TSDZ2 OSF for all displays, VLCD5-VLCD6-XH18, LCD3, 860C-850C-SW102.

[hetm4n]

I published on GitHub the pre-release for 860C that I am testing, https://github.com/emmebrusa/TSDZ2-Smart-EBike-860C/tree/master/releases/v20.1C.2-860C, you can enable / disable Field Weakening, so those with temperature problems can check if it depends on that.
There are other interesting changes too, I will prepare a list.

I'm waiting for versions for SW102

 

[mbrusa]

There is also the one in the link, you have to try if it works. I don't have a display SW102.

 

[martin.shane]

HI
I know that 1.01 and 020 there are different firmware's. I try to find out how to set up everything .
Regarding TQ calibration on 0.20.1C . After set up all parameters calibration should be enabled or disabled?
I just don't know how it works.
Disabled- whatever i put there is not in use anyway?
Enabled- It must be enabled to use data from calibration process?
Just wondering.

 

[Elinx]

....
I just don't know how it works.
Disabled- whatever i put there is not in use anyway?
Enabled- It must be enabled to use data from calibration process?
...

Yes, this is how it works.
If you disable the calibration, it uses the default setting (160 range, 67 step)
Enabling after calibration means mostly that there are some lower values.

 

[martin.shane]

Thank U

 

[plpetrov]

If you disable the calibration, it uses the default setting (160 range, 67 step)
Enabling after calibration means mostly that there are some lower values.

I am still not sure which value is which and how to do the calibration. If I interpret correctly your example:
- 160 - is the value of the torque sensor if 0 weight is applied on the pedals.
- 67 steps is the difference between the value with applied certain weight on the pedals and value with no weight (in your example 160).
I am still hesitating if those 67 steps are the value for the full weight of the rider or for some calibrating weight, like it was mentioned previously in the thread - 20,5 kg if I remember well.

 

[Elinx]

.....
- 160 - is the value of the torque sensor if 0 weight is applied on the pedals.
- 67 steps is the difference between the value with applied certain weight on the pedals and value with no weight (in your example 160).
...

-No, 160 is the ideal range of the torquesensor between min and max. value, it could be min/max 0/160 but also 120/280 too.
Every torque sensor has another, mostly smaller, range, so it is adviceable to do the torque software calibration.
The min.value is without weight, max. value is with your weight on the pedal.
With hardware calibration you can improve this range, but you only need this if the torque range is very small.

-No, 67 step is 0,67Nm/ADC bit or 0,4kg/ADCbit (0,17m pedal). If you calibrate this with a weight of 20-25kg it give a new (mostly lower) step value.
It must be such a low weight because the lineairity of the torque sensor is about max. 40kg (mostly lower), so you can see your human pedalforce on display. This calibration isn't neccesary if you aren't interested in knowing the humanpower.

 

[JohnAnanda]

I published on GitHub the pre-release for 860C that I am testing, https://github.com/emmebrusa/TSDZ2-Smart-EBike-860C/tree/master/releases/v20.1C.2-860C, you can enable / disable Field Weakening, so those with temperature problems can check if it depends on that.
There are other interesting changes too, I will prepare a list.

I just installed it on the SW102.
The startup time when turning the display on is much quicker, I don't even have time to read the booting screen !
One issue is that whenever I exit the configuration menu, I always get an error 2 : torque sensor. Switching off and on the display clears the error.
Frequent random freezes in the configuration menu are still there.
I only made a short test run around the block, but it seems the motor is quieter : it's almost silent up to 200W, and starts being noticeable at 300W.. I'm a bit noise obsessed so I'm really happy with that
I haven't played with the new options yet.

Again, thank you so much Mbrusa, Mspider65 and Casainho !

 

[martin.shane]

Mbrusa, Mspider65 and Casainho ! Thank U for U hard work!
Uploaded and is really fast boot. Also i have made 1 min test and it is quiet Heavy rain started so no more fun for now
I see that Hybrid mode is gone. Yes?
FW is switched off.

 

[Mr.Flibble]

I did not explain myself well, my test was not oriented to the temperature, it is the first time I use the sensor, I don't even know if it works well.
The comparison was to verify the response of the motor with Field Weakening enabled or disabled. For how I use the bike, low cadence and low acceleration, I didn't notice any differences.
I don't want to say that Field Weakening is useless, there are those who need to pedal at a high cadence and accept less efficiency and those who do not accept it.
I published on GitHub the pre-release for 860C that I am testing, https://github.com/emmebrusa/TSDZ2-Smart-EBike-860C/tree/master/releases/v20.1C.2-860C, you can enable / disable Field Weakening, so those with temperature problems can check if it depends on that.
There are other interesting changes too, I will prepare a list.

