I have a question about the an application where I want the odrive to provide a power level below spark ignition. This means that voltage will be at 2v and current between 100 and 200mA. I will also be heavily dependant on the anti cogging functionality.
I learned that some industrial drives have difficulties providing those low power outputs, so let’s try Odrive.
I will probably be designing the output stage before the inverter myself as I need to add some safety measures to ensure this low output.
Are there any reasons why Odrive would have difficulties providing low power, keeping in mind that I need to use the anti-cogging.
The default current shunts will need to be replaced, as they have ~ 500mA of noise. So you’d need shunts that are probably… 100x larger? That gives you 5mA of noise and about 1.2A of full range.
And of course ODrive needs at least 8V to run the board. The voltage seen at the motor wires is different, but it (and the encoders) won’t run at 2V.
Do not use ODrive in an explosive, dust ignition, or other hazardous environment.
I would recommend a linear drive, not a PWM drive (such as ODrive and the vast majority of industrial drives) for this application.
Reason being: ODrive cannot output less than 10V, never mind 2V. It can output something that looks like 2V by outputting 10V at a high frequency square wave, with 20% duty cycle.
It also cannot guarantee to limit the current. If the inductance is low, the current will rise faster than the controller rate (8kHz) and the ODrive can easily overshoot its current setpoint.
Industrial PWM drives will have the same limitations.
You could reduce this overshoot and make the current controller more stable by adding 3x large inductors, one per phase. The same would work for industrial PWM drives.
However, no PWM drive even with high inductance load can guarantee to a safety certifiable standard that it will never overshoot its current. Whats worse, is that the inductance you add for controllability will also cause sparking in the case of a break in the circuit. So PWM drives are just a big problem here, I think.
A linear drive works by using transistors in the linear region, throwing away the excess voltage as heat. So they are very inefficient (if you supply 10V and want 2V then you will throw away 80% of the energy in the transistors themselves, so you need a big heatsink)
But they have the advantage of being noise-free, with perfect open-loop voltage control. They don’t require high inductance loads, and they can guarantee to never overshoot on voltage or current.
e.g. BL Series Linear Amplifiers – Aerotech US
Minimum load inductance: 0
That’s how you know it’s a linear drive. That and the big heatsink for a low power rating.
you guessed correct
“Do not use ODrive in an explosive, dust ignition, or other hazardous environment.”
it is not the Odrive but the motor that will be in explosive atmosphere. The voltage and current will be limitied be external components, but any overshoot that might occur can also blow the safety components.
I was mostly triggered by the coggging compensation options the odrive offers and the option the self develope the board further for this specific application. I was also pointed into this direction :
But that overshoots the budget I have a feeling Aearotech might also be in that category ( will find out)
My initial thought was to re-design the output stage (make it linear) and to feed the current/voltage information back into the Odrive. Would that be an option or do I now greatly oversee many complicated things?
Indeed, so there is a safety function on ODrive to stay inside its configured current range, and not suddenly apply 9V, ie its minimum supply voltage, to the motor. You will have to validate/certify this requirement yourself, since ODrive is not a safety certified product.
But depending on what your external protections are (100mA fuses would be a good start - I was thinking also zeners, but they could confuse the current controller), it might be possible to say that the ODrive cannot blow them if you power it from 9V DC, for example.
I have ‘used’ the S700. (tried it, alongside drives from Elmo, Copley and Moog, and eventually sent the S700 back to Kollmorgen. It is a very big and loud machine (but yet still uses PWM so I believe) that must be powered from at least 110VAC (200V DC Bus) but ideally wants 415V mains (600V DC Bus). The front-end X/Y capacitors are so big that they will trip most RCDs.
I think they will struggle to run a motor at 2V 100mA.
Also they trumpet a lot about functional safety, but i’m not sure if they will certify anything that is not a Kollmorgen motor from their catalogue, with a Kollmorgen encoder. And all that will set you back about 10-20k per axis.
That could work, but would involve a LOT of software (firmware) redevelopment, since the ODrive is very much a PWM controller.
I wonder: If you replaced each phase output with a capacitor to ground, (i.e. turn it into a DAC) and wired it from there into a linear amp, it might work.
You would need to move the current shunts to the real output and perhaps add your own amplifier…
Yes, for this application the normal safety components are zener diode and current limiting resistor, protected by a fuse to ensure the power gets cut and the zener does not go above it’s rated values.
In fact maybe due to the inline resistor the voltage to be provided gets pumped to about 10volt from which the resistor cuts 8 volt away. Any chance this extra load tempers the odrive a bit ?
I follow your idea about connecting capacitors to the outputs and taking the resulting voltage into an amp, but does that still give any ( the correct ) current feedback back to the odrive ?
Another thing with this solution is that I should limit the power before the inverter otherwise I have three field lines to limit, and the power of the three lines combined have to be under the power limit in that case only 1/3 of the power pet line. ( Hope you understand)
I am also very happy with the faulhaber drives, these can do magical things on verry low power levels, but they are still PWM…
