Hello everyone,
I have my controller with HW board V1.2 working very well on bench with a 15Kw three phase induction motor.
The problem I have is in the choice of encoder technology.
Initially I started with a capacitive type encoder (CUI INC AMT10) but I could not solve the noise introduced by the motor in the encoder signals, I played a lot with different types of shielding but I never got something stable.
Then I tried with an industrial type encoder (I do not remember the model) which is of 3600 pulses per revolution, with that it worked well, but when it exceeded 2200rpm the motor began to hit and to have very abrupt and anomalous operation. What I assumed was that the pulse frequency was too high for the STM32 processing.
Now I make a 60 teeth toothed wheel with a printer optical encoder, something very homemade and not very robust, and it works perfectly, I could take my motor up to 9000rpm! and it runs very smoothly.
I think that the optical option in this environment of a lot of inductive noise is the best option, but I can't get something robust and industrial to work in automotive conditions.
What do you recommend?
What would be the "optimum" number of pulses to use?
I appreciate any comments
Thanks!
Encoder for sine
- johu
- Site Admin
- Posts: 5821
- Joined: Thu Nov 08, 2018 10:52 pm
- Location: Kassel/Germany
- Has thanked: 162 times
- Been thanked: 1044 times
- Contact:
Re: Encoder for sine
On board there is a roughly 60 kHz low pass for the encoder signal that's why the 3600 teeth version topped out. You can increase the cutoff frequency but in general I'd roll with a low pulse encoder like the one you made, since precision is not needed.
Support R/D and forum on Patreon: https://patreon.com/openinverter - Subscribe on odysee: https://odysee.com/@openinverter:9
Re: Encoder for sine
Hi!
Yes, I tuned that low pass filter acording to the 3600ppr frequency encoder, and can see nice pulses with osciloscope but still fail at 2200rpm.
-
- Posts: 3266
- Joined: Mon Dec 24, 2018 1:37 pm
- Location: Slovenia
- Has thanked: 80 times
- Been thanked: 234 times
- Contact:
Re: Encoder for sine
I figured long time ago 60 slot optical sensor is optimal also for lower speed. We are talking about ACIM motors yes?madius wrote: ↑Thu Nov 03, 2022 4:14 pm
I think that the optical option in this environment of a lot of inductive noise is the best option, but I can't get something robust and industrial to work in automotive conditions.
What do you recommend?
What would be the "optimum" number of pulses to use?
I appreciate any comments
Thanks!
See here how i made my first sensor when my AB OEM sensor failed .
https://mazdamx3ev.wordpress.com/2015/0 ... er-senzor/
When i was satisfied with performance i used a lot of my daughters nail polish over the sensor so it is protected from corrosion. Nail polish can also be removed by acetone if you need to make some repairs to sensor later...
If you have too much noise then possibly it is not encoder fault. You should make sure your inverter is correctly setup for your motor.
Re: Encoder for sine
nice build!arber333 wrote: ↑Thu Nov 03, 2022 7:39 pm I figured long time ago 60 slot optical sensor is optimal also for lower speed. We are talking about ACIM motors yes?
See here how i made my first sensor when my AB OEM sensor failed .
https://mazdamx3ev.wordpress.com/2015/0 ... er-senzor/
When i was satisfied with performance i used a lot of my daughters nail polish over the sensor so it is protected from corrosion. Nail polish can also be removed by acetone if you need to make some repairs to sensor later...
If you have too much noise then possibly it is not encoder fault. You should make sure your inverter is correctly setup for your motor.
that's exactly what I have built now, but I 3D printed the wheel because it was just for testing.
Yes yes, it's an ACIM motor.
- Romale
- Posts: 452
- Joined: Fri May 20, 2022 4:16 pm
- Location: Romania
- Has thanked: 216 times
- Been thanked: 49 times
Re: Encoder for sine
I used the camshaft gear from the engine timing system. only I ground a little in the lathe so that the length of the tooth surface was equal to the distance between the teeth. the sensor itself was used from a Tesla car, it is cheap and very reliable. it turned out 42 pulses per revolution and high reliability for all conditions of use.
Previously, I used a sensor on one microscope of my own manufacture (as5047p) and programmed it for 50 pulses per revolution. this also gave high-quality work, but requires protection from the external environment.
Previously, I used a sensor on one microscope of my own manufacture (as5047p) and programmed it for 50 pulses per revolution. this also gave high-quality work, but requires protection from the external environment.
evil neodymium