Is there any way to add an option to link Field weakening to rpm ?
ie: Field weakening on/off, if on PWM or Motor RPM option.

 

[mbrusa]

...
I see that Hybrid mode is gone. Yes?
FW is switched off.

Gone where?

Is there any way to add an option to link Field weakening to rpm ?
ie: Field weakening on/off, if on PWM or Motor RPM option.

Field Weakening cannot be linked only to rpm, first the motor must reach PWM max, otherwise efficiency is lost.
If in addition to PWM max it is also bound to a motor rmp threshold, there would be no gradualness.
Field Weakening makes sense if it works well otherwise it's better to disable it.

 

[martin.shane]

Mbrusa- I mean in set up mode under assistance menu, but now I see that hybrid mode can not be set up like others but it is available for use like before so all is ok.
Sorry for that

 

[mspider65]

Field Weakening cannot be linked only to rpm, first the motor must reach PWM max, otherwise efficiency is lost.
If in addition to PWM max it is also bound to a motor rmp threshold, there would be no gradualness.
Field Weakening makes sense if it works well otherwise it's better to disable it.

You are right.
In my latest firmware, FW has no threshold based on motor rpm.
But unlike previous versions, it adds a time offset instead of an angular offset.
This means that the angle that is added varies and increases with the motor rpm.
This contrasts the back EMF generated by the motor which also increases with the motor rpm
For example, when FW reaches the maximum time offset, it adds 2.5 degrees at 2000rpm and 5 degrees at 4000rpm.

 

[mbrusa]

Thank you for the explanation.
Clear and precise.

 

[mbrusa]

Mbrusa- I mean in set up mode under assistance menu, but now I see that hybrid mode can not be set up like others but it is available for use like before so all is ok.
Sorry for that

The assistance parameters of the hybrid mode are the same as in the Torque and Power modes, combined at the same level.

 

[martin.shane]

0.20.1C2 tested in 23 km

Battery - 630 Wh. Start with around 82%. After 16km 62% of battery left.
Speed- adv- 24km/h, most of the time around 33-36 km.
outside temp 11 C
Different modes tested.
FW - disable
Max motor power 600
Max batter cur. 16
Torque calibration on , and off ( i haven't notice too much difference in number and in feel of the bike)
No problem with temp.
Very low noise , only in very high gear and hi cadence but much lower then before( some kind of resonations)
Assist level- most of the time 4.

Only one thing what i have notice that whatever mode i use i do not feel the difference between 4 and 9 assist level. Even in battery cur. and motor power. It drains the same amount . from 5.5 to 12., most of the time around 9 amps
In 0.20.1C1 it was a big difference between 4 and 9. And max speed I could make was 43,5 km/ h on 26 inch bike.
Now max 36 km/h no matter how hard I pedal

 

[blazo]

I published on GitHub the pre-release for 860C that I am testing, https://github.com/emmebrusa/TSDZ2-Smart-EBike-860C/tree/master/releases/v20.1C.2-860C, you can enable / disable Field Weakening, so those with temperature problems can check if it depends on that.

Thanks mbrusa,
OK I am after some temperature testing: My setup : tsdz2 48v, 52v battery, 860c and lcd3 with same connector to swap fast.
I tested bike on tarmac test route 4km uphill. I set the motor limit to 525W, and use power assist on max level to keep the power output around max. Motor temperature on bottom of the hill is the same - 10c, and I note temp on the top. I try to keep cadence and my effort when pedaling the same. I made also tempearture mod because I use my tsdz2 in mountains.

Temperatures:
20.1C.2 (860c) (power limit 525W) with field weakening -- 75c
20.1C.2 (860c) (power limit 525W)without field weakening -- 74c
0.20.0beta1 (lcd3) (power limit 550W, so even a little higher) -- 55c

Than I tested 20.1C.2 (860c) and set up my max power to 400W. And I got 65c but the ride uphill was much slower.

So it is not the field weakening thing.
I really don't know what it can be. Maybe it is something like lower efficiency in sinus controllers vs square controllers in hub motors when rotating top rpm? (I don't know what changes you guys did in firmware, but motor is super quiet now - like sinus vs square). It's just me thinking..

