DGND

[Buechsenschuetz]

I'll try to answer your last messages alltogether.

First, you are always asking for direct answer. I already mentioned that in some cases it is NOT possible to say 'yes/no' or 'correct/not correct', since even setups which seem simple may turn out complex. As mentioned, evaluating a proper and suitable grounding scheme for a certain application sometimes includes analysis of the power supply(supplies) and even board layout. I cannot do that from here.

All I can say is that in most cases you will see a (more or less minor) voltage between the DGND terminals of a pc i/o card and the pc earth/chassis. I would strongly recommend to use the DGND terminal as reference point, NOT the pc chassis/earth.

Also, I already mentioned that the BNC cable shield will introduce a second GND connection between the test fixture and GND of your acquisition system. This should be avoided if possible but if your system works well that way and if you do not have to make measurements in the sub-mV-range maybe it will work with acceptable performance.

I already mentioned that you probably will measure different voltages between DGND and different points of the chassis. The opposite is true in many cases, too: you will measure different voltages between pc chassis/earth and different GND terminals on different boards. That's the reason why it is recommend to use only ONE reference point for measuring. If this is not possible (as it is probably in your case), and if high accuracy is needed, you have to 'split' the DGND line and use opto isolators for digital signal transfer. But, as I mentioned several times, it depends on your application and demand of accuracy whether this is necessary or not.

I do not know your analoque acquisition board in detail. On many analogue boards you can tie the A- input directly to AGND if you have a single-ended analogue signal (as in your case). Some manufacturers recommend not to use a direct connection but via a resistor which should be equivalent to the DC input impedance of the analogue inputs to avoid any errors due to input bias currents of the analogue input. I have seen a comprehensive explanation on the NI website somewhere, I think it was on the same page where they discuss grounding schemes for analogue signals.

Sorry I cannot help you with all the details of your setup, but this is just impossible from the distance and without having a look at all the details. Even the most detailed description will not contain all the information needed. All I can give you are general guidelines based on some experience, and I do not have experience with all test setups and data acquisition boards.

[nyborn]

Hi Buechsens:

First of all, I just start working in the society so I don't know that much. Don't worry, I will keep it in mind for some questions that can not answers Yes/No.

In your previous mail, I would recommend to use the DGND terminal as reference point, NOT the pc chasis/earth because in most cases I will see a (more or less minor) voltage between the DGND terminals of a pc i/o card and the pc earth/chassis.

So, I measure a voltage difference between negative terminal of the power supply in the fixture and DGND using DMM. I got 0V. I know I my setup in this part is correct right now. But I want to get understanding here.

You said the DGND and the computer earth have the same potential. Right now, I measured 0.9mV. Could it say the same potential? If you say "yes." Then I got no question to ask from computer chasis to negative terminal of the power supply. But if you say "no". and the reason is voltage shift. then are there any problem between the ground of my test fixture and the computer earth. As you know, everything is connected together. Those 0.9mV will cause any problem. If no, why? Could you please explain?

You mentioned the BNC cable shild will introduce a seond GND connection between the test fixture and GND of your acquistion sytem. Why? Could you please explain?  And the sub-mV range that you mentioned. Sub means under. Sub-mV range mean under 0mV, is my understanding right?  What is the meaning of ground loop here? Because all I knw the definition of ground loop is the two different individually gound signal source. Could you please explain your perpective of ground loop here? You also mentioned if the measurement is in the acceptable range, that is OK, right? you don't need to care ground loop due to a subMV, why?

My measuremetn does not need high accuracy. so I don't need to use opto isolator Let say "no", that means as long as my measurment get the value I fullfil with my customer requirment, that is all right? I need to make sure with you. Would you please answer to me this question directly?

[Buechsenschuetz]

>In your previous mail, I would recommend to use the DGND terminal as reference point, NOT the pc chasis/earth because in most cases I will see a (more or less minor) voltage between the DGND terminals of a pc i/o card and the pc earth/chassis.

>So, I measure a voltage difference between negative terminal of the power supply in the fixture and DGND using DMM. I got 0V. I know I my setup in this part is correct right now. But I want to get understanding here.

I was referring to the pc earth terminal, NOT to the neg power supply terminal of the test fixture.

>You said the DGND and the computer earth have the same potential. Right now, I measured 0.9mV. Could it say the same potential? If you say "yes." Then I got no question to ask from computer chasis to negative terminal of the power supply. But if you say "no". and the reason is voltage shift. then are there any problem between the ground of my test fixture and the computer earth. As you know, everything is connected together.

