3

I really don't expect an analytic answer. I'm sure there are far too many variables to consider.

I'm working with a fairly simple circuit (Precision voltage, Instrument amplifier) but need to really an output signal that has have noise a low as possible. I suspect I have a ground loop problem but I'm not sure. Here's the background.

I have a multi-channel 24 bit DAC (data acquisition system_) -- differential or single ended. My approach to measuring the 'ground'/'current loop' voltage is to connect the 'ground'/ common return lead to the to both inputs for a differential input while the circuit is powered and collect the data over an extended time (an hour or so). Data is collected once per second so I have lots of data. I expect there to be noise on the 'ground' wire but I don't know what is reasonable. The data shows that there is a mean voltage of 450 microvolts with a max excursions about the mean of +/- 15 microvolts. A 25 sample running average of the data gives the same 450 microvolts with an average deviation for the data (an 3600 + points) of 7 microvolts.

These values are the maximum resolution advertised by the DAC equipment. The basic question is "Is this a Quiet ground or is more expensive equipment required"???

If this is not a quiet ground, can you point to design documents (circuit/PCB) to further reduce the noise.

gandolf
  • 31
  • 1
  • 2
  • 2
    "I'm working with a fairly simple circuit but need to really an output signal that has have noise a low as possible." --- Need to really what the circuit? This would also be a good time to determine your requirements better than "noise as low as possible". How much noise can you actually tolerate at the output? "As low as possible" implies you should be looking into cryogenic circuits and other extreme measures. – The Photon Jul 06 '13 at 05:36
  • I'm also having trouble understanding where you say, "My approach to measuring the 'ground'/'current loop' voltage is to connect the 'ground'/ common return lead to the to both inputs for a differential input." Could you make this more clear? Did you just connect the ground node of your in-amp circuit to both inputs of a differential DAQ system? A schematic or photo of your setup might help. – The Photon Jul 06 '13 at 05:38
  • Is your question "What is a quiet ground?" or is it "How do I measure ground noise?" I can't tell. – Phil Frost Jul 06 '13 at 12:44

1 Answers1

2

I'm not sure I understand how you set up your ground noise measurement. To me it sounds like you have this:

schematic

simulate this circuit – Schematic created using CircuitLab

If this is what you're doing, you are not measuring the noise on the ground node of the in-amp circuit; more likely you're measuring emi pick-up from the two leads connecting the circuit under test to the DAQ.

What you probably want to do (although it won't distinguish ground noise from other noise in your system) is connect the negative input of the DAQ to the in-amp circuit's ground, and the positive input to either the input or output of the in-amp circuit. This will give you an idea what kind of noise is present in the circuit. But be aware that if the input bandwidth of the DAQ system is not the same as the bandwidth of the in-amp circuit, you still won't be seeing exactly the same noise as will be present in the final system you're designing.

Is this a Quiet ground?

A "quiet ground" basically means there shouldn't be any large currents flowing through the ground, particularly between the amplifier and whatever sensor/device it is amplifying. For example, switching currents from switching power regulators (and probably you should prefer a linear regulator for the in-amp's supply), or ground currents of any microcontroller or large digital circuits shouldn't be routed near the in-amp circuit.

The Photon
  • 129,671
  • 3
  • 164
  • 309
  • Your drawing in correct. The approach may be flawed and I'm afraid your right about the EMI pickup. But perhaps it is useful as an indicator for 'background' EMI noise for other measurements in the circuit. – gandolf Jul 07 '13 at 16:09
  • Sorry. Wrong key and cut off he reply. Basically I need to identify a 100 sec fluctuation at a .7 millivolt level that I 'suspect' is being introduced into the ground plane via a current loop. Just don't have a good idea of where to start. – gandolf Jul 07 '13 at 16:15
  • @gandolf If you have some "prime suspects" for the source of the noise, try turning each one on and off, change the frequency, change the duty cycle, or whatever, and confirm that you have located the source. Then figure out how that source is coupled to your sensitive circuit. Then figure out how to break the coupling. If you suspect a ground loop is involved, try breaking the ground loop and see if the noise is affected. – The Photon Jul 07 '13 at 20:20