New "TSDZ2 Torque Sensor Central Motor"

[casainho]

Thank you link is OK now. There is just one thing: The new display firmware (see the post above with the bar graph) is also not available. Can you also check this?

That is not my firmware and I never tested it. You need to ask questions to his author.

 

[LeftCoastNurd]

its a shame this project has become so fragmented.

i've got a TSDZ2 w/a 860C, and am running 1.1.0, but when I poke around the various githubs now, i'm seeing like 0.56 and stuff? are those older or newer ?

 

[RichardPH]

hi, I've got two bikes with 350W 36V TSDZ2 both operating with the same Hailong 10s 4p batteries. One was purchased in 2017 the other 2020. The newer one appears to have its sweet spot at a higher cadence, which I prefer, when compared to the outwardly identical earlier one. Is this just my imagination or are they different in some way?

Also related, does the 250W version have the same/higher/lower cadence relationship?

 

[Elinx]

..two bikes with 350W 36V TSDZ2 ... 2017 the other 2020. The newer one appears to have its sweet spot at a higher cadence, ....

FYI there is no differnce between the 250W and 350W version. They are equal
Tsdz2 has only a 36V 400rpm and 48V 4000rpm version.

Imho there could be two reasons for that difference.

1. The 2017 version has for sure the older 8 bit version of the controller unit and the 2020 version could have the newer improved 32bit controller version which allow a higher cadence.
2. With all tsdz2 versions the torquesensor make the difference in response of the engine. That could be a reason that you feel a differnet behaviour.

NB
1. It is possible to measure the engine-plug for the speedsensor to know if you have an old or new controller.
2. If you have a Vlcd5 you can see the difference of the torque sensitivity inside the hidden menu.

 

[RichardPH]

Thanks Elinx. Until now I've been happy to just ride the 2017 version, but the possibility of changing things has made the [mechanical] engineer in me curious. Snag, I've not the first idea how to start, but your links have given me a bit to read and maybe work out the right questions to ask. Don't want to be a nuisance

 

[gordone]

What is latest recommended firmware to use for motor and display?

Display I will use is 850c/860c.

I understand latest firmware don’t support these display so wondering latest to use?

 

[casainho]

Great looking new display OpenSource firmware specific for TSDZ2, original message from author:

Hello everyone.

I wasn't satisfied with the open source UI for the SW102 display, so I've made a different version. Functionally it's the same but it looks better IMHO. I encourage you to try it
It's mostly compatible with casainho's original code - you can switch between my version and the 1.1.0 release with OTA update and your settings will be preserved.

This is the addres of my github repo: https://github.com/anszom/SW102_LCD . You can find slightly more information there, including short usage and installation instructions. A prebuilt update image is available on the releases page.

Here's an animated demo of the firmware' s functionality (this was recorded with an "emulator" build. It's difficult enough to take good pictures of an OLED, let alone do it while riding a bike. The real display runs as smooth as you can see in the gif. And below you can see some actual photos.

Main screen. Please disregard the butchered display casing, I've had to entirely rebuild the damn thing after I've fried the CPU with wrong voltages.

Config menu:

Assist level configuration:

Torque sensor calibration:

As a bonus, it should be easy to port this to casainho's DIY display, as the electronics are quite similar.

 

[Elinx]

What is latest recommended firmware to use for motor and display?

Display I will use is 850c/860c......

There could be problems with different versions of the 850C and SW102 display, but imho the 860C display is not a problem.
The latest version is made by mbrusa, which you can find here.

 

[gordone]

....
But it looks like my assembly is not identical?

Ofcourse it isn't, because you have dismantled the wrong side of the motor for the first 7 steps.

The side you are showing are the optional last two pictures on the "how to" step 8, to illustrate a bit more cooling by shortening the plastic cover.

I really ask myself, if it is wise that you do such a modification.
There are easier methodes to improve the heat dissipation by just filling the gap between motor and case cover with conductive pads as Wpenner did here.

hehe, thank you.

 

[anszom]

Yes, the SW102 is not supported anymore. The only display is this OpenSource one: https://opensourceebike.github.io/

I think that's going to be a problem for many would-be users. Re-programming an off-the-shelf display is one thing, but building one from scratch is likely going to be a problem for many. That is, unless you are planning to sell ready-to-use display kits?

 

[casainho]

Yes, the SW102 is not supported anymore. The only display is this OpenSource one: https://opensourceebike.github.io/

I think that's going to be a problem for many would-be users. Re-programming an off-the-shelf display is one thing, but building one from scratch is likely going to be a problem for many. That is, unless you are planning to sell ready-to-use display kits?

You are correct, this is not for would-be users. This is for me and advanced users alike, this is for my current ebikes (5 ebikes on my family) + for my 3 electric scooters and for my futures devices, who knows, maybe an electric motorcyle??

