1
\$\begingroup\$

I am currently building a cable and would like to possible reduce any EMI. Three signals ( GROUND, ANALOG SIGNAL AND VCC) are connected to the encoder. I plan on twisting the Ground and VCC to cancel EMI and shield the twisted wire with aluminum foil. What should I do with the Analog Signal to reduce EMI? I assume the cable to be around 15 ft long. Note: There are motors, processors and radios around.

\$\endgroup\$
  • \$\begingroup\$ Where is the EMI coming from? Is it crosstalk or radiated? How long is the cable? \$\endgroup\$ – Voltage Spike Jun 12 at 18:41
  • \$\begingroup\$ What is the device you are feeding the encoder signal to? \$\endgroup\$ – DKNguyen Jun 12 at 18:43
  • \$\begingroup\$ feeding to DAQ. \$\endgroup\$ – Sam Shurp Jun 12 at 18:44
  • \$\begingroup\$ What's causing the EMI? A motor? Or are you getting cross talk from one line appearing on the other line? If you could show us a scope image of the interference, that would help. \$\endgroup\$ – DKNguyen Jun 12 at 18:45
  • \$\begingroup\$ I am currently building a cable and would likely to possible reduce any EMI. There are radios, processors and motors. \$\endgroup\$ – Sam Shurp Jun 12 at 18:47
2
\$\begingroup\$

A three-wire cable is not ideal for such a situation; if small amounts of power are needed, and the power is not switched or current-modulated, the return current in the GROUND wire might be constant enough not to couple power noise into your signal, but the better solution is NOT to have a 'common' wire for those two functions. Remember, the I*R voltage drop in this wire IS added to the signal.

So, a four-wire cable (and if there are connectors, those too have some resistance, so use four-position connectors) will work better. You can use common-mode chokes on the signal-wire pair, and a true differential amplifier as the signal receiver, so it will only have a 'ground' connection at ONE end, and no ground-loop pickup will result. The twisting of the data-wire pair will help reject B field (adjacent motors etc.) pickup, but (at some extra expense) a coaxial cable will do even better. If high voltages were present, capacitive noise coupling to the twisted pair is likely, and the coaxial cable is the best choice.

This is not an unusual situation, and conbination coaxial/wire-pair cable is available commercially (Belden #549945, for instance), or if you want something off-the-shelf, PC video cables (VGA) or S-video cables (two coaxial pairs and overall shield) are readily available. Assembling your own conductors is problematic, because an overall sheath is usually required, and molded-on sheathing is preferred to wrapping with lots of black tape...

There's no way to prevent radio pickup, but common-mode chokes can block it from your receiver without hurting the difference signal that is detected. Other RF filtering depends on your data frequencies being different from the interfering signal, but is often required.

\$\endgroup\$
1
\$\begingroup\$

I plan on twisting the Analog Ground and Analog Signal to cancel EMI and shield the twisted wire with aluminum foil.

Sounds good.

What should I do with the Vcc to reduce EMI?

I don't think there's anything you can do really, except for placing the VCC far away from your Analog Signal.

But, I'm not 100% certain how you've thought you are going to connect ground, so I've made art in paint showing how you should connect the ground to the different parts. Connecting it up as in my paint art below should protect your signal from your VCC and other radiation noises due to good ground shielding.

A slight problem is that the shielded wire will have a characteristic impedance. This means that if your analog signal contains frequencies in the 400 Mhz region ((C*2/3)/15ft ~ 43 Mhz, reflections starts getting important when you can fit ~10 periods inside a transmission line). Right, if you got those frequencies, then you should buy a proper shielded cable and do some proper transmission termination at the entrances of the shielded cable.

Edit: Though, now when I think about it, I think it's a little bit overkill to twist it as well. You should be fine with just shielding it and connecting the shield to only one side.

enter image description here

\$\endgroup\$
  • 1
    \$\begingroup\$ There are a lot of answers to a single problem. Most answers kind of makes sense. I will twist the signal wire and returning signal wire and shield them. Then completely shield the all the wires together. \$\endgroup\$ – Sam Shurp Jun 13 at 17:45
  • \$\begingroup\$ @SamShurp Keep in mind that you don't have to accept an answer straight away, you can just accept in a week or two, maybe others with some more expertise drops an answer or two. I've only dealt with these things in theory and in simulations. I've so far never dealt with it physically in real life. \$\endgroup\$ – Harry Svensson Jun 13 at 20:22
0
\$\begingroup\$

Twisting the wires won't do anything here since you don't have a differential pair. I don't think shielding will do much either unless terminated properly. If you add shielding it's going to have to be properly 360 degrees terminated to the enclosure shields at both ends and that can be tricky.

My recommendation is for you to you build an active cable that has a single-ended to differential amplifier (like an AD8476) at one end with a differential amplifier at the other end, unless your DAQ can accept that directly. By directly, I mean that your DAQ has a balanced/differential input whose input impedance for both lines is equal so you get common mode rejection. That converts your signal to a balanced signal that you transmit over a twisted pair.

Might still not be enough though at 15 feet...but as far as I'm concerned that's the minimum that you need.

\$\endgroup\$
  • 1
    \$\begingroup\$ "Twisting the wires won't do anything" Since the signal is flowing through the Analog Signal and returning through the Analog Ground, you expect the magnetic field to cancel if you twist. That was what I thought. \$\endgroup\$ – Sam Shurp Jun 12 at 18:59
  • 1
    \$\begingroup\$ There's a bit more to it than that. The input impedance of both lines has to be balanced for that kind of thing to work or else the currents induced in the wires won't be equal. That's why you convert it to a balanced transmission before transmitting it on twisted pair. \$\endgroup\$ – DKNguyen Jun 12 at 19:00
  • \$\begingroup\$ Will think over your answer, and also wait for answers from others. Thanks \$\endgroup\$ – Sam Shurp Jun 12 at 19:02
-1
\$\begingroup\$

Twisting your signal with your ground just gets it that much closer to Vcc noise. Keep in mine that current spikes on Vcc have ground as their return path. It would be better to split the ground at the source into 2 wires and make 2 twisted pairs. Also, remember to connect the shield (aluminum foil) to ground as well, or it won't have much effect.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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