el_walto wrote:I'm interested in the Torque Sensors. Would be nice if i could build pedal and go ebikes for my parents. Not sure if there is going to be anything decent available from ebikes.ca torque sensor wise to go with the CA.
For sure, we have a small number THUN torque sensors on hand (like maybe 6 or 7 left for beta testers) and will be ordering them in some quantity very soon so that we can offer a CA V3+THUN bundle, with pre-made wiring harnesses to simplify the assembly. The input settings for the torque and PAS sensor in the CA itself are pretty flexible, so it's not tied to any specific device and should accomodate anything on the market that has a linear voltage/torque relationship. You indicate the scaling factor in Nm/V (the thun is 100, but we set this to 200 so that it doubles the torque to account for the single sided torque sensing nature of the device):
- TrqScale.jpg (9.96 KiB) Viewed 1759 times
Plus you can zero the torque offset, much like the zero amps routing for the current offset:
- Trq Offset.jpg (10.48 KiB) Viewed 1759 times
For the pedal cadence sensing, you set the number of pulses per pedal rotation:
- PAS Poles.jpg (9.73 KiB) Viewed 1759 times
And if the cadence sensing has a quadrature encoder, then there is a polarity selection to indicate which is direction is forwards and which is reverse pedalling:
- Dir Polarity.jpg (9.78 KiB) Viewed 1759 times
All of my testing so far has been done with the THUN X-Cell sensor, for the most part it is pretty good but there are a few slight shortcomings that came up:
1)
Magnetic Field Sensitivity: My demo bench was welded up from steel frame bike parts and a lot of the tools in my garage are slightly magnetized, and as a result the test jig has some residual magnetism too. It turns out that the sensing technology is extremely sensitive to external fields, so on my demo setup even with no torque on the bottom bracket, I would see a variation in about +- 10Nm in the signal output just by changing the spindle angle. When I touched my slightly magnetized wrench to the THUN spindle to remove the crank say, then the signal would jump off the scale. On an alloy frame bike this is all of minimal consequence, but I can envision situations with steel frames that have had magnets stuck on them in the past where a degaussing of the bottom bracket area might be necessary.
2)
Offset hysterisis, The specified accuracy error of the THUN sensor is 2%, which looks really good at first. However, that is not the linearity error but 2% of full scale. Since it reads up to +- 200Nm, then the output can be off by +- 4Nm, which is a lot if you are only doing light pedalling. If you are averaging like 15-20 Nm which is pretty typical for casual effort, then the accuracy error can reach 20-30%. You can see this first hand by measuring the voltage from the THUN sensor with no torque, then standing and hoping on the pedals, and then looking at the zero torque reading again. If it was 2.49V the first time, it might read 2.52V or so afterwords. Spin the pedals 180 and hop on them again, and the zero torque reading might be 2.47V. That means that for very accurate reading of the human power, it will be important to periodically check and re-zero the thun's torque offset. For just PAS mode it's no issue, but for someone using it as a training aid it's not quite as suited as I had hoped.
3)
Spindle length options Thun offers 120, 127, and 136mm spindles which is fine for the large majority of bikes. We thought it would be fun to install one on the Dogati bike after the Taipei show and have maybe the worlds first multi kW pedalec! But the Dogati needs a >150mm spindle for the cranks to clear the motor drive assembly, and lots of other mid-drive ebikes (like the Cycleone-USA stuff) require speciality long spindles too. I'll ask Thun what the minimum numbers would be to have a longer spindle option tooled up, because if we can have it work with the majority of mid-drives as well then that would be sweet.
Here is what the sensor looks like, btw:
- THUn BB.jpg (11.69 KiB) Viewed 1759 times
The ST-01 torque sensor being shown by GreenTrans looks like it has a few benefits, namely that it senses both left and right crank torques and also has a wire exit on the side of the bottom bracket rather than through the center, so you don't need to drill a hole through the frame to feed the wire out:
http://www.greentrans.com.tw/eng/downlo ... alogue.pdfHowever, the dual sided sensing means that a special adapter needs to be fit to link the chainring to a splined sleeve on the spindle. You can't just use an off-the-shelf crank and chainring on the right side if you want to "see" the right side pedal torque. And when I talked to them, they were really only interested in providing a whole package solution and not so much in supplying just the torque sensor. I'll try to follow-up but don't hold too much promise.
The other supplier is Schaeffler FAG, who do the torque sensing BB's for the Bosch mid drive module. In the press release they also list it as dual left/right pedal torque sensing:
http://www.schaeffler-group.com/content ... id=3447045But I heard indirectly that this is only available to Bosch, and the one that they showed me at the show was single side sensing like the Thun, but instead of with a -200 to +200 NM range, it was a 0-100 Nm range. So it would probably be more accurate at measuring human watts at lower power levels, but would hit saturation when someone is really pushing on the cranks.
Definitely Thun as a company has been most responsive and the easiest to work with so far, so we'll carry their sensor to start, but it's good to see that there could be a few other options for people in this space. The last torque sensor we looked at was YST:
http://www.yst-corp.com.tw/products_vie ... 1104150001Theirs has a non-linear varying resistance to torque relationship, which isn't very practical for device interfacing.
-Justin
