This is about for low level signals interference/noise being observed with single ended data acquisition inputs but not with diff inputs. Basically there are three data acquisition boards, where one of them is a single-ended USB board, the other one is a single-ended earth grounded PCI board(can be set for diff ended as well) and the third one is diff-ended USB board. All are monitored by a desktop PC. Sampling rate is set to 1kHz for all. And all boards have the same resolution.

When I couple a source to inputs of these boards normally if there is no interference and noise I don't get issues for DC signals like 5V DC.

But sometimes I observe noise issues with single ended PCI board so I wanted to investigate about these empirically.But for this I needed a source with interference. Luckily I guess I found one. A function generator signal is read differently by the boards depending on they are single or diff ended.

I use this function generator as source for input signals to the data-acquisition boards. I set it to 10mV pk-pk 3Hz sinusoidal with no offset. So a good measurement should measure a similar signal which should resemble a 10mV pk-pk 3Hz sinusoidal. I use both STP and BNC coaxial cables. In this example I will use BNC coaxial cables.

First I use the function generator with 3-prong power cord. The function generator's output signal ground is connected to the mains earth through the power cord just like in a scope.

So first I use it with 3-prong power cord. Then I measure with the PCI single-ended earth grounded board and here is what it measures:

enter image description here

Above you can see the offset error and distortion in the signal.

And now I set up the same measurement with the single-ended USB board. Notice its ground is not earthed. And here I get the same problem:

enter image description here

There is offset error with significant distortion.

I noticed this distortion gets bigger with the cable length.

And finally here is the result from the diff-ended board's measurement:

enter image description here

Above there is almost no offset error and the noise does not distort the signal much. Looks like random noise. I also get similar measurement if I use the PCI board as diff ended input config.

And after all these I concluded diff-ended inputs are more immune to noise.

But then I used two-prong power cord for the function generator and the single-ended boards started to measure very accurate even a bit better than the diff-ended.

My questions are:

1-) I use cables around 15 meters up tp 50 meters. And seems to me like if there is no interference single ended is great. But when there is noise the dif-ended is better. In my case what could be the reason that using three-prong power cord for the function generator causes tis interference?

2-) How does this noise appears as offset and distortion in single-ended and does not appear at the diff.ended daq board?

3-) I want to show it to people to convince to use diff.ended board for better measurements. I was looking for long time a noisy interference scenario to test two systems single and diff ended but I failed to create interference. Can my setup be used to demonstrate the difference of these two systems(single vs diff ended)?


As I mentioned if I use 2-prong power cord(i.e no interference), I obtain the single-ended(blue) data even better than the diff-ended(red) case as below:

enter image description here

  • \$\begingroup\$ Comments are not for extended discussion; this conversation has been moved to chat. \$\endgroup\$
    – Dave Tweed
    Commented Mar 18, 2018 at 14:14

1 Answer 1


There is a power supply inside the function generator. With a power transformer to isolate (or float) the Function Generator output signal (your 10mv at 3Hz). When you use the 3-wire cord, you are grounding lots of pieces of metal, and the long cables likely are picking up 60Hz hum that does to ground, thus making the 10mV ride atop 60Hz power line trash.

Your very slow sample-rate (what is your sample rate) converts the 60Hz trash into apparently random spikes. The consistent (-) going trash polarity may come from the polarity of rectifier charge surges.

Friend of mine dubs vinyl records onto harddrive; the turntable has a 5th wire, which is a non-insulated copper braid tied to the metal tonearm and metal baseplate under the motor. The 2 cartridge channels --- left and right --- edit the tonearm in twisted pairs (not shielded) that run over and very close to the metal baseplate, before becoming coax-cables out of the turntable to the RIAA preamp. If this 5th wire is dis-connected, the RIAA preamp outputs milliVolts of 60Hz. If the 5th wire is tied to the RIAA preamp "GND", hum becomes microvolts.

You have the challenge of learning to handle electric fields and perhaps magnetic fields, in a precision signal measurement system.

One way to launch into understand is by drawing some sketches of what connects to where, including the cable distances/shield_resistance/shield_inductance, and then inject trash of various frequencies that need to find a RETURN path. All possible paths will be explored, proportional to conductance.

  • \$\begingroup\$ Sampling rate for the daq boards are set to 1kHz, isn't that enough to reveal 60/50Hz? Another thing is that why am I not seeing that kind of distortion when I use the diff-ended input data acquisition? \$\endgroup\$
    – floppy380
    Commented Mar 18, 2018 at 11:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.