Analog Signal Noise

First you need to ascertain, or tell us you know, that the noise is indeed coming from the high voltage mains lines. If that is the case, then moving the cable closer to mains lines should increase noise, and vice versa.

If that is the case, then the coupling is capacitive, not inductive. The twist would have eliminated any inductive coupling. The grounded shield should eliminate the capacitive coupling. To what is the shield ground terminated? It should be terminated to the circuit ground at the receiving end of the cable.

If the noise is not from the high voltage mains, but instead coming in from an rf transmitter, then the shield must be terminated at both ends, and a shield drain wire is no longer a good ground - you should terminate the entire shield at the cable connector strain relief (assumed conductive) and the connector should be making a good ohmic connection (milliohms) to the equipment case, not the circuit ground. In other words, if the problem is rf, do not run the shield drain wire through the connector on a pin, take it to the equipment enclosure and let that act as a Faraday shield to keep the rf from getting inside the equipment.

That's about all I can give you with the question as you posed it. More info on the nature of the noise will result in more precise answers from readers.

Good points, emc c. I should add this: if there is rf interference on the twisted cable you could cut it off by using a ferrite bead (ring type, dedicated for the interfering frequencies). Wind the twisted wires (together, as if they were "one wire") around the ferrite bead. You could, also, add small caps (e.g. 100pF or so), after the ferrite bead, between each of these input signals and gnd, forming an rf filter.

Another point is that you should have a differential receiver in order to exploit your twisted pair cable. There is no differential noise due to the twisted pair and the common mode noise is rejected due to the differential receiver. Hence you, finally, get a very "clean" input signal. Otherwise, by using just a "single ended" receiver, your twisted pair cable is essentially useless.

what noise? visable, audiable or other interference signal?

how high i s the voltage? 36v, 56v, 220v or 100kv?...

You also need to make sure you have balanced driver and input circuits. It's no use using twisted pairs if audio devices are unbalanced.

Balanced audio usually uses XML connectors (round three pin), or 6.5 mm phone plugs with Tip Ring Sleeve (TRS). If you have RCA connectors you are un-balanced (at least you circuit is :-).

Peter.

Have you tried using a signal isolator?

They act like a DC transformer allowing the input & output to float, stopping any ground loops etc. They're fairly cheap.

A model I use a lot is ADAM 3016. it takes 0to5V, 0to10V,-10to10V, 0to20mA etc In and transmits 0to10V, 4to20mA etc Out.

Yes, this is probably the easiest route. It should succeed no matter where your noise is coming from. Just be sure to shield and ground the A/D converter.

Sounds like holding the paintbrush and moving the barn.

No, it's really about the same in cost and effort as is figuring out where the noise is coming from and then designing around it. Especially in noisy environments it's much easier to convert the low voltage analog signals to digital and just avoid the noise.

If there are RF issues, you WILL see them in your A-D converter unless you shield and filter that too. I agree with the steps outlined in the first couple of posts, and those should be done first before going the digital route. I've run dozens of signals (sonar, audio, RF communications, video, analog temperature, humidity, linear displacement sensors, etc.) in a confined environment co-existing with high power, high frequency (ultra-sonic) PWM motor controllers. Proper shielding and ground termination worked in virtually all cases. Only in three cables did I have to make the "snake swallow some ferrite toroids". Several passes of twisted pair through the right ferrite toroid can work wonders on CM noise.

Good luck with your quest.

Some great ideas already posted.

Have you considered using differential inputs over coax with the shields grounded. Any "interference" will enter both signal cables and will be self canceling at the receiving end as the two signals are added together again, just giving the original signal back.

For example some SCSI cables are made in this way (not all!), but the sending and receiving electronics are also necessary to make it work correctly.

If you are interested, I could do a search for you on the internet for some circuit arrangements if you are unable to find anything yourself....

You did not give us enough information about your signals to let me give examples. remember "GIGO"!

Assuming that you have differential drivers and receivers: you might consider isolating the signal with a transformer at the receive end:-

Greetings.

When you run your cables near power try to run them at 90 degree angles.

When using shielded cable only ground at the end near the source and insulate at the far end.

Check your source and make sure that your problem is not getting in at the source.

Shield the source if possible.

The simplest solution would be to go to your local electrical supply shop and buy some flexible metallic conduit which you can have cut to length. Run your cable through the "flex," then run the flex along side the wireway. This will shield your comm cable from electro-magnetic noise from the high voltage cables.

Could you use a curent signal transmission in stead of a voltage? The disturbancies will be a lot less important.

I have to assume a couple of things. First of all, I will assume, that your analog signal does not have a DC component. Then you can drive a transformer, that drives a (standard 600 Ohms) twinline within a coax type shield. At the receiving end another transformer brings it back to the single ended configuration. The beauty of the configuration is that the ground on both ends do not have to be the same. This is a standard way to cable up sound systems without annoying noise and hum from different power sources and grounds. The other one is using fibre (usually plastic) + LED for the same purpose = to break power and ground loops.

Use coax

Coax is extremely bad advice unless the spectrum of the analog signal is all above 10 kHz. The typical analog signal has a spectrum down to dc - that is, it has a non-zero dc component. At dc and clear through the mains frequency (50 or 60 Hz depending on location) and its harmonics, the coax shield does not act as a shield - it does not separate the signal riding on the inside of the shield from the noise on the outside. If the coax is grounded at both ends, which it must be in order to allow the circuit to function, 50/60 Hz and harmonic currents will flow due to ground potentials being different at both ends. The potential drop along the shield due to these ground loop currents will sum in series with the desired signal and add noise. It is for this very reason you use a shielded twisted pair for this type of signal, which is what the original post stated was being used.

Many of the previous answers were attempts to help solve the noise problem, but using coax would make it much worse.

RG6 dual core with ground spiral twisted around the envelope. Works great for clean power or audio.