You can find the guy’s posts on the forum and contact him, I forget the name of the company. He posted on the forum offering paid work to consult on a motor driver, I don’t think he mentioned the application in that post though.
As I discussed with him, if you want the rotational motion to be highly consistent there is practically no point in closed loop for the actual motor control. You could still use an encoder to grab the position of the motor with good reliability in an instant. However for actual motor control, if you use closed loop that will introduce variations in RPM that will always be greater than what you get with open loop, for a simple rotating mirror with no other load and good bearings etc. With open loop, your rotational speed is very accurately determined by the frequency of the 3 phase AC signal, which is pretty closely linked to the crystal oscillator.
With closed loop you will have a PID or something and things will ramp up to the commanded RPM, overshoot a bit, bounce around if it’s disturbed, theoretically it will never quite match it etc.
They guy did an interview with me and I told him all kinds of stuff he just didn’t want to hear, like the reality that simply contributing to the open source driver project and then using that is dramatically more sensible than trying to develop your own like that. And he was bend on using closed loop and still getting tremendous rotational speed accuracy etc., not good engineering.
I have ordered some motors from Constar before for testing (specifically the B1230 line). If you’re in Europe they refer you to their distributor Kocomotion. For small oder amounts you’re looking at about €30-40 per motor, plus shipment. Usually takes 8-10 weeks to arrive. I’m pretty sure they only do B2B tho.
My opinion about the AS5048 is split: The SPI mode is fast, but the PWM mode sucks big time. On the pictures it looks like they don’t provide SPI-wiring?
OTOH the AS5600 has a mediocre I2C mode which is OK for slower motion and a fast, but noisier analog mode.
If you’re OK with a max speed of ~50rad/s you could get away with the cheaper AS5600.
Maybe the dev team is working on improving the I2C protocol? There were some pointers, what could be improved lately…
I am using 1MHz instead of 400kHz and didn’t run into problems yet.
Thanks @o_lampe .
The 1105 and 1806 come up with the as5048a only. They aren’t available with the as5600, not sure why (bigger form factor of the standard as5600 PCB?).
On these cheap bundles, in my experience, the default encoder wiring is usually pwm, but you can solder the spi instead.
Can you please elaborate on the pwm issue with the as5048a? In the past, I usually went with pwm on the as5048a (by pure laziness) and never had any problem.
AFAIK, the loop time is slowest. At least, when you use interrupts.
I only tried it once, when I was new to sFOC. Couldn’t get it to work, but was partly my inexperience’s fault.
@runger , at last I’ve received the quotation for the Constar BO1410NB2B:
10 pcs: USD30
100 pcs: USD17
1000 pcs: USD10.5
This is 3 times more expensive than the brands I’ve mentioned above in this thread. For this reason, and also because Constar doesn’t provide bundles with magnetic encoders, I will not follow up with them.