I am not planning to sell this devices. But I wish shops like Electrify Bike or Eco Cycles, that are selling TSDZ2 motor and displays with our OpenSource firmware, could decide to produce and sell this display to the users, and they could even customize the display enclosure with their brand. My wish is that production of this EBike technology could be dispersed and done more locally and not only on China or German - but keeping to be OpenSource, with shared OpenSource knowledge and not proprietary and very expensive like German Bosch for instance.

Anyway, the display is very easy in the DIY perspective and I am trying to improving the building and testing documentation as possible.

 

[Nrgbod]

I've just ordered/ received two sw102 displays for use on mine and my sons tsdz2 motors but now can't find the sw102-full-x.y.z.hex file to flash the bootloader for the first time to allow bluetooth firmware uploads. All the github links now give 404.
Can somebody please pm the file to me to allow me to get started so I can try the osf.

 

[anszom]

Here is sw102-full-0.19.11.hex that I've found on my disk. It's a bit old and so I don't think it would be compatible with the current motor firmware, you should update it to a recent release.

 

[Nrgbod]

Here is sw102-full-0.19.11.hex that I've found on my disk. It's a bit old and so I don't think it would be compatible with the current motor firmware, you should update it to a recent release.

Thanks anszom, that's the one I needed, just flashed it and the display is now running the simulator :thumb:
I'll give the Bluetooth updater a try tomorrow.

 

[Whitenoise]

Hi all - I would like to record available data from the TSDZ2 to a text or csv file for later analysis. How can I do that? I’ve searched a lot on the subject but the only viable option I see is using a Grin Cycle Analyst and Analogger, which is a pretty big $$ investment and mostly duplicates already available functionality.

Can I instead load one of the OSF to my motor and use an android phone or the Analogger (records to SD card)? It would be nice to have 5 or 10Hz data sampling rate, but I could live with less frequent.

My main goal is to develop some predictive tools to estimate remaining range in real time, as I have found even the expensive OEM bikes lacking in that regard (pedaling a fully loaded touring ebike after running out of battery is not fun ).


Sent from my iPhone using Tapatalk Pro

 

[casainho]

Hi all - I would like to record available data from the TSDZ2 to a text or csv file for later analysis. How can I do that? I’ve searched a lot on the subject but the only viable option I see is using a Grin Cycle Analyst and Analogger, which is a pretty big $$ investment and mostly duplicates already available functionality.

Can I instead load one of the OSF to my motor and use an android phone or the Analogger (records to SD card)? It would be nice to have 5 or 10Hz data sampling rate, but I could live with less frequent.

My main goal is to develop some predictive tools to estimate remaining range in real time, as I have found even the expensive OEM bikes lacking in that regard (pedaling a fully loaded touring ebike after running out of battery is not fun ).

Currently it is 20Hz data sampling rate from the motor controller to the display. Display then is sending the data to the mobile phone by Bluetooth. We have our OpenSource app for Android.

Would be great if someone could develop something about machine learning. We can also add an IMU / accelerometer to the display, and the NRF52 on the display support some basic machine learning.

See here how to build the display or technical details for developers: https://opensourceebike.github.io/

 

[Whitenoise]

Thanks casainho! Building display is a bit of a stretch for me but not if it’s the only option! I was hoping there is some OSF option with my existing VLCD5 display but purely for data logging. Using machine learning algorithm is definitely my plan since otherwise you need very detailed physics models and the data to put into it. But I was planning to use software similar to Matlab which works in desktop only. It would be interesting to use ML on the custom display!!


Sent from my iPhone using Tapatalk Pro

 

[ornias]

It is interesting to read this article on the Edrivenet.com site
Among other things it is written this:

"September 2021: Bafang has sued Tongsheng for patent infringement. Bafang has won this lawsuit, therefore the Tongsheng TSDZ2 officially can’t be sold on big platforms like Aliexpress and Amazon anymore. Because of this, it can be hard to find sellers of the Tongsheng TSDZ2 on these platforms. Sellers on Aliexpress and Amazon now don’t use the official name anymore to avoid getting their listings removed by the platforms."

https://edrivenet.com/tongsheng-tsdz2-vs-bafang-bbs02/

Very good writing in edrivenet.com about the differences between bafang and TSDZ2. Thank you for shearing.

edrivenet is made by people that don't even know the difference between peak and nominal power.
It's wise to not-put too much faith in them, when it comes to motors
To be frank, i've not been able to find any source for this, that was not derived from edrivenet.

I would be very interesting to find more, because all of bafangs motors at the time of TSDZ2 developement, where clones of already existing motors as well.

 

[ornias]

I was looking at the controlboard (v1) specifications and the 60v limit.
Assuming one would replace the 63v rated caps, is there any other reason why one couldn't use a 18s battery pack (about 75v peak voltage when charged)?

All other parts seem to support a max voltage of either 75v or 80v.

Assuming the same powerdraw (so lower amps), any idea how the coil would respond to that? More heat, less heat, more torque, less torque, more rpm, less rpm...etc...?

