New "TSDZ2 Torque Sensor Central Motor"

[arka]

Casainho - So finally you showed us the reason why you are doing such cool custom firmware - it is your son with whom you have great fun on bikes
Unfortunately, it is not possible to mount the front derailleur to this frame, because there is no way to screw it. To install the TSDZ2 I will also have to cut the frame holders on the chain tensioner and think about how to mount another front tensioner ...


View attachment 3

View attachment 1

I ride in the rain like that and try to make everything water resistant. At the moment my wife's Giant is the same like You 40T-11T cassete cheap from Ebay Sunrace
I thought about making a battery permanently in the triangle of the frame + optional to add a water bottle for long routes.
As for overheating, it would be great if the field with the current temperature value appeared on the display, because the graph does not show precisely what the temperature is :-(
In 3 months summer begins for me I can't wait to start traveling with my son on trips

 

[casainho]

Oh, my bicycle frame is totally different from yours, yours seems much more robust maybe for enduro. My frame came originally with switch for the 2 front gears.

As for overheating, it would be great if the field with the current temperature value appeared on the display, because the graph does not show precisely what the temperature is :-(

You need to read more carefully the wiki documentation about configuration - motor temperature variable is available to be selected on the graph as also on any of the 4 numeric fields.

 

[arka]

Oh, my bicycle frame is totally different from yours, yours seems much more robust maybe for enduro. My frame came originally with switch for the 2 front gears.

As for overheating, it would be great if the field with the current temperature value appeared on the display, because the graph does not show precisely what the temperature is :-(

You need to read more carefully the wiki documentation about configuration - motor temperature variable is available to be selected on the graph as also on any of the 4 numeric fields.

I have an 850C display, but I want to look on the wiki, thank you.

 

[beardydave]

Hello!

Posted elsewhere, it was suggested I put here...

I'm having problems snapping BB spindles at the groove where the circlips locate. I am a heavy and strong cyclist, I've destroyed plenty of BB and pedal spindles and cranks in my time, but I've done three of these in the last year (approx one every 1,000 miles) and I'm a bit fed up of it. So I've asked around and found a couple of companies that are willing to look at producing a stronger version.

I was planning on getting it made identical to the existing one but in hella strong steel and without the circlip grooves, which clearly act as a stress riser exactly where you least want one. I was planning to instead retain the bearings (or locate the spindle, depending how you look at it) using thin sleeves with wavy washers locating against the inside of the crank arms.

Has anyone tried this before?
Any other solutions to this problem?
Any comments / advice on this proposal?
And finally would anyone be interested in buying one of these if I do get them made? 10 of them would be cheaper per unit than 2! No idea on actual cost yet.

Cheers

 

[skestans]

I'm having issue converting the TSDZ2 stock 48V to 52V using the Linux `stm8flash` tool:

 $ sudo ./stm8flash -c stlinkv2 -p stm8s105?6 -w ~/Downloads/52vdatamemory.s19
Determine FLASH area
Due to its file extension (or lack thereof), "/home/user/Downloads/52vdatamemory.s19" is considered as MOTOROLA S-RECORD format!
Address 00004000 is out of range at line 1

Anyone knows how to write to this address?

 

[ezrider1199]

Does anyone have experience with gearing? I currently run a 48T chainring and a 12-32 7-speed cassette. Im thinking dropping to a 40T and a 11-28 cassette. I believe going smaller on the front chainring will transfer more of the torque to the wheel. Maybe 15% reduction in chainring teeth will be a 15% increase in torque? Anyways the number seems significant enough to try it. I will be losing ~1.25 MPH off the top speed which is fine since you gain efficiency. Also better chain line with a smaller chainring

 

[Lubin]

Regarding Torque sensor calibration, I fixed my issue which was a misplaced point within the 850C interface.

I am now running with the 0.6.4 firmware which works great, more sensitive; well done! I modified my calibration curve to ensure a minimum slope to limit the sensitivity of the inverse function (I set one last curve segment with a high slope protecting the inverse function from a possible sensor raw value shift, possibly due to temperature). I am not noticing anymore the superman behavior where human power suddenly get high. I suspect the root cause was sensor noise + high sensitivity of the inverse function.

I am sometime seeing one issue at initialization which make the human power measurement far too high (raw torque sensor value is ok). In such case, rebooting the system fix it. Once it works, it's all ok (never had to reboot during a trip).

