It just occured to me that when operated as intended the user is being pushed beyond his ability to resist the motor. So what happens if the user slightly reduces their effort during the move? With decreased load, the speed of a motor increases right?
With brushed motors that is certainly true and could cause injury because the user was already at max effort, now the system has higher momentum so it will require greater than max effort to slow the system to its original speed.
FOC brushless being electronically commutated can limit the commutation speed increase, but the instantaneous acceleration caused by reduction of slip angle still happens. It is controlled to a degree under FOC.
Steppers advance by a known fixed amount at a torque proportional to current. We cannot control the instantaneous slip angle, but we can guarantee that it is less than or equal to the step angle.
I think that makes steppers the safest to experiment with initially, but brushless may be better once the actual reaction speeds and forces are known.
First of all, thanks for all the additional data you provided, really appreciated.
My plan is to monitor the motor speed and limit it to a safe amount. Basically apply the calculated torque for a specific exercise and monitor the speed. If the speed exceeds a safe limit, decrease torque until speed is back to safe limit. If speed slows to 0, possibly increase torque slightly to keep moving again.
With proper gearing and a decent encoder on the motor shaft, it should be easy to monitor even small changes to the actual end user speed. But, yes, good points, thanks for raising them
You’re welcome. I’d like to see what you come up with.
The problem with relying on speed monitoring when human safety is involved is sensors can fail. Encoders can slip on shafts. When they fail, they indicate the motor has stopped when it hasn’t. If that happens, your machine might try to crank up the speed.
If the user is knuckle deep into a pulley, that is not a desirable failure mode.
I had it in my head that steppers would allow you to take a step, check if everything is good, then take the next step only if everything is OK. But as I type this I realize that logic is flawed because if the sensor is goobered, it doesn’t matter that there is a well defined period to check the sensor. It is still goobered in the same way and will fail the same way.
But that probably wont happen. You’re probably good.
Assuming a magnetic encoder with a magnet on the shaft, you can monitor the proper working on the loading phase, then monitor it while moving with torque applied. If the magnet falls from the shaft, the encoder will give an error reading, and an emergency stop can be triggered.
Since the encoder is monitored hundreds if not thousands of times a second in a tight FOC loop, I cannot possibly see a failure mode than puts the user in danger. I mean, it’s good that you made me think about it, but I also think that it can be easily addressed now that I’m aware of it.
With a mechanical encoder, I agree that problems can happen. That’s why a magnetic encoder (SPI or I2C) is preferred
Well, I had to learn it the hard way, but encoder failures are not always binary. There are situations, when the sensor readings are just slightly off because of a slipping magnet, which is just loose enough to move a little bit in its fixture. Since then I regularly check the sensor at startup, by spinning one turn back and forth, performing some plausibility checks on the sensor readings while doing so and monitoring the magnetic field strength as well as sensor error flags. You should be really careful about that if human beings are in danger. In my case it did only cost me driver board at a certain point (plus a lot of time).
I’m not claiming that sensors don’t fail, just that it can be monitored at least in this case.
And the user is never really in danger: the motor operates only if the user is actively closing a contact (pushbutton or similar), and all movements are eccentric and with a safety end switch (max movement).
But, yes, thanks for all the warnings, always good to be on the lookout for problems.