 

[LeftCoastNurd]

I was looking at the controlboard (v1) specifications and the 60v limit.
Assuming one would replace the 63v rated caps, is there any other reason why one couldn't use a 18s battery pack (about 75v peak voltage when charged)?

All other parts seem to support a max voltage of either 75v or 80v.

Assuming the same powerdraw (so lower amps), any idea how the coil would respond to that? More heat, less heat, more torque, less torque, more rpm, less rpm...etc...?

when dealing with an inductive circuit (eg, the motor coils), you really need some margins between peak rated voltage and operating voltage, something like 25% at a bare minimum

 

[Ianane]

G'day all, Thanks for your sharing over the years. I bought a 48v 500w TSDZ2 about 6 weeks ago, largely due to the strength of this community and the open-souce software developers whose work I'm looking forward to understanding one day, and was dissappointed when it didn't work. Made the 99.9 speed sensor gap error and fixed it but the best it would do is about a kilometer, and it was a great kilometer, and then a 'soft clunk' like something disengaged. If I powered the battery off, drained the circuit and the repowered up, I would get a whirring sound for a bit and then nothing. The first test ride I was on an errand and going down hill... all subsequent test runs have been uphill... I may be slow but I learn from experience
Anyway, I made videos of the issue, and my supplier, Cap Rouge, replaced the motor promptly, can't fault their service one iota and they left me with the original dud for spares or... fault finding and repairing.
I'm having no mechanical issues with the 2nd motor at all... though it doesn't significantly outperform a same specced front hub I have... that is a bit disappointing... not the least reason being that the TSDZ2, as an effort amplifyer, is saying nasty things about my effort.
So, the history out of the way, I can't find any referrence to whirring sounds from TSDZ2s here, or elsewhere. I suspect the torque controller is faulty but I'm guessing. Any suggestions greatfully received.

 

[-Pete-]

Hi, firstly thanks to all the developers and contributors to the TSDZ2 OSF, much appreciated! I've just fitted a 36V TSDZ2 with a VLCD5, and bought a Malectrics Arduino spot welder to make my own packs from new cells. Now I'm trying to consider ALL the possibilities, and I read that the lowest voltage is 24V but wondered if it's possible to go lower? 6S would be 19-24V and physically small, I can get build 5 mile packs for about £40 each using decent new cells and 6S balance chargers are cheap. Any thoughts?

 

[casainho]

Hi, firstly thanks to all the developers and contributors to the TSDZ2 OSF, much appreciated! I've just fitted a 36V TSDZ2 with a VLCD5, and bought a Malectrics Arduino spot welder to make my own packs from new cells. Now I'm trying to consider ALL the possibilities, and I read that the lowest voltage is 24V but wondered if it's possible to go lower? 6S would be 19-24V and physically small, I can get build 5 mile packs for about £40 each using decent new cells and 6S balance chargers are cheap. Any thoughts?

You need to test by yourself the lowest battery voltage possible on the motor controller.

You will also get lower cadence, is up to you to test and see if you like the results.

 

[Elinx]

...... wondered if it's possible to go lower? ........ Any thoughts?

You can consider to use a 1500W step up boost converter, that can give a relative high current.
In that case you can keep the output 36V or 48V, independent of the lower input Voltage.

 

[ornias]

I was looking at the controlboard (v1) specifications and the 60v limit.
Assuming one would replace the 63v rated caps, is there any other reason why one couldn't use a 18s battery pack (about 75v peak voltage when charged)?

All other parts seem to support a max voltage of either 75v or 80v.

Assuming the same powerdraw (so lower amps), any idea how the coil would respond to that? More heat, less heat, more torque, less torque, more rpm, less rpm...etc...?

when dealing with an inductive circuit (eg, the motor coils), you really need some margins between peak rated voltage and operating voltage, something like 25% at a bare minimum

"peak rated voltage"
I never know a lawyers label could define the laws of physics.

That was a joke ofcoarse, what I meant to say is that "rated" does just mean "Some idiot put a number on a label, which might or might not be done using a repeatable and/or acceptable international standard.

It's also not really an answer either, as I was refering to the possible (side) effects. Not "rules of thumbs". I think i'm smart enough to understand that something rated for 48v is not supposed(!) to run at anything much higher than that.

On the other hand the 36v coil also runs quite decent at 48v, so it's not support clear what the actual coil ratings actually are in this case, as we also dont get any certification reports. Besides the point that they are 250w nominal rated, using an older measuring method (which is also the reason they are actually closer to 500w nominal).


---

Anyway, on topic:
After some research I came to realise that the torquerating of the coil is fixed regardless of voltage and only scales with current. So upping the voltage doesn't magically create a higher torque rating.

So, the primary question I'm still figuring out:
Assuming the same power output (so lower current with higher voltage), how would a 16s, 17s or 18s battery affect heatproduction of the motor?
Lets say 75v 12/13A