One suggestion: my torque sensor value is 185 at rest while default value is like 280. It takes a while to set the 18 point. Maybe it is possible to ensure that all points entered are monotonically increasing. The benefit is twofold:

• ensure no misplaced points

• provide a quicker way to enters points, particularly when far from the stock setting

Lubin

 

[qwerkus]

After exhaustively searching the site, I finally found the product page for the egg shaped steel wire rope grip. But you were saying you think something like the zinc plated duplex steel wire rope grips clamps would be better? I don't think that would work for my needs . I'm struggling to find a suitable torque collar for my TSDZ2, I have a square 40mm x 40mm (aprox 1.5" x 1.5") boom pole that I will be mounting the motor and torque collar to as I'm doing the conversion on a recumbent. Would really appreciate some suggestions.

Yeah cables are the more flexible solution to hold the motor unit in place. I found that clamps with only one screw are not strong enough, but maybe your experience will be different. Definitely go with stainless steel: the area around the BB is usually the bikes closest place to the ground and gets hit by all kinds of dirt and debris. Overall I think the bafang solution where the motor would just push against the downtube is the better one; unfortunately it makes fo a weird look with the tdsz2, with everyone wondering if the motor is broken...

 

[ancan]

Hi,
I just changed from the LCD3 to the 850C this weekend, and must say it's a much better feeling from the motor, and it also feels a lot stronger if that's posssible. Great job!

I'm however struggling with some calibration issues. First ride to work yesterday, I completely zipped away on assist level 1, cutting five minutes of a 40 minute ride.
Before the trip back home I adjusted the lowest assist levels to that 1=0.01 and 2=0.02, and while assist level 1 is more reasonable now, the step to 2 is just too big and the higher assist levels feels like you're on a moped.

I calibrated the torque sensor last evening, getting resting values at around 140, and maxing out on 358/393 for my 85kg. This did no big difference to the assist levels.
One thing I've noticed is that my "human power" seems to be around 300w in my comfort zone, and I wonder if that's a bit high. Agreed, I've yet to learn how to ride a bike slowly and tend to push, but I'm not that fit nowdays.

Have I missed something in the setup? Is there a way to tune-down the power-level overall, so you can get better granularity in the assist level settings?

(And I agree on Lubins comment on torque adjustment, which is a bit tedious if your sensor reads so much lower than the defaults. Perhaps it could do 10-steps at a time if you hold down the button?)

 

[casainho]

Hi,
I just changed from the LCD3 to the 850C this weekend, and must say it's a much better feeling from the motor, and it also feels a lot stronger if that's posssible. Great job!

I'm however struggling with some calibration issues. First ride to work yesterday, I completely zipped away on assist level 1, cutting five minutes of a 40 minute ride.
Before the trip back home I adjusted the lowest assist levels to that 1=0.01 and 2=0.02, and while assist level 1 is more reasonable now, the step to 2 is just too big and the higher assist levels feels like you're on a moped.

I calibrated the torque sensor last evening, getting resting values at around 140, and maxing out on 358/393 for my 85kg. This did no big difference to the assist levels.
One thing I've noticed is that my "human power" seems to be around 300w in my comfort zone, and I wonder if that's a bit high. Agreed, I've yet to learn how to ride a bike slowly and tend to push, but I'm not that fit nowdays.

Have I missed something in the setup? Is there a way to tune-down the power-level overall, so you can get better granularity in the assist level settings?

(And I agree on Lubins comment on torque adjustment, which is a bit tedious if your sensor reads so much lower than the defaults. Perhaps it could do 10-steps at a time if you hold down the button?)

With the feedback already from some users, I am thinking to increase assist level to resolution of 0.001, this means that you can have the 0.01 (3 motor amps at your 300W pedal human power) and then increase on +10% step, to 0.011, so, 3.3 motor amps for your 300W pedal human power. But I think steps increase of 50% or not lower than 33% is the min the user will want.

About the torque sensor table, yes, I think it is a good idea to increase in steps of 10 units at long press.

This are things I want to do in short time.

About the human power value, I think it should be now the same across all users that calibrate the sensor. Still, let's see if someone can do a comparison with a calibrated measuring system, like Azur user that was thinking to buy a pedal power meter that is included on the pedals - I expect only a factor of maybe 0.8 or 0.6...

