BAFANG ULTRA: Torque sensor transformation curve

AndZab

100 mW
Joined
May 27, 2020
Messages
40
Hello, there is something a little unclear about the Ultra torque sensor calibration.
It seems that with no pressure on pedals, base input voltage is around 750 mV (TqVoltate)
Said that, it looks like default torque settings have no sense at all.
Infact, default do not take care of the sensor base voltage, that is around 750 mV.
If I well understood, with defaults is like to press 18,75 kg as you move the pedals with minimum force.
This can mean too much power at wheel when start pedaling.

Michael's Frey well studied this fact, you can read his conclusions and "smooth settings" at this link:

http://frey-bafang.patransformers.com/2020/10/09/bafang-ultra-programming/

You can compare the two sensor calibration settings, where Michael correct Bafang mistake:

Ultra_CompareDefault-MichaelDATA.png

But you can better see the difference with the 2 conversion curves comparing:

Ultra_CompareDefault-Michael.png

What you do think about all these calibration curves ?
Regards
 
2 thoughts. First, it's not hard to improve on the standard Bafang settings with your own preferences in mind.

Second, keep in mind we are all going to have our own set of 'druthers when it comes to the Ultra, or any of the BBSxx series for that matter.

There is not going to be a generic set of parameters that will completely satisfy all riders. For instance, there's too big a difference in what a road bike rider and a hard mountain bike rider are looking for out of their bikes. Compared to them, my 'druthers are way more about low speed response!

From what I can see, reading through some of the stuff written by those that will likely have forgotten more than I'll ever know on the topic, there is no consensus when it comes to "perfect" or the hot setup most likely to work for everyone.

For those with the patience, I think a focus on changing just parameters to suit them personally will work out best in the end. That's me though. -Al
 
AHicks said:
2 thoughts. First, it's not hard to improve on the standard Bafang settings with your own preferences in mind.

Second, keep in mind we are all going to have our own set of 'druthers when it comes to the Ultra, or any of the BBSxx series for that matter.

There is not going to be a generic set of parameters that will completely satisfy all riders. For instance, there's too big a difference in what a road bike rider and a hard mountain bike rider are looking for out of their bikes. Compared to them, my 'druthers are way more about low speed response!

From what I can see, reading through some of the stuff written by those that will likely have forgotten more than I'll ever know on the topic, there is no consensus when it comes to "perfect" or the hot setup most likely to work for everyone.

For those with the patience, I think a focus on changing just parameters to suit them personally will work out best in the end. That's me though. -Al

I understand your point, but I'm referring to another matter.
In my opinion, one thing is the calibration of the sensor (done by a customizable curve), and another is personal assist tastes.

The first, is only about the conversion from sensor mV to real Kg force you are pressing on the pedal.
This has to be precise, as I see it, or you'll not understand the following step.

I mean, the following area where you input the "personal tastes", is this one:

Ultra_TorqueAssistSetup.png

Maybe all this confusion about this matter, is because crank SpdXX parameters work over a bad calibrated curve.
That's may opinion, anyway...
Regards
 
Ok, I see what you are talking about now. You've charted both sets of figures (Mike's and Bafang's) to see what they look like on a graph. Interesting!

I have to ask though, where are you getting the mV values you are using on the graph?
 
AndZab said:
Hello, there is something a little unclear about the Ultra torque sensor calibration.
It seems that with no pressure on pedals, base input voltage is around 750 mV (TqVoltate)
Said that, it looks like default torque settings have no sense at all.
Infact, default do not take care of the sensor base voltage, that is around 750 mV.
If I well understood, with defaults is like to press 18,75 kg as you move the pedals with minimum force.
This can mean too much power at wheel when start pedaling.

Michael's Frey well studied this fact, you can read his conclusions and "smooth settings" at this link:

http://frey-bafang.patransformers.com/2020/10/09/bafang-ultra-programming/

You can compare the two sensor calibration settings, where Michael correct Bafang mistake:

Ultra_CompareDefault-MichaelDATA.png

But you can better see the difference with the 2 conversion curves comparing:

Ultra_CompareDefault-Michael.png

What you do think about all these calibration curves ?
Regards

1. Bafang OEM setting was not activating PAS till 750mV. So changing base voltage is useful
2. Voltage span vs pedal Kg is basically sensitivity adjustment. Since motor is very powerfull, smooth operation is achieved by programming larger voltage span over same Kg range.
3. Another layer of complexity is that MAX Kg force applied by rider will greatly change with cadence and where in min-max of muscle range rider is. Example: for your body it's easy to control 1-2kg force but not 10-11kg.

programming is improvement over OEM but not linear and not in line with sensor output. To calibrate it as a scale, just put know weights on pedals and record readings. Just remember, it will read differently depending on how pedals are positioned (LOL).

Way too many variables to summarise.
 
AndZab said:
AHicks said:
2 thoughts. First, it's not hard to improve on the standard Bafang settings with your own preferences in mind.

Second, keep in mind we are all going to have our own set of 'druthers when it comes to the Ultra, or any of the BBSxx series for that matter.

