0
\$\begingroup\$

One of my previous question revealed that a variable voltage test device's power supply which is an AC/DC adapter, and is SMPS type which causes common-mode voltages. A solution given by tying the supply ground to earth ground. This works only if the AIGND of the data-acquisition device is earthed.

Anyway below is about my question:

Now I found another workaround which I cannot explain the reasons. To begin with the issue was due to the parasitic leakage of the SMPS adapter which causes common-mode voltages to appear all the way at the device terminals wrt earth. I show this in Figure 1 below:

enter image description here

Now if I tie the ground of the device to the AIGND of the differential input data-acquisition channel as in Figure 2 below(green line), I still have 50Hz noise:

enter image description here

But if I tie the ground of the power adapter to the AIGND of the differential input data-acquisition channel as in Figure 3 above(green line), 50Hz common-mode noise completely disappears.

Even more strangely is that: when I make a continuity test between the adapter and device ground(GNDA and GNDB in Figure), they are already connected to each other.

So theoretically tying from GNDA or GNDB to AIGND shouldn't be different.

My questions are:

1. Why does the common-mode voltages disappear when I connect the adapter ground to AIGND?

2. The continuity test tells adapter and device ground already connected. Bu then how would you explain the difference between the results of Figure 2 and Figure 3?

3. Can I safely use this method in this type of situations? I mean tying the power supply GND to AIGND.

enter image description here

I mean I have a setup when the switch above is open. And when the switch is closed the common mode noise-disappears even though there is continuity between the GNDA and GNDB. Can this technique be used safely?

\$\endgroup\$
0
\$\begingroup\$

The continuity test tells adapter and device ground already connected

Doing a continuity test does not reveal the true impedance at 50 Hz (or higher) between GNDA and GNDB. Who can say whether the output terminals of the device contains (internally) a common mode choke that is being disrupted by a differential current when you grounded GNDB. There could be resistance that is still significant that your tester regards as a "pass" for continuity.

Why does the common-mode voltages disappear when I connect the adapter ground to AIGND?

That's the sensible place to ground the system - then you have no AC currents passing through the "device".

Can I safely use this method in this type of situations? I mean tying the power supply GND to AIGND

Prevention is better than cure and doing this prevents those AC currents entering the "device". This would be the sensible approach.

\$\endgroup\$
  • \$\begingroup\$ Oh I was still editing the question. Please see the third question again. \$\endgroup\$ – HelpMee Jan 26 '18 at 13:49
  • \$\begingroup\$ I see Rbias and this can be needed when the "device" has a floating output and hence needs a light galvanic connection to AIGND but, for getting rid of the AC interference, you should use a direct connection to GND (It's not clear if AIGND is "good enough" as that GND connection but I'm presuming it is) \$\endgroup\$ – Andy aka Jan 26 '18 at 13:52
  • \$\begingroup\$ If AIGND is not heavily strapped to real GND, connect GNDA to real earthy ground independently. \$\endgroup\$ – Andy aka Jan 26 '18 at 13:54
  • \$\begingroup\$ Even Rbias is zero(direct connection) like in Figure 2 there is 50Hz and its harmonics as noise on the signal. The only solution is Figure 3. Thats what I observed. Or Figure 4 when switch is closed. \$\endgroup\$ – HelpMee Jan 26 '18 at 13:56
  • \$\begingroup\$ AIGND manufacturer says should not be connected to earth, he says our ground is very clean so dont use earth. \$\endgroup\$ – HelpMee Jan 26 '18 at 13:56

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.