Torque sensor question

[tbar23]

I had an interesting issue this morning. It has been ~6 weeks since my last ST2S commute (I don't like riding when snow banks narrow the roads too much). This weekend I took the opportunity to install a longer stem with a -15* angle to get a more aggressive / aero position on the bike. Following the cockpit modifications I did an ~21 mile ride in Mode 2 without any problems.
When I got home, I gently rinsed the bike being careful to stay away from the battery compartment and trying to stay away from where the hub motor plugs in. Given the low temps in my garage, I probably should have spent some more time drying the bike, but I did give it a quick wipe down. This was all on Sat or Sun.
Okay, this morning I took the bike out for my commute. As soon as I pulled out of my driveway, I felt that something was off. I wasn't getting the same level of assist that I was used to, but after cycling through the modes a few times, it seemed to be "ok".
However, about 8 miles later, I lost all assist. I stopped and removed the rear wheel, disconnected wiring harness, reconnected and put everything back together.
The issue seemed "better", but not perfect. The more I rode, the more it seemed like something was off with the torque sensor. If I worked hard (lots of torque), it seemed like I would get assist. Otherwise, not so much. This was primarily in Mode 3 and a little Mode 2.
Finally, I pulled over with a plan to increase torque sensitivity.
I had totally forgotten about the torque calibration.
I performed a cal and cranked up the sensitivity.
That appeared to solve everything.

So - my question for the group: has anyone else had torque sensor issues? Is the sensor particularly susceptible to water? I need to find the torque sensor on the bike so I can attempt to troubleshoot, in the future, as necessary.

 

[Johnny]

I had an interesting issue this morning. It has been ~6 weeks since my last ST2S commute (I don't like riding when snow banks narrow the roads too much). This weekend I took the opportunity to install a longer stem with a -15* angle to get a more aggressive / aero position on the bike. Following the cockpit modifications I did an ~21 mile ride in Mode 2 without any problems.
When I got home, I gently rinsed the bike being careful to stay away from the battery compartment and trying to stay away from where the hub motor plugs in. Given the low temps in my garage, I probably should have spent some more time drying the bike, but I did give it a quick wipe down. This was all on Sat or Sun.
Okay, this morning I took the bike out for my commute. As soon as I pulled out of my driveway, I felt that something was off. I wasn't getting the same level of assist that I was used to, but after cycling through the modes a few times, it seemed to be "ok".
However, about 8 miles later, I lost all assist. I stopped and removed the rear wheel, disconnected wiring harness, reconnected and put everything back together.
The issue seemed "better", but not perfect. The more I rode, the more it seemed like something was off with the torque sensor. If I worked hard (lots of torque), it seemed like I would get assist. Otherwise, not so much. This was primarily in Mode 3 and a little Mode 2.
Finally, I pulled over with a plan to increase torque sensitivity.
I had totally forgotten about the torque calibration.
I performed a cal and cranked up the sensitivity.
That appeared to solve everything.

So - my question for the group: has anyone else had torque sensor issues? Is the sensor particularly susceptible to water? I need to find the torque sensor on the bike so I can attempt to troubleshoot, in the future, as necessary.

What you said happened to me too and calibration solved it. Seems rarely calibration can go off and that is why there is the calibration button on the menu. I also don't rest my feet on the pedals when I turn on the bike.


You can see the specs of the TMM sensor on this page https://www.idbike.com/sensors
It is actually rated ip66 hence should be quite safe against water(however if water manages to freeze around the sensor, I guess it may throw the calibration off since it is a pressure plate). Its operational temps are -10 to 60C, storage temps are -40c to 60C.
It is actually more sensitive compared to the sensors in mid drives and it can detect a much wider range (on mid sensitivity it can detect up to 300nm).

That being said, it is located on the frame, close to the rear axle hence it has to be torqued to spec(I assume that you have a torque wrench) when taking the wheel out and putting it back on. (One of the headaches of this bike).

I don't know if yours is the same but on mine the speed sensor is incredibly sensitive. Detects even the slightest push. Coupled with the torque sensor it could have made a great power meter, I don't know why Stromer has not implemented that functionality yet.

 

[bluecat]

I don't know if yours is the same but on mine the speed sensor is incredibly sensitive. Detects even the slightest push. Coupled with the torque sensor it could have made a great power meter, I don't know why Stromer has not implemented that functionality yet.

This would create a cyclic loop; speed is a result of the TMM4 measuring.

The "hidden secret" of the Stromer riding feeling is the perfect integration of the sensitive TMM 4 signal into the motor power curve.

 

[tbar23]

Thanks @Johnny. It was about -3 or -4C when I left this morning, so still in range. I’ll double-check torque on the axle - thanks for the reminder.
Speaking of speed sensors - I have noticed that mine differs from my GPS reading by at least 1 mph (1.6 kph) - Stromer reads high.
@bluecat - I don’t think it is necessarily a cyclic loop. I’m pretty sure the torque sensors typically detect rpm based on the natural power stroke of a rider. So they are really outputting two independent variables - torque and rpm, which allows the system to calculate power.
I’m with @Johnny - bummed that “Stromer has not implemented that functionality yet”!