My questions are :
1. Is there any newer firmware out there for 860c display with the motor control code like 0.20beta1? I would really like to use better display.
2. Can anyone make version with the motor control code from 0.20beta1 but all the new features and support for 860c lcd?

Once again thank you all for awesome work, if anyone would like to help some tsdz2 MTB bikers it would be so cool

 

[seattlesockey]

I continue to use and enjoy v20.1c-860C. Thanks mbrusa!

I could use a little help to better understand the Power assist level and Hybrid mode.

The Power assist wiki description refers to “value%/2”. My interpretation is the motor assist power is one-half of the assist level multiplied by the power applied to the pedals. However, the example in wiki indicates assist power is twice the assist level. (100 w @ pedals, 150 assist level = 300 w motor). How does the Power assist level multiply the power at the pedals in order to set the motor power?

I am better off writing the code than the manual
"value% / 2" refers to the value of the assistance level.
"% value" is the% ratio of human power to motor power.
So perhaps it is more understandable:
assistance level value * 2 =% value
motor power = human power * value% / 100
Division by 2, is to hold the maximum value in 8 bits (memory problems).
Attention, to have an acceptable value of human power it is necessary to calibrate with the weight.

My second question is about Hybrid mode. Since Hybrid is a combination of Power and Torque modes, are the assistance levels in Power and Torque intended to be set at the same values for each assistance level?

Not necessarily, they can also be different. You can decide how much each modality contribution should be.

...
By the way Mbrusa, I'd be curious to know around which cadence does the power mode start to be prevalent over torque mode in hybrid.

It is not possible to define a cadence for changing modes, the range is very wide.
It depends on the ratio of the assistance parameters of Torque and Power (at the same level), moreover even after defining the assistance parameters, the mode change point is dynamic, it depends on the force applied to the pedals, the higher the thrust and the higher the cadence of the mode change.
A double change could also occur, with a light push it would switch to Power at a low cadence, then by increasing the push force to maximum, it could switch back to Torque and then again to Power at higher cadence.

@mbrusa - Thanks for the explanation!

 

[Mr.Flibble]

Field Weakening cannot be linked only to rpm, first the motor must reach PWM max, otherwise efficiency is lost.
If in addition to PWM max it is also bound to a motor rmp threshold, there would be no gradualness.
Field Weakening makes sense if it works well otherwise it's better to disable it.

You are right.
In my latest firmware, FW has no threshold based on motor rpm.
But unlike previous versions, it adds a time offset instead of an angular offset.
This means that the angle that is added varies and increases with the motor rpm.
This contrasts the back EMF generated by the motor which also increases with the motor rpm
For example, when FW reaches the maximum time offset, it adds 2.5 degrees at 2000rpm and 5 degrees at 4000rpm.

So that is effectively linking FW to motor RPM ?

That solution sounds like it would solve the issue of losing power on steep climbs where your cadence drops.

 

[Mr.Flibble]

.....
- 160 - is the value of the torque sensor if 0 weight is applied on the pedals.
- 67 steps is the difference between the value with applied certain weight on the pedals and value with no weight (in your example 160).
...

-No, 160 is the ideal range of the torquesensor between min and max. value, it could be min/max 0/160 but also 120/280 too.
Every torque sensor has another, mostly smaller, range, so it is adviceable to do the torque software calibration.
The min.value is without weight, max. value is with your weight on the pedal.
With hardware calibration you can improve this range, but you only need this if the torque range is very small.

-No, 67 step is 0,67Nm/ADC bit or 0,4kg/ADCbit (0,17m pedal). If you calibrate this with a weight of 20-25kg it give a new (mostly lower) step value.
It must be such a low weight because the lineairity of the torque sensor is about max. 40kg (mostly lower), so you can see your human pedalforce on display. This calibration isn't neccesary if you aren't interested in knowing the humanpower.

I'm recalibrating my TQ sensor after changing the controller, and I don't understand what the ADC step figure is, or how it is obtained either.
Is it the range between no weight on the pedal and the 20-25 kg weight reading?
ie: zero weight reading minus 20-25 kg weight reading = ADC step figure?

 

[seattlesockey]

.....
- 160 - is the value of the torque sensor if 0 weight is applied on the pedals.
- 67 steps is the difference between the value with applied certain weight on the pedals and value with no weight (in your example 160).
...

-No, 160 is the ideal range of the torquesensor between min and max. value, it could be min/max 0/160 but also 120/280 too.
Every torque sensor has another, mostly smaller, range, so it is adviceable to do the torque software calibration.
The min.value is without weight, max. value is with your weight on the pedal.
With hardware calibration you can improve this range, but you only need this if the torque range is very small.

