7i76e and opto ancoder TCST2103

I have found one that seem to be isolated from DC GND/+19 to 230V and earth.

Little progress, the spindle does a little move before my house go out of power so i suscect they are not really isolated...

I have some trouble for understand why the +15v from KBMG is not enough because i found some correct voltage on the sig input.

I'm another time completely lost with this stupid spindle

Adding a 2200uF 25v capacitor to the +15 from kbmg allow a working spindle !!

Hi

Another day another problem

Now with encoder that display very unstable speed (without pid), the spindle seem to be stable.

Quadrature with filter = 1


Quadrature with filter = 0


Counter mode with filter = 1


Counter mode with filter = 0


With my oldest board (BBB + furaday cape encoder managed by si8640bc and PRU encoder) i does not have this problem.
(with my opto tcst2103 this cape need pull down 560ohm from encoder input A/B/I)

Did you think this is some electrical noise or maybe not enough precisely mounted encoder A/B ?

Hal and ini attached


Also see that higher speed = more stable speed but not fully clean

It might be electrical noise issue, the last plot seems to have a periodic error that may
be a wheel or quadrature issue.

Since this is a low resolution encoder, lowering the encoder input bandwidth will help
if this is an electrical noise issue.

One way to do this is enable the encoder filter and lower the sampling rate to say
500 KHz:

setp hm2_7i96.0.encoder.sample-frequency 500000

You can check for wheel or quadrature errors by slowly rotating the spindle and
plotting the A and B signals:

hm2_7i96.0.encoder.00.input-a
hm2_7i96.0.encoder.00.input-b

In halscope. These signals will also reflect the filtered encoder inputs so can be checked for noise

950RPM filter = 1 sample-frequency 500000


3350 RPM filter = 1 sample-frequency 500000



450 RPM filter = 1 sample-frequency 200000 Counter mode 1


With the condensator hm2_7i96.0.7i83.0.0.fieldvoltage when spindle run is from 13.4 to 13.7

Did lowering the sample frequency fix the near 0 velocity spikes in the velocity?
If not, that suggests a quadrature error

Note that you should only run at say 30 RPM max and 1 KHz sample rate
otherwise you have a data rate much faster than the sample rate so relatively
meaningless data

For spike i'm unsure for now, need more test but i think not fully better

With counter mode this is without quadrature and i see also some spike


edit : i see something strange the encoder.input-index is allways = 1, i suspect this is reverted
check with multimeter :
aligned to the drill = 0v
aligned to undrilled = 5v

connected like the NEW schematics is according to the result, but maybee it is needed to invert something ?

Full wirering from my older cape single ended input for A B I

I have also one time wirering and used a Nema encoder A+ B+ I+ to this cape encoder input.

One change is that you had a 560 Ohm load before and now have a 2000 Ohm load
This may have affected the quadrature symmetry. You may be able to adjust this by lowering the LED current.

Note that you cannot actually see the quadrature signals accurately in halscope unless
you rotate the spindle very slowly

BTW you should not use any "D" term in a velocity control PID loop, just FF0 P and I