Oooops - I should have said 'DGND and computer earth are THEORETICALLY at the same potential'. Unfortunately in the real world they are not. This is due to voltage drop in the wires from pc power supply to mainboard. Also, what you can measure is only DC voltage difference. There may be even heavy voltage spikes between DGND and computer earth. So in most cases I would NOT recommend to use the computer earth terminal at all. All circuitry of the dig i/o board is referenced to the DGND terminal of that board, and this terminal should be used as reference terminal for circuitry like your test fixture which is connected to the dig i/o board.

> Those 0.9mV will cause any problem. If no, why? Could you please explain?

0.9mV is a minor fraction of the TTL logic zero level which should be lower than 0.8V. So when using digital circuits only using computer earth probably will cause no problems. But I think there is no reason to do so if you can do it better very easily, as described above.

[Buechsenschuetz]

>You mentioned the BNC cable shild will introduce a seond GND connection between the test fixture and GND of your acquistion sytem. Why? Could you please explain? 

Usually the BNC cable shield is connected to AGND of the test fixture. I can see from your diagrams that AGND (i.e. GND terminal of analogue output) and DGND are identical in your test fixture since you have just one power supply with one neg terminal which is used as GND for digital and analog circuitry.

As I mentioned, on most data acquisition boards AGND is tied to pc power supply GND somewhere, and the same applies for DGND of digital i/o board.

So if you connect neg supply terminal of test fixture to DGND, and GND of analog output to AGND (via the cable shield) it is evident that you have TWO connections between neg supply terminal (= GND) of test fixture and pc. The GND loop now runs from DGND of dig i/o board to neg supply/GND of test fixture, over to GND of analogue output to AGND of analogue board and back to pc power supply GND which is connected to DGND of dig i/o board.

> And the sub-mV range that you mentioned. Sub means under. Sub-mV range mean under 0mV, is my understanding right? 

Sorry for not clarifying this, 'sub-mV' usually is used as a designator for order of magnitude. So 'sub-mV' means 'signals with less than 1mV amplitude'.

> What is the meaning of ground loop here? Because all I knw the definition of ground loop is the two different individually gound signal source. Could you please explain your perpective of ground loop here?

See above. A loop usually is a 'circle' of (wire) connections. I hope I pointed out that you get a 'circle' or 'loop' of GND connections with the set up described. Any magnetic field can generate noise in such a loop.

> You also mentioned if the measurement is in the acceptable range, that is OK, right? you don't need to care ground loop due to a subMV, why?

As I mentioned several times, you must have a close look at your setup, down to the power supply design and printed circuit design to check all this. I cannot do this from here.

>My measuremetn does not need high accuracy. so I don't need to use opto isolator Let say "no", that means as long as my measurment get the value I fullfil with my customer requirment, that is all right? I need to make sure with you. Would you please answer to me this question directly?

All I can say that if you get results with acceptable accuracy your set up is correct for this purpose. Maybe you will get trouble when higher accuracy is needed, but it is impossible to judge this from the distance.

Anyhow, never mind if you are a newcomer. All of us have been newcomers in a certain period of their professional lifes. Experience does not come over night (and not from a few answers in a forum), I hope that you will not be blamed if your first data acquisition does not work perfectly right from the start - any boss should know what he does when he/she hires a newcomer.

[nyborn]

Hi Buechsens:

I still have something that need you to verify and answer so that I can get answer from the previous mail.

In the previous mail, you said "I was referring to the pc earth terminal, NOT to the neg power supply terminal of the test fixture.'

Here you meant,  use the DGND terminal as reference point, NOT the pc chasis/earth to measure the voltage difference between DGND and the pc earth terminal. Am I correct?

I measure the voltage difference between the negative power supply and DGND. I got 0V. I think I got correct setup here for digital DGND part.

The blue text is what you said:

Oooops - I should have said 'DGND and computer earth are THEORETICALLY at the same potential'. Unfortunately in the real world they are not. This is due to voltage drop in the wires from pc power supply to mainboard. Also, what you can measure is only DC voltage difference. There may be even heavy voltage spikes between DGND and computer earth. So in most cases I would NOT recommend to use the computer earth terminal at all. All circuitry of the dig i/o board is referenced to the DGND terminal of that board, and this terminal should be used as reference terminal for circuitry like your test fixture which is connected to the dig i/o board.

[nyborn]

First I got the real world is not 0V. I will get 0.9mV like this vale. I got it. The second is that you said the voltage between pc earth and DGND is DC voltage.

Is that because this voltage here is talking about digital here, I need to verify here? I wonder the pc main power supply is AC, would it be DC after the current or voltage go into computer? I need to make sure with you. no matter is AGND or DGND. I don't understand if you took DGND as a reference point, why are you not afraid of the voltage drop (Shift voltage) and heavy voltage, is that because DGND is 0V by designer and pc earth is 0.9mV or 1mV depend on the area of chasis? Could you explain the pc earth voltage? I need to slove this problem here. I don't know if you can see it.