-No, 67 step is 0,67Nm/ADC bit or 0,4kg/ADCbit (0,17m pedal). If you calibrate this with a weight of 20-25kg it give a new (mostly lower) step value.
It must be such a low weight because the lineairity of the torque sensor is about max. 40kg (mostly lower), so you can see your human pedalforce on display. This calibration isn't neccesary if you aren't interested in knowing the humanpower.

I'm recalibrating my TQ sensor after changing the controller, and I don't understand what the ADC step figure is, or how it is obtained either.
Is it the range between no weight on the pedal and the 20-25 kg weight reading?
ie: zero weight reading minus 20-25 kg weight reading = ADC step figure?

@Mr.Flibble - Go to the Technical menu and read "ADC torque step calc" and enter that value in Torque sensor menu under "Torque ADC step". Do this after calibrating torque sensor.

 

[Elinx]

......

I'm recalibrating my TQ sensor after changing the controller, and I don't understand what the ADC step figure is, or how it is obtained either.
Is it the range between no weight on the pedal and the 20-25 kg weight reading?
ie: zero weight reading minus 20-25 kg weight reading = ADC step figure?

As I said before, the torque sensor is not lineair. So if you want an exact display reading of your humanpower (on the pedals) it is neccesary to know which reading you get with which force.
If you see the wiki, you understand with this diagram what I mean. The left curve is linear till about 25kg.

If you see the left diagram you see a total a min.value of 292 and max. value of 386, so the range of the torque sensor is 386-292 = 94, which is a low value compared with the default 160, I mentioned before.
The rise of the curve at the start 0-25kg is a lot more than the end 25-80kg. The measurements are divided in steps.

With 67 we have a rise of 0.67Nm/adc-step, meaning 0,67÷9.81÷0,17 = 0.4kg/adc-step for the first 25kg (0.17 is the length of the pedalcrank 17cm)

For the display reading, of human power, you calibrate with a weight 20-25kg.
For the step calculation, for this case, I take 19kg out of the table.
360-292= 68 adc-steps for 19kg, that is 19÷68 = 0.28kg/step and 0.28x9.81x0.17 = 0.46Nm/adc-step, so for this torquediagram I should configure 46, which is a lot lower than the default 67.

I hope you understand a bit more about the torque sensor calibration and why it is needed.

(edit: notation symbols changed)

 

[Mr.Flibble]

......

I'm recalibrating my TQ sensor after changing the controller, and I don't understand what the ADC step figure is, or how it is obtained either.
Is it the range between no weight on the pedal and the 20-25 kg weight reading?
ie: zero weight reading minus 20-25 kg weight reading = ADC step figure?

As I said before, the torque sensor is not lineair. So if you want an exact display reading of your humanpower (on the pedals) it is neccesary to know which reading you get with which force.
If you see the wiki, you understand with this diagram what I mean. The left curve is linear till about 25kg.

If you see the left diagram you see a total a min.value of 292 and max. value of 386, so the range of the torque sensor is 386-292= 94, which is a low value compared with the default 160, I mentioned before.
The rise of the curve at the start 0-25kg is a lot more than the end 25-80kg. The measurements are divided in steps.

With 67 we have a rise of 0,67Nm/adc-step, meaning 0,67/9,81/0,17=0,4kg/adc-step for the first 25kg (0,17 is the length of the pedalcrank 17cm)

For the display reading, of human power, you calibrate with a weight 20-25kg.
For the step calculation, for this case, I take 19kg out of the table.
360-292= 68 adc-steps for 19kg, that is 19/68= 0,28kg/step 0,28x9,81x0,17= 0,46Nm/adc-step, so for this torque diagram I should configure 46, which is a lot lower than the default 67.

I hope you understand a bit more about the torque sensor calibration and why it is needed.

Thx.

So the answer to my question is "yes the adc step figure is no weight on the pedal reading minus 20 kg on the pedal reading" ?

Is "/" a "÷" ?
Is "," a "." ?
Where has 9.81 come from?

 

[mbrusa]

Thanks mbrusa,
OK I am after some temperature testing: My setup : tsdz2 48v, 52v battery, 860c and lcd3 with same connector to swap fast.
I tested bike on tarmac test route 4km uphill. I set the motor limit to 525W, and use power assist on max level to keep the power output around max. Motor temperature on bottom of the hill is the same - 10c, and I note temp on the top. I try to keep cadence and my effort when pedaling the same. I made also tempearture mod because I use my tsdz2 in mountains.