 

[ezrider1199]

Would it be feasible to have a step variable at the top of the assist level config? just an open question

But I think steps increase of 50% or not lower than 33% is the min the user will want.

 

[casainho]

Would it be feasible to have a step variable at the top of the assist level config? just an open question

But I think steps increase of 50% or not lower than 33% is the min the user will want.

Please clarify. What is you need? And why?

 

[ezrider1199]

It's ok i was thinking it would help with setting the levels but its sort of adding a little complexity. Forget it

Would it be feasible to have a step variable at the top of the assist level config? just an open question

But I think steps increase of 50% or not lower than 33% is the min the user will want.

Please clarify. What is you need? And why?

 

[Retrorockit]

I'm starting to adapt the third bike for TSDZ2, of course under Custom firmware. It will be the 2015 Mondraker Summum on 27.5 inch wheels, a downhill bike:
Zrzut ekranu 2020-01-23 o 20.05.24.jpeg

So far I've made two such constructions for wife and kid:
Zrzut ekranu 2020-01-24 o 11.58.22.jpeg
Zrzut ekranu 2020-01-24 o 12.01.03.jpeg

My driving style is very aggressive, at full speed in the mountains and in the woods. Can you advise me or anything else I can predict. assumptions:

1. Shark 52V battery suspended under the frame as low as possible so that the wheel does not rub. Protected with aluminum sheet as in the picture:
Zrzut ekranu 2020-01-23 o 12.26.13.jpg

2. To increase speed on flat ground, I will add 52T forward which I will manually change if necessary. I will add them like a user from this forum https://tiny.pl/tsrvj
Zrzut ekranu 2020-01-23 o 20.24.50.jpeg

3. I wonder whether to change the cassette on the back to 11-40T or whether it could not be better to use Nuvinci N380 / in which the gears can be changed under load or when stationary. According to the manufacturer's specifications, the hub is adapted to work under 75Nm load and max peak 90 Nm. I have such a hub https://tiny.pl/ts9tx or https://tiny.pl/ts9gb
Zrzut ekranu 2020-01-24 o 11.48.38.jpeg

4. I have a temperature sensor installed, but with my aggressive driving style, it overheats very quickly. From this topic I see that it is best to dissipate heat to the housing through steel sheets https://tiny.pl/tsrbh
Zrzut ekranu 2020-01-24 o 11.29.09.jpeg

Can I ask you for suggestions what to add, what to change?

A good bike for that purpose.
As far as shifting goes with an 8-9 speed casette you have the option of normal XTR "high normal" derailers that can upshift under load for climbing, or running a "low normal" Rapid Rise derailler and a gripshift that can dump gears sitting still. The shift will be preselected and will occur when the bike moves again. But not both. I run the Rapid Rise which does upshift under load OK but is spring loaded that way and not a solid pull on the cable. IDK if it's possible to run the old Rapid Rise derailer with newer 10 speed shifters or not. Might need thinner idlers?
As far as gearing goes I don't think a 52T chainring will do much for you on a 27.5 bike.
I'm basing this on my own 1500W BBSHD with a 50x11t top gear on a 26" bike. This is good for 36mph @100 rpm cadence. 2 things the TSDZ2 can't do. I don't think the TSDZ2 has the power to use that kind of gearing with that wheel size. If you lug the motor the Amps will go up and overheating will be worse. Power squares with increased speed. IMO 25-28mph is about what a TSDZ2 is capable of. You might get more on a skinny tire aero race bike, but you won't have much control doing it.
Here is a gear calculator. You can solve for speed @ 90 rpm and see what gear will work for you.
https://sheldonbrown.com/gear-calc.html
As far as cooling goes thermal pads/paste are good for eliminating air pockets to help conduct heat "between" parts. But metal has several times higher thermal conductivity. The best solution will use both methods.

 

[michih.]

Hello!

Posted elsewhere, it was suggested I put here...

I'm having problems snapping BB spindles at the groove where the circlips locate. I am a heavy and strong cyclist, I've destroyed plenty of BB and pedal spindles and cranks in my time, but I've done three of these in the last year (approx one every 1,000 miles) and I'm a bit fed up of it. So I've asked around and found a couple of companies that are willing to look at producing a stronger version.

