0
\$\begingroup\$

A single ended earth grounded data-acquisition board has several inputs. Since it is single ended all input channels share the same ground when hooked up.

The thing is normally if there is no ground loop in the entire system and if I lets say plug in a new channel the other(already connected channels) channels' noise level and standard deviation is not affected.

But recently if I plug in a force transducer amplifier as an input to the same data-acquisition board the other channels' standard deviation increases for example from 0.005 to 0.032 ect. And when I observe what happens in FFT in real time is that when I plug in any of the force transducer amplifier as an input, other already connected input signals' white noise increases. There is no 50Hz related noise.

I can understand how 50Hz noise or harmonics appear in other inputs if there is ground loop. But in this case I couldn't find a ground loop and the noise type occurs in other channels seems like white noise.

How can this be explained?

Edit related to an answer:

Single-ended daq

enter image description here

Diff-ended daq

enter image description here

\$\endgroup\$
  • \$\begingroup\$ What else is common mode? or shared? Vref? Power? Gnd? Are they all force transducers/ \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Feb 14 '18 at 18:20
  • \$\begingroup\$ There is ghost voltage coming through power supplies around 80VAC as a common mode. Capacitive coupling. But I dont think that is the case because I have other instrument with same ghost voltage issues they dont affect the noise level in other channels. \$\endgroup\$ – floppy380 Feb 14 '18 at 18:22
  • \$\begingroup\$ Can you add a LPF to force transducer output like 27k, 0.22 uF? Or add 3nF from 0V to earth gnd? \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Feb 14 '18 at 18:25
  • \$\begingroup\$ Yes I can try tomorrow when I have access. I should draw the sytem and update here I guess also. \$\endgroup\$ – floppy380 Feb 14 '18 at 18:27
  • \$\begingroup\$ CMRR drops with rising f in IA's \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Feb 14 '18 at 18:28
1
\$\begingroup\$

Consider a standard 33 rpm turntable, with 2 output coax cables and that 5th wire connected to the metal base and the tonearm.

Why is this 5th wire needed? to provide a path for interference coupled across the power transformer used in DC supply for the turntable rotation motor.

How much current to handle?

Assume the power transformer has 3.16 cm by 3.16 cm common area between primary and secondary, with 1mm insulation. Assume relative permittitivity is 5. The capacitance is

C = Eo * Er * Area/Distance

C ~~ 50 pF/meter * 0.0316meter * 0.0316 meter / 0.001

C= 50pF/meter * 0.001 meter^2 / 0.001 meter = 50pF

Assume the power transformer is badly imbalanced, so we simply model the displacement current (the capacitive charge/discharge current) as

I = C * dV/dT = 50pF * d(160 voltPeak * sin( 2 * pi * 60Hz))

I = 50pF * 160 * 377 radian/second ~~ 50pF * 60,000 ~~ 3 microAmpers at 60Hz.

Your situation is probably similar. You have 3 microAmps exploring all possible paths to get back home. Draw some sketches of these paths.

============================================

Here is +10dB SNR white-noise, riding atop +1/-1 volt squarewave

enter image description here

\$\endgroup\$
  • \$\begingroup\$ Could you please check my edit? I made these drawings to be sure whether I got your answer. In my single ended daq drawing, I tried to draw what you tell. Is that correct? CM currents from power supply transformer causes potential swings in other channels due to Rgs(resistance of the wire path to AIGND) being common to all channels. And in the second drawing I tried to show that due to MUX the CM coming through transformers does not affect other channels since in this case the channels do not simultaneously share the same ground. Was that what you mean?? \$\endgroup\$ – floppy380 Feb 15 '18 at 17:07
  • 1
    \$\begingroup\$ The thing is the noise I see as increase is white noise not 50Hz. \$\endgroup\$ – floppy380 Feb 15 '18 at 17:09
  • \$\begingroup\$ How do you know it is white-noise? Spectrum Analyzer? \$\endgroup\$ – analogsystemsrf Feb 16 '18 at 2:44
  • \$\begingroup\$ Your edits look correct. 50Hz and 311 volts peakpeak \$\endgroup\$ – analogsystemsrf Feb 16 '18 at 2:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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