Temperatures:
20.1C.2 (860c) (power limit 525W) with field weakening -- 75c
20.1C.2 (860c) (power limit 525W)without field weakening -- 74c
0.20.0beta1 (lcd3) (power limit 550W, so even a little higher) -- 55c

Than I tested 20.1C.2 (860c) and set up my max power to 400W. And I got 65c but the ride uphill was much slower.

So it is not the field weakening thing.
I really don't know what it can be. Maybe it is something like lower efficiency in sinus controllers vs square controllers in hub motors when rotating top rpm? (I don't know what changes you guys did in firmware, but motor is super quiet now - like sinus vs square). It's just me thinking..

My questions are :
1. Is there any newer firmware out there for 860c display with the motor control code like 0.20beta1? I would really like to use better display.
2. Can anyone make version with the motor control code from 0.20beta1 but all the new features and support for 860c lcd?

Once again thank you all for awesome work, if anyone would like to help some tsdz2 MTB bikers it would be so cool

Well thanks for the test
Immediately a clarification, the comparison between v20.1C and 0.20beta1-LCD3, is not on equal terms, I have already written it.
The ADC conversion parameters of the current are different.
In all recent versions, one ADC step equals 0.156 Amp, in 0.20beta1-LCD3 it is 0.17 Amp.
The difference is 10%, that's not a little bit.
There is however another parameter with 20% difference and it is for the phase current of the motor used in the calculation of the FOC angle.
I will have to do some tests to understand how much it can affect the actual current / power.
Surely the current / power limit of the 0.20beta1-LCD3 version, at the same value, is lower than v20.1C.

Another consideration, it is not enough to set the power limit, the limit current is also important, it must be adapted to the motor / battery voltage, with the new version this limit is reached more quickly.

The test method must also be improved, in addition to the maximum power, the time taken or the average speed should also be considered. It is to understand where the extra current / power absorbed by the battery goes, because if it goes to the wheel it is not a defect, just take it into account by adjusting the assistance level.

Personally I am very satisfied with this new version of mspider65, in addition to being quieter the motor, in my opinion it consumes less than the previous one.

In these days I have done some tests on a route already done many times, in my way of using the bike at 50/60 rpm without forcing, with 10°C external the motor reaches a maximum of 40°C and I think the sensor works correctly.
One day I increased the assistance by two levels to try to keep the power close to the 500W limit, I only got there in steep uphill sections.
I consumed 50% more battery, on 22 km the average speed was 18 km/h instead of the usual 14 km/h (ascent and descent), the max temperature 50°C instead of 40°C, but contrary to what I expected I did much more effort.
Pedaling uphill over 80 rpm has crushed my legs, having the limit power as a goal to maintain is not for me.

No I'm not going back, the temperature is a problem only in conditions of high power, it is not the only aspect to evaluate.
I'll try to understand and improve if I can, but I'm not going back.

 

[Mr.Flibble]

.....
- 160 - is the value of the torque sensor if 0 weight is applied on the pedals.
- 67 steps is the difference between the value with applied certain weight on the pedals and value with no weight (in your example 160).
...

-No, 160 is the ideal range of the torquesensor between min and max. value, it could be min/max 0/160 but also 120/280 too.
Every torque sensor has another, mostly smaller, range, so it is adviceable to do the torque software calibration.
The min.value is without weight, max. value is with your weight on the pedal.
With hardware calibration you can improve this range, but you only need this if the torque range is very small.

-No, 67 step is 0,67Nm/ADC bit or 0,4kg/ADCbit (0,17m pedal). If you calibrate this with a weight of 20-25kg it give a new (mostly lower) step value.
It must be such a low weight because the lineairity of the torque sensor is about max. 40kg (mostly lower), so you can see your human pedalforce on display. This calibration isn't neccesary if you aren't interested in knowing the humanpower.

I'm recalibrating my TQ sensor after changing the controller, and I don't understand what the ADC step figure is, or how it is obtained either.
Is it the range between no weight on the pedal and the 20-25 kg weight reading?
ie: zero weight reading minus 20-25 kg weight reading = ADC step figure?

@Mr.Flibble - Go to the Technical menu and read "ADC torque step calc" and enter that value in Torque sensor menu under "Torque ADC step". Do this after calibrating torque sensor.

That's in the technical section on the 860c display?
If the machine already knows the figure, why do we need to enter it manually?

Thx