noise problem with differential analogue input

I wonder why you are using coax for a differential signal, this wont help, twisted pair should be better.What is the required bandwidth?

What CM rejection are you getting?

martin

What exactly do you mean, "I have a noise problem..." What sort of noise? Can you put an image of a `scope trace on a web page, illustrating the noise? If you short the two inputs together and bias them to an appropriate common-mode voltage (perhaps ground), do you still have noise? What sort of diodes are those, and why are they there? (Hint: if they are standard signal diodes as the symbol suggests, they won't do much to clamp the signal within the common- mode range of the amplifiers.) Have you done anything to insure the input signals stay within the common mode range of the amplifiers? (Do the input signals share a common "ground" with the amplifier circuit, and are they nominally near 0V?) One common mistake in applying this sort of amplifier is having an input signal with no ground reference, and the amplifier inputs "float" to a DC level that's inappropriate, outside the common mode range of the amps. A resistor from each input to ground can often cure that problem. The resistors in the schematic do not do that.

Like Martin wrote, it would be more usual to feed signals through a shielded twisted-pair. Since the "gain" of the amplifier is about

-12dB, I'd be surprised if you really even needed the shielding, if you're not in an electrically/magnetically noisy environment.

50 reads/sec (averaging about 194 ADC samples[10bits] each 20 milliseconds for each read)

Whats the ambient electrical noise like?

Ok, just had another look at the circuits, the diodes seem to be reversed biased al the time, so they dont actually seem to do anything.

i think you need to dump the coax, and try some screed cable, proper screened microphone cable would do for a test, or even CAT 5.

A quick tweak would be to drop the value of R76 a bit, and put a small pot in series, with 1% R's your CMRR will only be in the 40dB range, the pot will get you into the 70dB or more range, but this wont solve having he wrong cable type

martin

text -

Thanks for your post and Martin ones,

I try to put an image of a scope trace, tomorrow. All signals (6) are DC voltages between 0 to 10 volts. In fact, they are from resistors of voltage dividers that show voltage of some High Voltage capacitors. Ground of all capacitors (6) have connected to same point (earth and my boards chassis) that connect to G inputs. 5 capacitors (signals) have posetive charges and 1 capacitor (signal) has negative charge. If I short the two inputs together, my system shows (true) Zero voltage. Diodes are 13 Volts/1Watt Zener for opamp protection from possible high voltage inpouts.