0.1 or 0.5 kg (this is because it would be difficult for everyone to have an identical weight to be applied for the ADC reading);

Click to expand...

Excuse me for coming to this technical discussion. How does the driver's weight work? At least I sit in the saddle practically every time I cycle. So I don't push my pedal at full weight almost ever. And I think many others do the same when driving on an electric bike.

Click to expand...

Hi Dameri, maybe I didn't express myself correctly!
The fact is that to calibrate the torque sensor (assuming it is now linear) we need at least 2 points. The first is the value at rest detected during power on, we must enter the second which apparently varies from sensor to sensor.

Now, we can perform an ADC reading with a defined weight applied to the crank (example 20 - 30kg), or, since it would not be easy for everyone to have the same calibration weight to apply at the crank to perform the second reading, we can have 2 parameters to configure, the weight applied during calibration and the ADC reading with applied weight! In this way, since a straight line passes only by two points each one would have its own sensor perfectly calibrated regardless of the weight used for calibration!

Click to expand...

thineight said:

Hello casainho, i've just updated the wiki for the brake sensor installation
https://github.com/OpenSource-EBike-firmware/TSDZ2_wiki/wiki/How-to-install-brake-sensors
I am not sure I did the procedure correctly, anyway I added the note on top to pay attention during the boot to make sure the brake sensors are correctly installed and working, in addition I added the schematic with a good solution to keep the sensor disconnectable.

Unfortunately I think I cannot add pictures on the wiki, therefore I kindly ask you to add the following one to your wiki in order to complete the explaination.
Many thanks.

brake_sensor_schematic.png

Click to expand...

casainho said:

Done, the file is here: https://github.com/OpenSource-EBike-firmware/TSDZ2_wiki

Yes, files need to be upload to other place and works differently.

Click to expand...

skestans said:

I am new to this, and I'm wondering: is the Open Source firmware compatible with the 850c colors display, or is it only with the KT-LCD3? Are there any plans to make the 850c compatible with it?

Click to expand...

casainho said:

skestans said:

I am new to this, and I'm wondering: is the Open Source firmware compatible with the 850c colors display, or is it only with the KT-LCD3? Are there any plans to make the 850c compatible with it?

Click to expand...

I hope to release in a few days a first version for 850C, here is a picture I took recently from the one I am riding in my bicycle:

Click to expand...

thineight said:

Hello everyone,
this morning I borrowed a digital luggage scale and I made some quick tests on the force/sensor response.

It has a sort of hook, therefore I put the bike upright in normal position with the brake pressed, pedal horizontal (no laser level though :wink and force applied upwards on the pedal at the back (right).
I flipped the bike, rotated the pedals 180° and repeated the measurements on the left pedal just to have 2 sets of values.
The precision may not be that accurate but I think it gives the idea.

My rest value, read from menu 9.2, is about 50. To be noted that it was 48-49 few weeks ago when buba asked the first time.

Here below my chart, i think I can conclude that it can linearly appreciate about 20 kg.

Hope more data helps the development.
Cheers

Click to expand...

Popo15 said:

Hello good afternoon Buba, finished riding an engine with Throttle, I confirm that the Throttle does not work with version v.19 beta8 I have tested with previous versions example version Throttle v0.16.0 working, thank you very much for your work.

Click to expand...

casainho said:

I got my torque sensor well calibrated and even saved the max value of each pedal rotation and then applied the sinewave average factor, so, I hope this value is correct.

I had a bug on 850C that took me some days to solve... but finally the graph shows correctly the pedal human power and with that white line as a contour.

I had to sweat to get that 300W peak!! I had to put all my body of 100 kgs on the pedals and pedal hard up to get some decent cadence. Since now my torque sensor measures only up to 40 kgs and there is not yet an exploitation of the non linear zone, I guess the peak should be more like double of that value... Anyway, for pedaling seated, I think the values are correct. I can do between 100 to 150 watts continuous for maybe like 30 minutes.
I changed the gears and even if my cadence increase or decrease, the power measured were almost the same as I was trying to keep it the same, feeling that I was doing the same effort/work, so I think it really works well.

Note that black vertical lines are due to the mobile camera recording frequency.

Buba, I have the code ready to move to your code.

Maybe you are wasting time doing all that optimizations on LCD3, I guess it will not handle the options for torque sensor calibration (including the non linear range).

Click to expand...

buba said:

EDIT: One alternative is that you do not commit your changes and do not accept my pull request yet. I can add your changes in my branch and then you can accept the entire pull request when everything is finsihed. This would be the best for me as I then have more time to finish my development. And when I am finished you can accept the pull request and immediately test my changes in combination with your changes. Everything will work! Can we do that? I can add the changes you have done in the branch: pedal_power_average in the coming days.

If that works with you I would be very grateful!

Click to expand...

casainho said:

I want to release today firmware version for 850C LCD, for motor controller firmware version 0.19.0.

Click to expand...

casainho said:

buba said:

EDIT: One alternative is that you do not commit your changes and do not accept my pull request yet. I can add your changes in my branch and then you can accept the entire pull request when everything is finsihed. This would be the best for me as I then have more time to finish my development. And when I am finished you can accept the pull request and immediately test my changes in combination with your changes. Everything will work! Can we do that? I can add the changes you have done in the branch: pedal_power_average in the coming days.

If that works with you I would be very grateful!

Click to expand...

yes, I put my latest code on that branch. Get it from there.

Some notes:
- I decided to calculate in parallel the human pedal power value to be shown on LCD as the one used to control the motor do works well and I didn't to risky what is working well
- the calibration value is for my girlfriend bicycle, that seams to have a double sensitivity torque compared to others

Click to expand...

casainho said:

I want to release today firmware version for 850C LCD, for motor controller firmware version 0.19.0.

Click to expand...

buba said:

Popo15 said:

Hello good afternoon Buba, finished riding an engine with Throttle, I confirm that the Throttle does not work with version v.19 beta8 I have tested with previous versions example version Throttle v0.16.0 working, thank you very much for your work.

Click to expand...

Hello Popo15,

Have you enabled the throttle in the Configuration Menu under the Motor Controller Setup?

https://github.com/OpenSource-EBike-firmware/TSDZ2_wiki/wiki/Features-and-configurations-for-version-0.19.X#7_Motor_Controller_Setup

Click to expand...

Popo15 said:

buba said:

Popo15 said:

Hello good afternoon Buba, finished riding an engine with Throttle, I confirm that the Throttle does not work with version v.19 beta8 I have tested with previous versions example version Throttle v0.16.0 working, thank you very much for your work.

Click to expand...

Hello Popo15,

Have you enabled the throttle in the Configuration Menu under the Motor Controller Setup?

https://github.com/OpenSource-EBike-firmware/TSDZ2_wiki/wiki/Features-and-configurations-for-version-0.19.X#7_Motor_Controller_Setup

Click to expand...

good morning, if enabled and it does not work, thank you very much.

Click to expand...