TSDZ2 repair - need advice

Hi!
I manage to destroy my torque sensor, so I ordered a new one (had to wait for more than 10 weeks for the damn thing to arrive).
I have 2 questions:
- Static coil is spring loaded, consequently it is rubbing against the rotating coil (the force is not that big, but still...). Shall I put some grease between the coils, or leave it dry?
- I forgot where is the right place for the spring washer (I guess it should be at the retaining clip, but there is also possibility to put it between 2 bearings on the left side, HK2214 and 6902RS), so, better safe than sorry

Nick2 said:

Hi!
I manage to destroy my torque sensor, so I ordered a new one (had to wait for more than 10 weeks for the damn thing to arrive).
I have 2 questions:
- Static coil is spring loaded, consequently it is rubbing against the rotating coil (the force is not that big, but still...). Shall I put some grease between the coils, or leave it dry?
- I forgot where is the right place for the spring washer (I guess it should be at the retaining clip, but there is also possibility to put it between 2 bearings on the left side, HK2214 and 6902RS), so, better safe than sorry

Click to expand...

The coils rubbing is normal, essential even. I recently wondered about the grease too and after more research it seems the covering on the coils is supposed to be a low-friction material and grease is not required. I think the small amount of grease I did find in that area had leaked out of the HK2214 roller bearing since it doesn't have any seals. It had also got onto the two teflon discs which made them sticky and drag, but after cleaning it off they spin much more freely.

The spring washer goes on the spindle between the 6902RS bearing and the circlip. However while you have the opportunity it would be better to shim that space with flat washers instead. See this post and linked document.
https://endless-sphere.com/forums/viewtopic.php?f=28&t=79788&p=1512178&hilit=shim#p1512178

What happened with your torque sensor by the way? You may be able reduce the chances of it happening again.

Thanks!

Initially I had a pretty bad case of wobbliness, after less than 500 km's from new.
At the time, It already started to eat into the plastic cover.
I thought that the culprit was the sprag clutch, but it turned out it is not the case.
Anyway, i put 5 small washers between the big gear and the chain sprocket, as a temporary measure to increase the space a bit and stop rubbing against the plastic cover.
What i didn't realize is that the increased space allowed the big gear to slide invards on the sprag clutch outer surface and start putting pressure on the torque sensor coil, rubbing it against the casing and the static coil, until it broke.

I read some posts before about replacing the sprag clutch, stating that I need a hydraulic press to remove the big gear, so I didn't expect it can move on the outer race. I replaced the sprag clutch also, and it turned out that I could press the gear in place using the vise - quite easily.

I believe that the main issue is the change of design. Initially, Tong Sheng was using straight cut "big gears", and those were probably fine (in that "woblines" parameter), but at some point, somebody decided to use helical gears, to reduce the noise - and here lies the problem.

According to the available data, this motor can produce somwhere around 90 Nm of torque, so I will use this number for calculation. Radius of the "big gear" is about 58 mm. To produce 90 Nm of torque, motor must provide about 1550 N (it is like weight of 158 kg!). So far, so good.

Introducing helical gears, it also introduces axial force.
As far as I could see, the tooth angle is about 10 degrees (give or take), and that will produce about 17% of radial force (in our case, 263 N or about 27 kg weght). Based on the tooth orientaion, this force is oriented inwards at the bottom where small gear is. The whole axle/torque sensor assembly must endure the additional torque of 15 Nm in the axial direction. Looking from the axle perspective, this torque also changes the direction, as axle rotates. I believe it was not designed for that...

The more I hear about TSDZ2, the more I’m “impressed” by Tongsheng’s engineering (lack of).

Perhaps some retaining compound between the sprag clutch and big gear wouldn't be a bad idea. On mine it looks like there might be slight gap behind the spider where it could still potentially move. I currently have my motor apart and have been thinking how to improve things because I also had the chain ring wobble and rubbing over the pinion gear area.

It seems like essentially the helical gears want to screw together, pulling the big gear in towards the pinion gear. But not only did they change from straight-cut spur gears to helical, but they introduced more play at the same time. The second 6902RS bearing on the spindle was removed and the sprag clutch fit on the torque sensor was loosened. Previous a puller was required to remove the big gear and sprag clutch, now it slides off easily. Instead of adding a thrust bearing behind the torque sensor, we get two teflon washers.

The HK2214 roller bearings have a nice interference fit in the gear case but are sloppy on the torque sensor assembly. This may actually be necessary for the torque sensor to work properly. The 6902RS is the only bearing directly supporting the spindle. The inner race is a terribly loose fit and it's not great within the housing either. None of these bearings support axial load.

If you can remove the play you can go some way towards mitigating the wobble. You could use additional bearings, retaining compound and/or shims. I've been posting about this in the mega thread recently, starting from around here.
https://endless-sphere.com/forums/viewtopic.php?f=28&t=79788&p=1571403#p1571403

Simply replacing the bearings with name brand ones won't help (apart from longevity). The majority of play isn't within the bearings themselves. There is some play within the sprag clutch but nothing can be done about that.