There is not going to be a generic set of parameters that will completely satisfy all riders. For instance, there's too big a difference in what a road bike rider and a hard mountain bike rider are looking for out of their bikes. Compared to them, my 'druthers are way more about low speed response!

From what I can see, reading through some of the stuff written by those that will likely have forgotten more than I'll ever know on the topic, there is no consensus when it comes to "perfect" or the hot setup most likely to work for everyone.

For those with the patience, I think a focus on changing just parameters to suit them personally will work out best in the end. That's me though. -Al

I understand your point, but I'm referring to another matter.
In my opinion, one thing is the calibration of the sensor (done by a customizable curve), and another is personal assist tastes.

The first, is only about the conversion from sensor mV to real Kg force you are pressing on the pedal.
This has to be precise, as I see it, or you'll not understand the following step.

I mean, the following area where you input the "personal tastes", is this one:

Ultra_TorqueAssistSetup.png

Maybe all this confusion about this matter, is because crank SpdXX parameters work over a bad calibrated curve.
That's may opinion, anyway...
Regards

I think there are two different settings.
One is Torque VS pedal force (PAS)
Second is PAS VS cadence speed (RPM)

They both serve very different scenarios.
Sensor calibration is universal variable across all cadences. While PAS vs RPM is very specific to cadance.

Using both those variables you can fine tune system. Setting torques sensor to reflect ideal reading, does not improve tune, I tried it :) It's very linear, easy to set. Just use weights put on pedals and record mV output. Then change sensor calibration to reflect actual values. You get very true weight sensor but not very useful due to human pedal input vs PAS expectation been non-linear in nature. Some general read on the side about G510, Frey EX Pro long term review
 
2Slow said:
[
I think there are two different settings.
One is Torque VS pedal force (PAS)
Second is PAS VS cadence speed (RPM)

They both serve very different scenarios.
Sensor calibration is universal variable across all cadences. While PAS vs RPM is very specific to cadance.

Using both those variables you can fine tune system. Setting torques sensor to reflect ideal reading, does not improve tune, I tried it :) It's very linear, easy to set. Just use weights put on pedals and record mV output. Then change sensor calibration to reflect actual values. You get very true weight sensor but not very useful due to human pedal input vs PAS expectation been non-linear in nature

Hi, I believe we understood each other...

My procedure can be this: first calibrate sensor curve, putting known weights on horizontal cranck.
For example, 1,5 liter water bottles. 1 bottle, 2 bottles, 1 six pack, and so on.

When I have my calibrated curve, then regulate the cadence PAS, for all the 5 cadences.
Then, if unsatisfied, go back to sensitivity curve, and apply a multiplier to values, to increase or decrease sensitivity.
But you have to begin with a calibrated curve, or you'll understand nothing of what is happening...

Very interesting is the Base Voltage. It has to be put a little less of TqVoltate, so Kg pressure start from around zero, so you don't get start "bumps".
Hope you agree with that.

Regards
 
AHicks said:
Ok, I see what you are talking about now. You've charted both sets of figures (Mike's and Bafang's) to see what they look like on a graph. Interesting!

I have to ask though, where are you getting the mV values you are using on the graph?

I symply put on graph the values of Michael and Default, adding Delta Voltage on the X axis.
The big difference is that on Michael, Y points start from 0 Kg, and on Default Y points start from 18 Kg because Base Voltage is wrong.

This is what I understood, maybe I'm wrong...
Regards
 
AndZab said:
2Slow said:
[
I think there are two different settings.
One is Torque VS pedal force (PAS)
Second is PAS VS cadence speed (RPM)

They both serve very different scenarios.
Sensor calibration is universal variable across all cadences. While PAS vs RPM is very specific to cadance.

Using both those variables you can fine tune system. Setting torques sensor to reflect ideal reading, does not improve tune, I tried it :) It's very linear, easy to set. Just use weights put on pedals and record mV output. Then change sensor calibration to reflect actual values. You get very true weight sensor but not very useful due to human pedal input vs PAS expectation been non-linear in nature

Hi, I believe we understood each other...

My procedure can be this: first calibrate sensor curve, putting known weights on horizontal cranck.
For example, 1,5 liter water bottles. 1 bottle, 2 bottles, 1 six pack, and so on.

When I have my calibrated curve, then regulate the cadence PAS, for all the 5 cadences.
Then, if unsatisfied, go back to sensitivity curve, and apply a multiplier to values, to increase or decrease sensitivity.
But you have to begin with a calibrated curve, or you'll understand nothing of what is happening...

Very interesting is the Base Voltage. It has to be put a little less of TqVoltate, so Kg pressure start from around zero, so you don't get start "bumps".
Hope you agree with that.

Regards

Agree.
Base Voltage must be below TqVoltate, logically. But worth testing above as well, to see how software operates.
 
I was confused too about the base voltage setting. It seems that every bike manufacturer leave this parameter to 0. I figured out why. When it is set to 0, it means automatic calibration everytime you power up the controller. So if you put pressure on the pedals before you turn it on it will read a higher value and use this as the base voltage. It means that the sensor become much less sensitive. The main advantage of setting the base voltage is that it prevent bad calibration.
 
Back
Top