2
\$\begingroup\$

I have a chassis that is made from box steel beams, they are painted/surfaced. Attached to it is a fiberglass tub which can rotate. Attached to the fibreglass tub are supports and mounts where some absolute encoders (EMS22A30-C28-MS6) are attached. The encoders do not have physical conductivity to the steel frame. Where mechanical connections need to interact/anchor with the frame they are done via rope/nylon-like lines.

The electrical component of the system is an Intel NUC which is powered by an ordinary DC power adapter. The NUC then powers an Arduino Due via USB. Below is a (hopefully) representative schematic of the electrical components (done from memory because I don't have access to it for a few days). The wires that connect from the Arduino to the encoders are about 1.5m long. They are just single wire cables bundled together with molex connectors on the ends. schematic

I am finding that intermittently the encoders will report really strange values, like suddenly many bits are being truncated. Values of 300-500 will suddenly be seen as 10-100 ish. I became able to reproduce the issue if I:

  • Had a wheeled desk chair wheel along the vinyl-like flooring
  • Then that desk chair touched the steel frame with the metal parts of the frame of the chair

I've monitored the supply lines to the encoders during this event and sure enough there's a huge spike:huge spike

How do I mitigate this? I have tried earthing the chassis, which then ties signal ground to earth (via HDMI of the NUC through the monitor then via USB) however the problem still remains. The NUC is mounted to the frame chassis however the bracket isn't conductive (I could change that). It seems that potentially I should also be earthing the Arduino Due too, to the chassis however I'm unsure of this detail since the Arduino doesn't really have mounting holes that are connected to signal ground. I can connect one of the many GND pins, however not sure if most appropriate.

Finally, is this some kind of induced interference and potentially not grounding/earthing related? My cable runs to the encoders are currently just ordinary wire bundles, not twisted and not shielded. Would testing alternative cable methods be useful? Or should I look at some kind of choke/diode situation?

\$\endgroup\$

1 Answer 1

2
\$\begingroup\$

It's hard to say exactly what is causing the ringing. There are two pathways, one is static fields and the other is radio frequencies that sparks can generate.

The best thing would be to provide a ground or shield by using shielded cable to the encoder and if the encoders have a chassis. A shield would attenuate radio frequencies and also provide something like a faraday cage to block changes in electrostatic fields and RFI.

\$\endgroup\$
2
  • \$\begingroup\$ Yeah, I should go ahead and try shielded cable to see where that gets me. One thing that I'm unsure about though is that the encoders themselves do have a chassis, however it's not easily accessible to mount a shield anchor point to for the cable. I haven't checked if the encoder's signal ground is connected to the chassis yet or not. If there's no easy way for me to connect the shielded cable to the encoder, is it bad to only have one end of the shielded cable anchored (at the controller end)? Or is that pointless? \$\endgroup\$
    – Clammet
    Jul 8, 2021 at 1:16
  • 1
    \$\begingroup\$ No it's actually quite common to only have one end grounded as sometimes chassis can create common mode noise it cables. For sensors (in general) it is a good idea to have the shield connected to a chassis that surrounds the sensor (like a faraday cage) \$\endgroup\$
    – Voltage Spike
    Jul 8, 2021 at 1:22

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.