0.9mV is a minor fraction of the TTL logic zero level which should be lower than 0.8V. So when using digital circuits only using computer earth probably will cause no problems. But I think there is no reason to do so if you can do it better very easily, as described above

only use computer earth will cause no problem. That means if the setup put a wire from negative terminal of power supply to pc earth as a refernce point, not DGND? Would you please clarify me here?

Usually the BNC cable shield is connected to AGND of the test fixture. I can see from your diagrams that AGND (i.e. GND terminal of analogue output) and DGND are identical in your test fixture since you have just one power supply with one neg terminal which is used as GND for digital and analog circuitry.

[nyborn]

That means the AGND (GND terminal of analogue output) is the negative terminal of power supply in the fixture, am my understanding correct here?

If the answer is yes above, the ground is starting from the negative terminal of the power supply in test fixture to the pc earth. See attached graph. Is that correct?

I have 3.3mV for voltage throungh BNC shield cable This value is greater than 1mV. The ground problem is slove here, right?

Also, as you know i know the definition of ground loop which is two individual groud signal source. What is the difference between this ground loop and the other you mentioned.

You mentioned power supply designand printed circuit design to check. Are you talking about power supply design with earth or without earth, I need to be sure here and pricted circuit design, would you pls tell me what you are talking about here?

 I believe if you answer to me in this area, I can slove many problems. So, pls focus in my new message area

[Buechsenschuetz]

First, my recommendation still is to use the DGND of dig i/o board as reference point, NOT the pc earth. Although there is galvanic connection between pc earth and DGND, this connection is not well defined, and you nothing about the way the ground connection takes inside the pc. To avoid these problems you should always use DGND.

In case you are using shielded cables, the cable shield usually is connected to the connector shield which of course is connected to the pc chassis and pc earth when plugged in. I would strictly recommend NOT to connect the connector/cable shield to anything in your test fixture, you may create another ground loop here.

Maybe there are different definitions of 'ground loop' but the definition that I am familiar with is the one I described (you have got it correctly in your drawing, the GND connections make a closed circle or loop).

It is hard to say where the 3.3mV along the shield of your BNC cable come from. Remember that DMMs have very high input impedance and will pick up a lot of noise from the environment. Also, you never know exactly whether this is a DC voltage, or some unsymmetrical AC voltage along the shield. Of course the voltage along the shield will add to the analogue output of test fixture and cause some error.

Concerning the pcb layout, usually no designer makes an earth connection of the output of a power supply, if it is not absolutely necessary (it is NOT necessary in most cases). I was rather referring to the 'center point' of the internal GND connections of your test fixture. GND lines should be set up in a 'star' configuration, i.e. they should meet at one point if possible. This will be the 'real' GND terminal. However, in many cases these details are not very important if it is a small system, but for large systems with mixed analogue/digital circuitry proper layout of GND lines on the board is VERY important.

[nyborn]

Hi Buechsens:

I would like to solve the ground loop problem first:

The following is what you wrote:

As I mentioned several times, you must have a close look at your setup, down to the power supply design and printed circuit design to check all this.

My measurement does not need high accuracy. so I don't need to use opto isolator Let say "no", that means as long as my measurment get the value I fullfill with my customer requirment, that is all right? I need to make sure with you. Would you please answer to me this question directly?

All I can say that if you get results with acceptable accuracy your set up is correct for this purpose. Maybe you will get trouble when higher accuracy is needed.

It is hard to say where the 3.3mV along the shield of your BNC cable come from. Remember that DMMs have very high input impedance and will pick up a lot of noise from the environment. Also, you never know exactly whether this is a DC voltage, or some unsymmetrical AC voltage along the shield. Of course the voltage along the shield will add to the analogue output of test fixture and cause some error.

Concerning the pcb layout, usually no designer makes an earth connection of the output of a power supply, if it is not absolutely necessary (it is NOT necessary in most cases). I was rather referring to the 'center point' of the internal GND connections of your test fixture. GND lines should be set up in a 'star' configuration, i.e. they should meet at one point if possible. This will be the 'real' GND terminal. However, in many cases these details are not very important if it is a small system, but for large systems with mixed analogue/digital circuitry proper layout of GND lines on the board is VERY important.

SO, the power supply and pcb board is no earth. I got right DC voltage and current in R1 via BNC cable to NI5112. I tried different value of R1, I got correct too. I believe my setup right now is correct, right?

Another thing is to put R1 in the ground loop graph that I sent you in between AIGND and negative terminal of the power supply. It is also a ground loop from you definition, right? I need to make sure with you?

[nyborn]

Why are you so afraid of isolation of digital and analog part using opto isolator when measuring high accuracy? Could you please tell me?

If I put DGND to 7805 digital power supply ground, this is not relevant to my problem, is that right? That is what I think