DC fed long length wire interference

Since we are down around signal levels rather than power levels I feel competent to pipe in here.

Shielded wire should help with what you have observed.

Don't have an opinion on the op-amps.

Twisted is really designed to prevent coupling between the leads, and reject outside; AC and RF signals, so if you see AM or AC noise maybe - but it wouldn't help with what you describe.

The next step if shielded doesn't satisfy is to go to a different transmission method. Keep the wires and drive them with an encoder of some sort. Something with zero crossing signaling is good, something with a zero crossing offset signal is better.

There are any number of cheap differential signaling devices around.

Ceramic caps to case ground are what you are looking for. Assuming it is rf causing the problem, 0.1 uF is all you need. Minimum lead lengths to avoid added series inductance and possible tanking with capacitance. 10 nH per cm is a good rule of thumb on lead length added inductance.

emc_c

Twisted pair will help.
An alternative is to buffer the signal at the source end of the wire, so that the signal becomes a low impedance signal which is more noise immune.
You said #100mA if it's just going into an op amp it will be more like micro amps, maybe just a simple 470ohm resistor across the wires at the op amp end will calm it down, it depends on the surrounding circuitry.
Or just stop touching it

What is generating this 3-6v signal?
Del

The power is sourced from an ATX power box, using the 5 volt rail with a series resistor to reduce current so I don't heat up the thermistors, which is the source of the "signal" sent to the Op-Amp.

I have opted to avoid touching the wires () and to shield the (unavoidable) long length of leads with a grounded wire. Will introduce a ceramic capacitor close to the Op-Amp if needed as suggested earlier, which won't hurt...but will increase the overall parts count.

WTF is ATX?

OOPsy....was abbreviation for computer power supply...my bad (link)

...I assume too much sometimes

Differential drivers over twisted pairs for each signal are required.

In another blog I just updated it with such infos and links which you should find here. I hope this helps.

Let me know if the link did not help for any reason.....here is the text just cut and pasted:-

--------------------------------------------------------------------

Firstly, I don't know what the regulations are where you live with regard to laying signal and power cables together, but do follow that as a first.

I personally do not see a problem with putting them together provided that they are well insulated from each other AND all the signals are each sent as differential twisted pairs (two wires per signal) over the cable. There is a need for some extra electronics at each end.

You may find that something like the DS96176, RS-485/RS-422 Differential Bus Transceiver, which you will find here, may supply a quick and effective method of implementing such electronics....

...but only if regulations allow of course......

Using differential drivers is always a good way of reducing the effects of interference on signals.....here is a reasonable explanation of how they work, I hope it helps you further.

-------------------------------------------------------------------------------

Thanks for that. Most helpful indeed. The links all worked.

Will try solution fixes in increments, starting with an increase in voltage to ~15vdc feeding off from a PWM that is currently in circuit acting as a push-pull, with voltage sourced from the 12 volt voltage link from the ATX case. The feed from the transformer is filtered to clean DC, so no problem with that or controlling current.

Second step is shielded cables, with the grounded shield return path being seperate to other ground paths that have different voltages to run respective IC's. Each cable is linked to an Op-Amp with numerous Op-Amps sharing the circuit on the one PCB, so a low value ceramic capacitor won't hurt for each cable.

Third option if the second fails is to use differential drivers.

I am in the final stage of completing my bad-ass NiCad charger from recycled parts and hope to publish shortly the results, but need to fix up some loose ends...

All the previous suggestions are good.

For what it's worth, here is what I would do:

From your signal source start with a differential driver, use twisted shielded cable, terminate at the recieving end with an appropriate resistor & a low ESR (effective series resistance ) Cap. I would also use an IAMP (instrumentation AMP) as they have very high CMR (comon mode rejection). Analog Devices http://www.analog.com/en/index.html has a great booklet on the proper use of IAMPs In it there are many excelent suggestions on how to minimize noise and protect your circuits from other potential problems.

A couple of additional things:

The I AMP will help avoid "Ground Loops" , a major noise source. You can establish a new reference GND at the reciever with the IAMP. I would not count on the shield as it might have current flow, a noise source, running thru it.

Also, I AMPs have matched high input impedance where as the OP AMP circuit does not.

50cm wires. That's not that long. I didn't think there would be too much of a problem.

neither did I...touching the wires gave me cause for concern when weird things began to happen, making think of safeguards just in case...must be my electrifying personality

You have built what could be called, the simplest transformer possible.......that is the reason for your problems.......

the simplest transformer possible.

Wasn't that built by Two-loops Lautrec.?
Del

LOL!!!