In addition to the shim washer procedure I mentioned in my previous post, on the non-drive side you can add a second bearing as far out as possible. Then remove all play between the inner races and spindle and preferably between the outer races and housing too. Replacing the dust seal in the spider with a 6902 bearing may also help. I have posted about Enduro bearings which are for sealed for external use.

If your sprag clutch has any play on the torque sensor then you'll need to add shim washer there too, I will likely place them between the torque sensor and sprag clutch to give a bit more wobble room. Earlier motors actually had them there.

It's a lot of work, but at a minimum I'd do the shim washer procedure to the 6902 bearing and shim or glue it to the spindle with retaining compound.

Nick2 said:

I read some posts before about replacing the sprag clutch, stating that I need a hydraulic press to remove the big gear, so I didn't expect it can move on the outer race. I replaced the sprag clutch also, and it turned out that I could press the gear in place using the vise - quite easily.

Click to expand...

I couldn't get mine out using the vice, but I think I was trying to push it the wrong way. There's a lip on the inner side of the big gear opening that stops the sprag clutch sliding in too far. So I used a hydraulic press which did it easily. However I see there was retaining compound added to mine at the factory, did yours have anything on there? I'll definitely be adding some during reassembly now.

No, i did not find any traces of retaining compound. Also i got only one teflon disk originaly (replacement torque sensor came without it).
And one of 3 nipples on the coil was missing, so i used one from the old coil and glued it in place with epoxy.
All in all, pretty sloppy job.
Anyway, I bought 3 new bearings, of the highest quality that I could find, replaced the old one on the left side, adding another (there was just enough room for additional bearing and the rubber seal). On the right side I replaced the rubber seal with the third bearing.
So it is working again, who knows how long it will last this time.
I don't think I am going to fix when it fail next time - I will probably switch to Bafang or something else.

Nick2 said:

And one of 3 nipples on the coil was missing, so i used one from the old coil and glued it in place with epoxy.
All in all, pretty sloppy job.

Click to expand...

Was the missing nipple the one next to the wire like in this photo? It's a newer version and they have intentionally removed it, although my late model motor still has the 3 nipple version. Since it's the one closest to the pinion gear, I can only theorize that it must be something related to the rubbing problem. Perhaps they removed it to relieve the pressure from the big gear being pulled in there. So maybe we should actually be removing it if we have helical gearing?

Yes, it is the same like yours.
If they changed the design, at least they should provide some assembly instructions.
Besides, I beleve that those nipples are there to take the pressure from the cable entry point.
Anyway, it is too late now, it is glued by epoxy glue and I am not going to open it again.

Nick2 said:

Yes, it is the same like yours.

Click to expand...

Did the replacement come with three springs? I suspect this is all a known problem, removing the nipple is hardly likely to be a cost-cutting exercise.

Replacement come without springs, I had to reuse my old ones.

This seller only supplies two springs with the newer version, but it would be good to get confirmation from someone who had it from the factory.
https://www.aliexpress.com/i/4000680824444.html

Not coming with the teflon discs is a annoying, I have been told they are not available. One of mine is a bit weird and other has been scratched up by the case.

I wonder if not having that nipple and spring causes any change in operation of the torque sensor in certain situations such as when at a standstill. Potentially if the discs don't make proper contact then only cadence can be sensed until the motor starts to spin. We may see people having trouble getting moving when using the torque-only mode of the open source firmware. It would be interesting to see things working with the cover off to gauge just how much pull in there is on the big gear.

I must admit that I don't really understand this design. As far as I can see, coils are paralel to each other, and that basically makes a transformer. Since the winding is in the direction of the rotation, it doesn't matter if the axle is rotating or not - it should work just the same. But the whole point of the transformer is to avoid the direct contact between coils. If they don't mind having coils rubbing against each other, they could just put 2 sliding contacts, it would be cheaper and it will also save a lot of space.

famichiki said:

The HK2214 roller bearings have a nice interference fit in the gear case but are sloppy on the torque sensor assembly. This may actually be necessary for the torque sensor to work properly. The 6902RS is the only bearing directly supporting the spindle. The inner race is a terribly loose fit and it's not great within the housing either. None of these bearings support axial load.
ting from around here.
https://endless-sphere.com/forums/viewtopic.php?f=28&t=79788&p=1571403#p1571403

Click to expand...

Dang been trying to get it off thinking it was gone.. does feel a little rough though. But i cant get it past the torque sensor..?

charakaSamawry said:

Dang been trying to get it off thinking it was gone.. does feel a little rough though. But i cant get it past the torque sensor..?

Click to expand...

I replied to your other post with some instructions over here...
https://endless-sphere.com/forums/viewtopic.php?f=28&t=79788&p=1589910#p1589910

famichiki said:

charakaSamawry said:

Dang been trying to get it off thinking it was gone.. does feel a little rough though. But i cant get it past the torque sensor..?

Click to expand...

I replied to your other post with some instructions over here...
https://endless-sphere.com/forums/viewtopic.php?f=28&t=79788&p=1589910#p1589910

Click to expand...

Thank you for the detailed response!