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