 

[Johnny]

Speaking of speed sensors - I have noticed that mine differs from my GPS reading by at least 1 mph (1.6 kph) - Stromer reads high.

I think this is by design(even cars are designed to overestimate the speed, I think it is a legal issue, vehicles can not under read so companies leave some buffer by overestimating). Almost all mid drives read 1.5-2mph higher so I wouldn't be surprised by stromer's overestimate. There is also the tire size inaccuracy.

But in terms of how sensitive the the speedometer is, you may try moving the bike ever so slightly when you are off the bike and see if a speed is registered. In my case even the slightest movement registers speed.

@bluecat , as @tbar23 also stated we have the user generated torque measured at the wheels and wheel speed hence the two variables to compute rider's power input (after drivetrain losses). This makes a good hub based power meter and hub based power meters are pretty accurate. Maybe we should petition Stromer about this.

 

[bluecat]

I’m pretty sure the torque sensors typically detect rpm based on the natural power stroke of a rider.

You're thinking about a frequency analysis which detects the two dead centers? AFAIK, the software does the opposite: Smoothing this line to avoid a pulsing motor.

 

[tbar23]

You're thinking about a frequency analysis which detects the two dead centers? AFAIK, the software does the opposite: Smoothing this line to avoid a pulsing motor.

I suspect you are right, @bluecat - and as @Johnny mentioned, the speed is measured from another sensor.

 

[bluecat]

the speed is measured from another sensor.

The motor of a Stromer does not contain any electronic - except the Hall sensors. I assume, there are three of them in a 120º angle. The "speed" is something like a side product of these sensors.

 

[Johnny]

I assume, there are three of them in a 120º angle. The "speed" is something like a side product of these sensors.

I can't follow this statement.
In these motors there are 3 hall sensors, each one is connected to one of the three phases of the motor. At each phase one of these hall sensors will activate and that is how positioning data is acquired.

I wouldn't call this a by product since it is the purpose of these sensors to detect position/speed the motor for controlling it.

When you also have a torque sensor like the tmm here, you have a power meter in your hand.

 

[PedalUma]

Let the bike dry fully and cool down slowly after a ride to prevent condensation with capillary action and air pressure drops creating suction. And to prevent freezing of the internal condensation build up. Let it slowly outgas moisture as vapor pressure. On these bikes the sensor bends with chain pressure to send the voltage resistance signal. Metals' bending characteristics change with changes in temperature. That is how old in-home thermostats worked. A bit of crud will also throw it off.
Change the wheel diameter to calibrate the speed. For example a 622 wheel is 700c, 28" and 29" solely depending on the tire size. I saw a bike yesterday that was calibrated for 26" wheels but it had five inch tires making it 31 inches in diameter. Each rotation goes 97.34 inches, not 81.64 on that bike. It is in the photo of our ride yesterday and was off by almost 20%.
I got to see snow yesterday! It was exactly where is belongs. On a mountain 60K away. Nice to look at.

 

[bluecat]

When you also have a torque sensor like the tmm here, you have a power meter in your hand.

• You have the speed, which is the result of the riders and motors energy
• You need to have the motors energy consumption
• You need to have the motors efficiency at the given speed
• You have the riders force, translated by the drivetrain
• You need to have the drivetrain's efficiency at the given gear

I would say, with a bucket full of assumptions, Stromer could bring a "Riders Power" value to the display. For entertainment, with a two decimal precision.

 

[Johnny]

• You have the speed, which is the result of the riders and motors energy
• You need to have the motors energy consumption
• You need to have the motors efficiency at the given speed
• You have the riders force, translated by the drivetrain
• You need to have the drivetrain's efficiency at the given gear

I would say, with a bucket full of assumptions, Stromer could bring a "Riders Power" value to the display. For entertainment, with a two decimal precision.

You are right on the drivetrain efficiency, which should be taken into account but on the derailleurs it can be taken around %95.

I really don't know why you are putting motor consumption/efficiency etc in to the mix. We are interested in rider's power only.

Tmm measures the torque that is applied by the rider and motor rpm is measured by the hall sensors in the motor.

Torque x Rpm/efficiency = Rider's input power, as simple as that.

I can not exactly say how accurate it will be but I would guess it to be way more accurate than %20.

 

[bluecat]

Torque x Rpm/efficiency = Rider's input power, as simple as that.

From my point of view, the RPM (=speed) are the sum of the riders and the motors power. How did you overcome this?

 

[Johnny]

From my point of view, the RPM (=speed) are the sum of the riders and the motors power. How did you overcome this?

RPM is just the rotation per minute(angular speed), it is not a measure of power.

Here, in hub power meters in general, the power is measured by the torque at the axle times the angular speed at the axle.

With the direct drive hub motors, angular speed at the rear axle is equal to the rpm of the motor. So we are lucky with these dd systems that we already have that information.

 

[bluecat]

RPM is just the rotation per minute(angular speed), it is not a measure of power.

