Best way to get rid of this noise on encoders from motor?

[Triode]

So this trace is from the encoder on a large brushless motor that runs at low speed. 24V supply, about 6 amps when running, accelerating from a stop.

It seems to work pretty well just by putting a 10 nanofarad capacitor next to ground on the input, but I'm not sure that's the best way. And it may still be getting some false counts.

Because of the location of the motor driver the wire runs are about 0.6m, ~2 ft long.

That pulse is the motor driver making tiny adjustments back and forth I'm pretty sure. This is showing without the capacitor filter so you can see the full noise.

Filtered the noise is attentuated about 10x and the false counts seem to stop, I can't put a much larger capacitor on there because it would filter the pulses from the encoder too much.

Would an inline inductor help? Maybe sheild the signal wires and grounding the sheild to the driver ground?

At max speed the encoder pulses should get to be about 15kHz, 5 volts, 50% duty cycle. It works if they are rounded off as long as the edge of one phase is still above 3.5V or so when the other phase changes. Experimentally I determined that 47nF was too much for that to work. 10nF seems to work though it rounds it off a bit.

I could even put a capture board right by the motor with some filtering and to convert the signal to digital or current. But that seems a bit overkill.

I'm sure there is some well known best approach here. What would you all recommend?

 

[rjenkinsgb]

Screened twisted pair should do it, I'd have thought.

If there is still any noise present, a common mode choke should reduce it further - a ferrite sleeve, or core with a few turns of the entire screened cable through it.

 

[Triode]

Thanks, I haven't actually put shielded twisted pair on it yet. I will try that when some comes in.

But I wonder if it's being picked up in the wire, the encoders are right next to the coils, but I suppose they don't contain very long traces.

 

[MaxHeadRoom78]

What is the nature of the encoders, and the output form?
Are they quadrature output? 5v or 24vdc?
In motion control, typically differential line driver/reciever RS485 IC's are used.

 

[rjenkinsgb]

If it is a quadrature encoder with complementary outputs per phase, then it does really need an RS422 line receiver and terminating resistor on each pair.
I initially took it to be a magnetic / reluctance pickup...

It should still be using twisted pair cable though; most encoder manufacturers are very specific on shielding and grounding requirements.

If it's single ended, then you can still use the RS422 line receivers but bias the other input on each to the mid voltage of the signal.

 

[MaxHeadRoom78]

If single ended outputs, I just use a 8pin dual RS285 TX IC, and a 8pin RX IC. for both A-B outputs.
There are a few, 75ALS191 and UA9637 etc.

 

[Triode]

Thanks! that's really useful to know about