Motor Encoder Noise Issue - Please Help

[pmtwiss]

Hello,

I am using the Avago AEDR-8500-120 IC for a motor rotary controller, and I am having large noise transients on the 5Vdc into the IC, and on the encoder lines. I am a well experienced PCB design, but new to motors.

I have attached the schematic to this email. It's fairly simple.

The board is a 4-layer design. The encoder traces on on the bottom, while the motor signals are on the top. The 2 internal layers are solid ground place (mid1), and a misc signal layer with ground pour (mid2). I am following avago's recommendations. If I manually turn the motor by hand, the encoder signals are clean, but when we move the motor using power, the encoder signals are all corrupted.

I have a cable going to a commercial controller, and it seems the cable is not an issue due to the fact it's used with other various off the shelf motors.The noise is a decaying sinusoidal(66MHz), coming every 48KHz, while the motor runs at 24 KHz.

It seems I may be missing something. I have my ground stitched with vias, I have a pi network for the encoder IC. I think the noise is happening on the 5Vcln, as the noise on 5Vcln > noise 5V, which doesn't make sense given the filter. Yes I chose the ferrite bead correctly, using R,XL,Z table and frequency.

Due to my inexperience in motor electronics, I am looking for some guidance on how to figure out where the noise is coming from.

Please see attached schematic in PDF for review.

Please help if you can! And thanks,
-Peter

- - - Updated - - -

Hey Guys,

See the added PDF here showing some scope signals with the schematic, as well as PCB stackup.

Thanks,
-Peter

 

[D.A.(Tony)Stewart]

Common mode spikes can cause radiated noise from large motor commutation. Usually large CM choke is needed on motor cable to prevent ingress like those on VGA cables.

Also ensure measurement is done properly with very short probe ground or none using tip and probe ring to test pins added, or use two probes in A-B mode such that when connected to same point, you get a flatline. then you know your probe CMRR is good and move one to near chip ground.

If this still sees spikes, your 1uF Cap has poor ESR. Choose a better Cap.

If this still fails check your sensor is 1mm gap.

 

[pmtwiss]

That's what we suspected as well so we did a quick test and broke out the motor leads on a separate cable from the power/signal lines. The same noise was still present. Would a choke on the motor leads help if they are on separate cables?

 

[D.A.(Tony)Stewart]

That's what we suspected as well so we did a quick test and broke out the motor leads on a separate cable from the power/signal lines. The same noise was still present. Would a choke on the motor leads help if they are on separate cables?

yes also on encoder signals. ,!

Consider loading and/buffering the signal to suppress stray coupling also with appropriate impedance and LPF. the driver is not meant for long paths.

 

[HTA]

It is either cross talk on the line or board or ground loops. How and where are the two grounds connected?

Enjoy your trouble shooting work!

 

[FvM]

Your observations suggests that the interferences are caused by the motor PWM. No details have been yet revealed about the driver circuit or if it implements any output filtering.

Grounding the motor case might possibly help, also inductors or low pass filter at the driver output. Pay attention that the motor wires or motor case don't couple directly to the encoder. The encoder ground must be isolated from motor case.

 

[D.A.(Tony)Stewart]

That's what we suspected as well so we did a quick test and broke out the motor leads on a separate cable from the power/signal lines. The same noise was still present. Would a choke on the motor leads help if they are on separate cables?

With proper short probe short ground leads or coaxial probe, it is pretty easy to isolate if it is conducted or radiated noise. A shorted probe with ground loop to probe can sniff radiated noise waved near offending cable. conducted noise can be eliminated by low ESR filter cap. radiated noise by CM choke and small differential RF cap. (100pF)

 

[HTA]

Another check is to see if the encoder signals are not proper terminated or high impedance on the receiving end becase crosstalk can be higher.

Enjoy your trouble shooting work!