0

A single-ended source is connected to an inAmp of a data-acquisition board as follows:

enter image description here

As you see above, the shield is shielding the twisted shielded pair cable. The shield is tied only to the earth ground. AIGND is not connected to earth ground.

I want to make a very simplistic simulation in LTspice, to mimic this scenario when there is a 50Hz common mode noise comes through the power supply to both lines as common mode interference.

Below is my attempt:

enter image description here

There is something missing or wrong in above model.

I want to model the circuit where the common-mode interference comes through the SMPS power supply. Power supply need not to be modeled juts the noise like in my circuit. But there is also shield and the AIGND is not earth grounded.

How can I fix this to make it more realistic?

My second attempt:

enter image description here

edit:

enter image description here

GNZ
  • 1,729
  • 5
  • 28
  • 59
  • I've read this three or four times but I'm not clear about what you want. Where is the SMPS supposed to be situated? – Andy aka Jan 25 '18 at 17:23
  • SMPS not needed just a signal source which does the same effect of it. Suu Vs2 in the LTspice circuit. Imagine that is a transducer. And imagine the SMPS power supply of this transducer is leaking 50Hz common mode voltage all the way to the inamp lines. That is what I want to model. It is to convince some people the effect of it. – GNZ Jan 25 '18 at 17:27
  • I tried and added another attempt. I dont know whether is better. – GNZ Jan 25 '18 at 17:29
  • If the common mode voltage is high the inAmp outputs very weird. And 10k creates imbalance and makes it impossible to zero the 50 Hz noise. – GNZ Jan 25 '18 at 17:32
  • OK, your first attempt is OK if the SMPS is in the sensor but you will find that the series capacitance is more like 1 nF with very little resistance. At 50 Hz and 230 volts AC this can inject a current up to 70 ish uA. – Andy aka Jan 25 '18 at 17:32
  • If you are using output resistors of 100 ohms then the grounding resistor of 10k will create a small error. To solve excess common-mode use capacitors from each line (at the receiver) to ground - try 10 nF to start with BUT beware - these will affect the frequency response of your sensor. At DC it will be fine but at 160 kHz you will start to get roll-off. Now this almost certainly won't be a problem but, if you need to use 100 nF capacitors the roll-off starts at 16 kHz and might begin to affect performance. – Andy aka Jan 25 '18 at 17:37
  • Did whatcha say didnt work mate: https://i.stack.imgur.com/k9YvA.png Output jerks – GNZ Jan 25 '18 at 17:48
  • I didn't say 10 nF for C_leakage did I!!! Anyway, what were the effects before and after. "Didn't work" means nothing to me... quantify! – Andy aka Jan 25 '18 at 17:50
  • Apply it to your first circuit not your 2nd circuit. – Andy aka Jan 25 '18 at 17:51

1 Answers1

0

enter image description here

The two capacitors near the InAmp remove common-mode AC content caused by the 1 nF in the SMPS modelled to the left. Some SMPSs will have significantly less than 1 nF and may be tamed by 10 nF decouplers but other SMPSs might have 2.2 nF internally and will need greater help hence 100 nF or greater may be needed.

Bear also in mind that if 1 uF capacitors are needed thay will reduce the output bandwidth of your source because of the 100 ohm balanced drive.

Typically, 1 uF and 100 ohm produces a cut-off frequency of 1600 Hz. However, if you are only worried about slow speed signals from your source 10 uF capacitors can be employed.

One thing to watch though - those 10 uF capacitors will produce a differential signal if they are not perfectly matched.

Andy aka
  • 456,226
  • 28
  • 367
  • 807
  • I think in this circuit on the right side Earth and the AIGND should be wired. Earth symbol is hand drawn lines. But I did that connection now and doesnt work man. Output jerky. I can tell you 1nF leakage is so small to see that common mode range effect. Make it 100nF and common mode voltage becomes around 40V and those parallel caps doesnt rescue. Try simulate maybe you can figure out why. Im not familiar as much as you. – GNZ Jan 25 '18 at 18:32
  • Alright 10u worked. Before I had tried upto 100nF only. – GNZ Jan 25 '18 at 18:33
  • You said in your question that "AIGND is not connected to earth ground" so I went with this idea. Are you changing your mind? – Andy aka Jan 25 '18 at 18:43
  • It is not connected in real, but for the sim to work it is connected through that 1nF capacitor on the left side. If you dont connect that on the roght there will be no loop you cannot capacitively couple the common mode voltage in the sim. – GNZ Jan 25 '18 at 18:44
  • Vinterefence is tied to earth but that is not ground in the sim, that earth symbol is just lines I drew – GNZ Jan 25 '18 at 18:46
  • Just noticed this answer would be glad to hear from you about this new question: https://electronics.stackexchange.com/questions/377365/analog-signal-transmission-when-the-inputs-to-differential-amplifier-are-truly-i – GNZ May 31 '18 at 00:00