DGND

[Buechsenschuetz]

I am sorry but I cannot answer these questions from the distance, and I do not know anyone who can...

Checking proper GND connections, especially when analogue signals are involved, is not very easy, and sometimes you even have to check the board layout of the device in question. It depends on the accuracy and/or level of the analogue signals you have whether a grounding scheme works well or not. 'Works well' means that you can get acceptable performance even without the best grounding scheme.

Of course you would be lucky if both analogue and digital GND (AGND and DGND) would be floating, i.e. AGND and DGND would be completely independent. But this can only be achieved by separating digital and analogue circuitry by opto-isolators etc. I had to fight with a mixed digital/analogue setup for some time and finally decided to redesign it completely following this guideline. Our accountant was not amused (by the extra costs) but the big advantage is that the system works which is a major advantage....

Sorry but I think it is close to impossible to check your system for possible GND loops between analogue and digital circuitry from the distance.

[nyborn]

Hi Buechsens:

I understand what you meant to slove problem from a distance regarding the flowing of analog signal and digital signal. But thanks for your big word, acceptable performance even without the best grounding scheme. 

That mean I don't need to care the different ground of digital part and the analog part with BNC cable (hard ground.) As long as the current go through the BNC cable finally and the digital signal current and voltage go to the digital IC correct, it is acceptable performance. Am I correct? I need a direct answer.

The following is what you wrote:

Of course you would be lucky if both analogue and digital GND (AGND and DGND) would be floating, i.e. AGND and DGND would be completely independent. But this can only be achieved by separating digital and analogue circuitry by opto-isolators etc. I had to fight with a mixed digital/analogue setup for some time and finally decided to redesign it completely following this guideline. Our accountant was not amused (by the extra costs) but the big advantage is that the system works which is a major advantage....


I have a sugestion and I don't know if that work. That is I put a bias resistor 10Kohm from AI- to DGND on CB68LP board, can it seperated the digial ground and analog groud. See my attached graph to the setup

[nyborn]

I used DMM to measure the floating GND and earth (computer chasis). I found 0.9mV, very small value. Could it be consider as 0V?  Is that what you said about the following:

Since there is no voltage source between floating GND and earth there will never be any voltage between DGND and earth, Or, more precisely (since DGND is connected to earth in your pc and on the same potential), there will be no compensation current flowing into (or out of) the DGND terminal.

I am quite sure if what I am doing is exactly what you said. Could you please tell me?

 Is this 0.9mV always there becasue of the resistance of the wire? I need to verify with you.

2) I have a dimm light problem on my fixture. Attached is my graph. Without connecting DGND to the fixture. When the swtch on the fixture turn off. There is no light on reset. But after I connected DGND to the fixture, I have dimmed light. Is this normal? Could you pleas explain?

[Buechsenschuetz]

There is no direct answer - I simply can't judge this problem from here.

The problem is that on some i/o boards (like one from a different manufacturer we use) AGND and DGND are tied together at a certain (and properly located) point on the board. Also, on some analogue data acquisition boards AGND is connected to the host pc GND and will be connected to DGND of dig i/o boards (which in general is not isolated) via the internal GND connection inside the pc.

Since on your test fixture there is only one GND (common GND for digital AND analogue signal - they are fed by the same power supply), installing one connection between test fixture GND - dig i/o board DGND (as required for digital signal return line) and a second one between test fixture GND - AGND of i/o board (or other board with non-isolated AGND terminal) it is possible that there will be a ground loop via the shield of the BNC cable.

The only way to avoid this completely is to use galvanic isolation (via opto-isolators) of either the digital or analogue signal(s). Isolation of digital signals is much easier.

Anyhow only a test will reveal whether this is necessary. It depends on the absolute value of your analogue signal and on the accuracy required.

[nyborn]

Hi Buechsens:

I really need you to give me direct answer in the following. It is very important to me please.

I used DMM to measure the floating GND and earth (computer chasis). I found 0.9mV, very small value. Could it be consider as 0V?  Is that what you said about the following:

Since there is no voltage source between floating GND and earth there will never be any voltage between DGND and earth, Or, more precisely (since DGND is connected to earth in your pc and on the same potential), there will be no compensation current flowing into (or out of) the DGND terminal.

I am not  quite sure if what I am doing is exactly what you said. Could you please tell me?

 Is this 0.9mV always there becasue of the resistance of the wire? I need to verify with you.

[Buechsenschuetz]

>I used DMM to measure the floating GND and earth (computer chasis). I found 0.9mV, very small value. Could it be consider as 0V?  

Not really... it depends what you want to measure. Of course a 'base line shift' of 0,9mV very probably will not affect operation of digital circuits.

Did you measure with the floating GND (= GND of test fixture) connected to DGND of i/o board? I was rather referring to (non-existing) voltages between OPEN (i.e. not connected) floating GND and DGND.