Who claimed this? Please show me, how you make the calculation with the known values:

• Stromer Motor 300 RPM (approx. 40 km/h)
• Stromer TMM (torque sensor) value 1.5V

Feel free to use an arbitrary value (NM) for the TMM. I'm only curious about the calculation.

 

[Johnny]

Who claimed this? Please show me, how you make the calculation with the known values:

• Stromer Motor 300 RPM (approx. 40 km/h)
• Stromer TMM (torque sensor) value 1.5V

Feel free to use an arbitrary value (NM) for the TMM. I'm only curious about the calculation.

Without knowing sensors calibration this information is incomplete(the zero torque value is necessary, also you should say whether the sensor is in the high or low precision mode).
But let's make the calculation anyways:

According to the spec sheet on the website, at high sensitivity TMM sensor has 40nm/V, zero value range is 0.5-2V. If we assume zero value to be 0.5V, at 1.5V you will be measuring 40nm at the rear axle.
At 300rpm, which should be around 25mph for a 27.5" wheel, we are looking at a whopping 1260W!!!

Let's now assume something more reasonable, let the resting output(zero torque voltage) be calibrated to 1.3V. At 1.5V this translates to 8nm. If this is measured at 300 rpm that will translate into 250W at the rear axle.

Assuming %95 derailleur efficiency we get 250/0.95 = 263.1 W as the riders input.

 

[bluecat]

Without knowing sensors calibration this information is incomplete(the zero torque value is necessary, also you should say whether the sensor is in the high or low precision mode).

I fully agree with this.

I show you two examples of TMM voltage measurements. On both, the Stromer stays still, brake is pulled and one feet is on the pedal to create a value deviates from zero load. I see a 90kg rider who is out of the saddle and with his full weight is on the crank (25cm long) making 220NM. At 300 rpm, he's not in the direct gear, let's assume a 1:4 transmission. This will end up in 55NM at the rear axle. But most time, riders are in the saddle, so 8NM are realistic.

So, please show me the calculation how you came from 8NM at 300 RPM to 250W riders power (what is a remarkable power for Joe Sixpack).

 

[tbar23]

Ooops! Missing 2*Pi (Thanks to @Johnny for the correction.): 8 N-m at 300 rpm is NOT 250 W. It is more like NOT 40 W. 300 rpm is 5 rps so 8 N-m at 5 rps is just 40 N-m/s. Multiply by 2*Pi to get 251.3 W. This is what I get for trying to do math in a ski resort parking lot

And I think you have your transmission ratios inverted. The chainring multiplies the human torque (unless you’re in an MTB gear of something like 34:42). In all Stromer gears, the human rpm is lower than the hub rpm. So the 250 W human needs 32 N-m at 75 rpm (8 N-m at 300 rpm at axle).

This is a pretty good overview:
Power and torque

 

[Johnny]

8 N-m at 300 rpm is NOT 250 W. It is more like 40 W. 300 rpm is 5 rps so 8 N-m at 5 rps is just 40 N-m/s or 40 W.

It IS. Be careful about the units. Even in the link you have provided there is a small moment calculator app that will confirm this, try it out yourself. There is also a log scaled graph that you should easily see that this is correct.

The formula in terms of rps is : 2 PI x Angular speed (in RPS) x Torque (in NM). You are missing the 2 PI there.

So, please show me the calculation how you came from 8NM at 300 RPM to 250W riders power (what is a remarkable power for Joe Sixpack).

The formula is Power (In Watts ) = Torque (in nM) x Angular speed(In RPM) /9.55 . This is same formula above, we just converted it to be used with rpm, 2 PI x RPS = (2 PI /60) x RPM = RPM/9.55.
Plug the values in you get,
8x300/9.55 =251.3 ~ 250 W.

You take into account drivetrain losses after this(ie 251/0.95 ~264W).

For the two examples you are given, assuming that TMM is measuring 40nm/V,
first one 0.468x40=18.72nm
second 0.210x40=8.4nm.

You can plug them in the formula and make the computation yourself now.

 

[tbar23]

It IS. Be careful about the units. Even in the link you have provided there is a small moment calculator app that will confirm this, try it out yourself. There is also a log scaled graph that you should easily see that this is correct.

The formula in terms of rps is : 2 PI x Angular speed (in RPS) x Torque (in NM). You are missing the 2 PI there.


The formula is Power (In Watts ) = Torque (in nM) x Angular speed(In RPM) /9.55 . This is same formula above, we just converted it to be used with rpm, 2 PI x RPS = (2 PI /60) x RPM = RPM/9.55.
Plug the values in you get,
8x300/9.55 =251.3 ~ 250 W.

You take into account drivetrain losses after this(ie 251/0.95 ~264W).

For the two examples you are given, assuming that TMM is measuring 40nm/V,
first one 0.468x40=18.72nm
second 0.210x40=8.4nm.

You can plug them in the formula and make the computation yourself now.

Thank you @Johnny. It’s funny because my whole drive home I was thinking about the fact that my Stromer tends to add 500-600 W (on average) to my ride. With my missing 2*Pi that would have been over 100 N-m, and I knew the motor couldn’t do that.