I was planning on getting it made identical to the existing one but in hella strong steel and without the circlip grooves, which clearly act as a stress riser exactly where you least want one. I was planning to instead retain the bearings (or locate the spindle, depending how you look at it) using thin sleeves with wavy washers locating against the inside of the crank arms.

Has anyone tried this before?
Any other solutions to this problem?
Any comments / advice on this proposal?
And finally would anyone be interested in buying one of these if I do get them made? 10 of them would be cheaper per unit than 2! No idea on actual cost yet.

Cheers

hello,
me and a few others would be interested.
we have already destroyed some.
at the moment i have replaced the seal with ball bearings.
this is a support in front of the groove! because the waves break on the left.
but I would prefer a stronger wave.

mfg michael

 

[christiank]

hello,
me and a few others would be interested.
we have already destroyed some.
at the moment i have replaced the seal with ball bearings.
this is a support in front of the groove! because the waves break on the left.
but I would prefer a stronger wave.

mfg michael

I would be interested in 4 Axles.
IMHO the groove for the seeger-ring is not needed.
If 2 bearings are added to the left the spacing/movement of the axle can be adjusted by spacers between the last bearing and the crank. jm2c

To introduce myself, here are my bikes:
Liteville301MK4 + TDSZ2 0.18.2
https://www.pedelecforum.de/forum/index.php?threads/neues-projekt-liteville-301-mk4-tsdz2.62613/post-1292713
Custom Gravel/Treckingbike + TDSZ2 0.18.2
https://www.pedelecforum.de/forum/index.php?threads/bilder-von-ausfahrten-im-fr%C3%BChling-sommer-herbst-und-winter.31434/post-1360484

 

[sysrq]

Does anyone have experience with gearing? I currently run a 48T chainring and a 12-32 7-speed cassette. Im thinking dropping to a 40T and a 11-28 cassette. I believe going smaller on the front chainring will transfer more of the torque to the wheel. Maybe 15% reduction in chainring teeth will be a 15% increase in torque? Anyways the number seems significant enough to try it. I will be losing ~1.25 MPH off the top speed which is fine since you gain efficiency. Also better chain line with a smaller chainring

I'm using half-step gearing (typically more suitable for touring) with 42t/45t chairing and 8 speed Cs-hg31 11-32t cassette with 11t cog removed and 34t or 36t added. That's seems like an only way to get acceptable 130mm OLD chainline while retaining durability and cadence fine tuning for better usability with the motor off or on. Most steps between 16 destinctive gears are 7%, 8% and one is 10% or 12% if I remember correctly. Wider steps more suitable for MTB are still available at the back, while closer steps are better suitable for cruising on false flat roads.
As far as I know original 42t dished chairing is the most optimal for best chainline while not being too big. Closer steps between low gears should improve the motor assistance on hills since if the gear is too low and cadence too high then there will be not enough pressure on the pedals for torque sensor to detect and when slightly too high then the motor will bog down and overheat.

 

[jpm600]

Does anyone have experience with gearing? I currently run a 48T chainring and a 12-32 7-speed cassette. Im thinking dropping to a 40T and a 11-28 cassette. I believe going smaller on the front chainring will transfer more of the torque to the wheel. Maybe 15% reduction in chainring teeth will be a 15% increase in torque? Anyways the number seems significant enough to try it. I will be losing ~1.25 MPH off the top speed which is fine since you gain efficiency. Also better chain line with a smaller chainring

I'm using half-step gearing (typically more suitable for touring) with 42t/45t chairing and 8 speed Cs-hg31 11-32t cassette with 11t cog removed and 34t or 36t added. That's seems like an only way to get acceptable 130mm OLD chainline while retaining durability and cadence fine tuning for better usability with the motor off or on. Most steps between 16 destinctive gears are 7%, 8% and one is 10% or 12% if I remember correctly. Wider steps more suitable for MTB are still available at the back, while closer steps are better suitable for cruising on false flat roads.
As far as I know original 42t dished chairing is the most optimal for best chainline while not being too big. Closer steps between low gears should improve the motor assistance on hills since if the gear is too low and cadence too high then there will be not enough pressure on the pedals for torque sensor to detect and when slightly too high then the motor will bog down and overheat.

