Wierd Motor (magnetic) noise

Do the sensors/daq allow them to be used differentially? Perhaps switch to CAT5 twisted pair?

What frequency is the noise? 60 Hz harmonics?

The noise is around 200 kHz and seems fairly band-limited probaly 185 to 215 or so. Now, I could put a low pass filter as our signal frequency is close to 1 MHz, but what if in another location, the noise source frequency is different

No, the DAQ is unfortunately single-ended. The sensors can be kept differential until they get to the DAQ device. I tried to use a diff. amp (AD8137), assuming the the CM noise on the conductor and the double-braid shield, which is also ground would be the same, but it is not so. And my differential amplifier keeps ringing as well.

Have you checked for ground loops ?

Well there is an isolation transformer on the DAQ system input, so that should take care of loops, there is also a transformer (signal) that is going to be added in series to take out DC noise

So you have loops in your sensor common wires, ideal for picking up magnetic noise. Do you also have loops in your shields? Are you using the shield as a signal common?

Seems he said he has ground loops, indicating that he has not read "Grounding and Shielding" by Ralph Morrison, which is the grounding and shielding 101 class you should have had in school but probably didn't.

Interesting.

So if your signal is around 1MHz can we assume the sensor is active (not a strain gauge or RTD etc.)?

I'm wondering if the sensor output drive is having trouble driving a capacitive load and your getting some amplifier oscillation in the sensor? That would explain the relatively narrow frequency range of the oscillation and the fact that it gets worse as the cable gets longer. A possible test would be to add capacitance to your short cable to mimic the long cable and see if you get the noise (oscillation). Is the sensor output high or low impedance?

The noise is around 200 kHz and seems fairly band-limited probaly 185 to 215 or so. Now, I could put a low pass filter as our signal frequency is close to 1 MHz, but what if in another location, the noise source frequency is different

Very often the obvious, isn't !

Yes, and Taking out the loop at the sensor end makes the problem worse.. And the shield is hte signal common. Now, I am thinking of a triax cable with an extra shield whcih I can connect to system ground- and then take out the loop at the sensor end. At the DAQ end they are made single-ended by the DAQ input input.

It's quick to clamp on a common-mode choke, those ferrite two-half- bead/plastic shell things, if you just want to reduce the problem. Once the common-mode is within input compliance, your instrument amplifier will reject the rest of it. We hope.

I've also hidden common mode filter chokes inside preamplifier boxes, and with hookup wire you can probably fit several turns in a ferrite or iron core, which will give some performance improvement.

If the noise/pickup is truly a differential signal, it will respond better to use of twisted pair wiring. At worst, you can always mount the preamplifier at the sensor and drive the amplified signal on twisted pair wiring; that almost always fixes pickup problems.

use shield for both common and high leg of the device.

the center conductors is what you want to use. ground the shields only at one end . your problem should go away. this is why they have equipment grounds and chassis grounds..

Or, get your self some 2 led signal wire with shield and only ground at one end.

in that case the input is balanced... try a balanced cable.

Bye. Jasen

When you use the shield as signal common it is no longer an effective shield, just an outer conductor. You could try shielded twisted pair for better performance, grounding the shield at one end only, with the pair connected only to the sensor and not to any ground or the shield at the sensor end. (If the shield is allowed to carry current it will couple to the internal conductors.)

For really good noise immunity you will need to switch to a DAC with differential inputs; you can never completely eliminate unintended parisitic circuits when you run low level signal cable in noisy environments, but you can make the noise pickup mostly common mode fairly easily, and reject most of it with the differential input. If you don't want to change your DAC then you can use a separate diff amp or in amp in front of it. For best results read the book and do the analysis yourself :-).