0
\$\begingroup\$

I have two encoders that emit A/B signals for a total of 4 wires. Not sure if this will be relevant but these encoders are not the same and have different output impedances and rise/fall time. I also have two motor power wires (nominal 9V motor, operating current usually below 0.5 A) as well as a 5V wire and a corresponding GND wire.

The problem that I am running into is when I bundle all eight wires together, I am getting crosstalk between the encoder lines that show up as oscillations. Because these oscillations exceed the maximum input voltage limits of its receiving device, a LS7366R quadrature counter, I am trying to dampen these oscillations somehow. Oscilloscope screenshots are shown below that illustrate the oscillations: Oscillation 1

Oscillation 2

The maximum input limits are -0.35V to 5.3V according to the LS7366R datasheet, and both of those oscillations exceed that limit. I am looking for a solution to combat this - I was initially going to put a ceramic capacitor to ground for the encoder channels, but I am worried about limiting my bandwidth (would this even work?). What other quick solutions can I implement?

\$\endgroup\$
4
  • 1
    \$\begingroup\$ What are you encoding that has MHz bandwidth?! \$\endgroup\$
    – Tim Williams
    Commented Oct 10, 2023 at 22:33
  • \$\begingroup\$ What encoders they are, do they have data sheets, and what kind of outputs they have, is there any interface circuitry in between? Simply putting a ceramic capacitor there alone can make things even worse, as the encoder may just need to sink and source more current. Please note that when the purple channel goes up, all other channels have a slight dip. That may be ground bounce, not coupling between wires. \$\endgroup\$
    – Justme
    Commented Oct 10, 2023 at 23:06
  • \$\begingroup\$ @TimWilliams the system is basically an input encoder which the motor is trying to track. Because of ratio between input and output, the motor spins very fast to catch up to the input encoder hence the quick pulses. \$\endgroup\$
    – petemoss0
    Commented Oct 11, 2023 at 18:28
  • \$\begingroup\$ @Justme yes they do have datasheets. Both of the encoders have line drivers due to the long cable length (~6ft). I actually did attempt to add a capacitor - I tried it with 0.1 uF because that was the smallest I had on hand, and I got some sort of ringing before the rising edge like this however I'm not sure about the exact mechanics of this ringing, do you have an idea? I think it could be reflection arising from impedance mismatch, as the capacitor leads seem too short to add any significant inductance effects. \$\endgroup\$
    – petemoss0
    Commented Oct 11, 2023 at 18:33

1 Answer 1

Reset to default
1
\$\begingroup\$

Crosstalk on cables can happen both between wires and through a shared ground.

When there are also signal integrity problems such as unterminated or mis-terminated lines, the resulting reflections are spiky high-frequency signals that readily couple to neighbor wires.

Some things to work on:

  • Make sure there is adequate ground return for the signals. If possible, separate the ground returns for each channel in the cable (consider using twisted pairs).
  • Match driver impedance to the cable - add series resistor. This reduces edge rate and also absorbs reflections.
  • Add parallel termination at the receiver. This stops reflections.

If these don't work, consider using differential drivers and receivers to buffer your signals.

\$\endgroup\$
1
  • \$\begingroup\$ thanks for your suggestion, parallel termination helped a lot. the other options of differential signaling I could not try because of aspects of design that were frozen. however I was still getting some oscillations here and there from the other lines, so I will consider using clamping diodes from here on out, but this answer helped me understand the problem a lot more. \$\endgroup\$
    – petemoss0
    Commented Oct 12, 2023 at 15:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.