I would say that changing from 48 -32 lowest gear to 40-28 lowest gear will actually only gain you about 4% increase in torque in that easiest gear. at top speed in 40-11 you will be about 9% slower and 9% more torque- compared to 48-12. I would think that more relevant is making sure that speed at ~80cadence matches what you would actually be riding at.

Ie. 48-32 at 80 cadence equals about 14.9km/h
40-28 at 80 cadence equals about 14.2km/h
ie not much difference.

if you had 40-32 at 80 cadence equals at 12.4km/h

if you regularly climb steep hills at ~12km/h then that will be more torque for human power and motor power and as speed increases you change to a harder gear.

If you usually ride up steep hills at 15km/h then you wont need that lowest gear and you might prefer the closer ratios of the 11-28 cassette.

When I get my motor installed on my cargo bike it will be 42t with 11-32 cassette. which is a speed range at 80 cadence of 38km/h to 13km/h and about 15% jump between most gear shifts which i am fairly comfortable with on my 1x10 mountain bike.

check out sheldon brown's gear calculator to play with numbers for yourself.
https://www.sheldonbrown.com/gear-calc.html

 

[casainho]

SW102 firmware working

Finally I got the SW102 firmware working for the most recent version of motor controller, the SW102 and 850C uses the same firmware base (thanks to Geeksville developer!!) and so the same motor controller firmware. I hope today to make a new firmware release for both displays, where the assist levels factor can be configured starting from 0.001 and in 0.001 steps.

This means that for instance if I do 200W of pedal human power in average, I will get assistance of motor current of 2 amps if assist level is 0.01 (0.01 * 200 = 2) and I can increment in steps of 0.001, meaning next step can be 0.011, or +10% of previous step - this should give enough granularity for users to be able to adjust for their preferred assistance level factors.

 

[Waynemarlow]

Caisanho, looking good and this will give a nice number of options re the displays.

Before I go opening my SW102 to program it for the first time, has there been any further development to not need to open the display to program it ?

 

[ImpulsePurchase]

Judging by the comments here:
https://github.com/OpenSource-EBike-firmware/SW102_LCD_Bluetooth/issues/35
It looks like a couple of people are having a look into it.

I'm not sure if I understood the limitations, but I thought that it was possible to load the current firmware via a bootloader but you lost the bluetooth initialisation built into the original module. I got the impression (possibly wrongly) that you could continue to upgrade via bootloader as long as you didn't want to use bluetooth in the future?

My soldering skills are pretty poor and successful flashing whilst holding pin contacts to the terminals is rare (or non existent) so I'm currently considering this to be beyond my skills

 

[Nfer]

My soldering skills are pretty poor and successful flashing whilst holding pin contacts to the terminals is rare (or non existent) so I'm currently considering this to be beyond my skills

I have flashed holding the pins two displays. It took 30 minutes or so. It takes some time to find the right spot of contact but everytime I got a good contact my laptop made a sound. It seems it detected something in the USB port.

 

[ImpulsePurchase]

Thanks. Good to know. It looked possible but I hadn't heard many people saying it was a successful method.

I broke my second LCD3 screen a week or so ago (turned my bike upside to fix a puncture mid journey in the rain and the screen cracked) so I picked up an 850C as it was the same price as a replacement LCD3 (must have been discounted local stock) so I have just finished loading that with the latest firmware. I'll look at working out how to flash my SW102 at some point as this is my preferred form factor but I have hopefully more time to plan it out now to make it as neat as possible (or waterproof in my case as the LCD3 had a problem with water ingress in wet weather).

 

[casainho]

850C and SW102 displays new firmware

850C_SW102_v0.6.5

This version was developed for the TSDZ2 motor controller firmware v0.54.0 (TSDZ2-v0.54.0.hex).

Changelog:
- updated SW102 firmware, which have the same firmware base as 850C
- assist levels now with increasing steps of 0.001
- corrected Startup BOOST level

Download here: https://github.com/OpenSource-EBike-firmware/Color_LCD/releases/tag/850C_SW102_v0.6.5

Follow the wiki How to install the Flexible OpenSource firmware: https://github.com/OpenSource-EBike-firmware/TSDZ2_wiki/wiki

 

[ezrider1199]

thanks casainho! i just flashed it and will take it for a spin later

edit:

just came back from a ride and everything is good. more granular assist is definitely a handy option for me (0.025 feels really good but will also try going lower). one small thing i noticed the trip odometer has red font which is hard to read on the black background