If you measure a voltage between floating GND (unconnected) and pc earth there MUST be a ground loop somewhere. (Note that DMMs with high input impedance will display random values, usually in the range of hundreds of mV, when connected to an open circuit).

> I have a dimm light problem on my fixture. Attached is my graph. Without connecting DGND to the fixture. When the swtch on the fixture turn off. There is no light on reset. But after I connected DGND to the fixture, I have dimmed light. Is this normal? Could you pleas explain?

This is hard to tell. It may be parasitic voltage floating around somewhere. Since your test fixture uses CMOS ICs maybe there are oscillations somewhere (but I don't think so since pull-up resistors are used on all inputs).

[nyborn]

Hi Buechsens:

I might not say the question clear. I would like to say clear here again

In my setup, I don't have to care the flowing of digital signal and analog signal because as long as I get the right value (current and voltage) throung BNC cable  to NI5112 and the LabVIEW can control the push buttons of the fixture. It is the acceptable performace, I think?

Could I ask you that you think even I have a BNC hard ground connected from the test fixture to NI5112 board, the digital push buttons parts must go to the fixture ground first due to the operaton of this flow. In this setup I tried put AI+ and AI- from CB-68LP to the resistor form the output of the current meter in the test fixture, there is no current going through because I put a BNC cable which is a hard ground. The current will all go through to the hard ground. In mixing with operation of digital push buttons in LabView and the analog current go to the BNC cable to NI5112, the situation will be different becasue The first thing need to do is push push button in LabView, the TTL signal must go to the ground of negative power supply, right? Then, go to the BNC hard ground. This is what I think in this special case. Am I right?

2) I know you are talking another equipment like optical equipment to seperate the digital ground and the analog ground. I still didn't get it why I put a bias resistor 10Kohm from AI- to DGND on CB68-LP board. It does not seperated, could you please explain simply? That way it will be easy to understand that part. This part is very importnat.

[nyborn]

Hi:

I still have a doubt.

The following is what you said:

I used DMM to measure the floating GND and earth (computer chasis). I found 0.9mV, very small value. Could it be consider as 0V?  

Not really... it depends what you want to measure. Of course a 'base line shift' of 0,9mV very probably will not affect operation of digital circuits.

If you treat a connection wire from DGND pin of the CB68LP board to the negative ground of the power supply of the test fixture. The DMM will go "beep" in diode mode. That means it is a same wire. If you just take any wire (like cable I use for connecting DGND to the ground of the negative terminal of the power supply, it give you 0 volt, why in this case, it give be 0.9mV?) I think if you answer me this question directly, I will slove serval problems please.

Did you measure with the floating GND (= GND of test fixture) connected to DGND of i/o board? Yes I was rather referring to (non-existing) voltages between OPEN (i.e. not connected) floating GND and DGND. (what do you mean by here? not connected floating GND and DGND. You and me understood MUST connect DGND to the negative terminal to the power supply of my floating source fixture. I don't get it? Are you talking about different thing? I am confuse here.

If you measure a voltage between floating GND (unconnected) (that mean no wire from DGND pin of the CB68LP board to the negative terminal of the power supply of the test fixture, right?) and pc earth there MUST be a ground loop somewhere.(It seems I am right, if so, I did it and it was exactly what you said. Thanks for your explaination. (Note that DMMs with high input impedance will display random values, usually in the range of hundreds of mV, when connected to an open circuit). (Did you remeber the tutorial I show you and you said ground loop is caused by two different individual ground signal source. How come there is a ground loop?) I am afraid I don't mention here, you will not see. So, I need you to see if the first blue line is right? Then, I will know next.

> I have a dimm light problem on my fixture. Attached is my graph. Without connecting DGND to the fixture. When the swtch on the fixture turn off. There is no light on reset. But after I connected DGND to the fixture, I have dimmed light. Is this normal? Could you pleas explain?

This is hard to tell. It may be parasitic voltage floating around somewhere. Since your test fixture uses CMOS ICs maybe there are oscillations somewhere (but I don't think so since pull-up resistors are used on all inputs).

Yes. It makes sense to me when you said parasitic voltage floating around somewhere because I tried the setup. Sometimes the setup has no dimm light and sometimes it does. Yes, I believe you are saying those 4.7Kohm resistor, right? I don't think I have any oscillations. How does this parasitic voltage caused? Is it caused by the digital I/O lines connected from PC to the fixture because there were no this problem without connecting it. How come this parasitic voltage will appear when matching with NI boards and will this be affect my measurement and my circuits in the test fixture? I really need to know becasue it is very important.

[nyborn]

Hi Buechsens… :

I am very sorry to reply like that. I thought about it. I think I should say that I am afraid if I don't give this feeback to you. The next day you will get lose of the picture of my problem. I am very sorry.

Please forgive what I did.

[nyborn]

Hi Buechsens:

The black text is what you wrote. I still don't get it.  Could you please answer to me?

Since there is no voltage source between floating GND and earth there will never be any voltage between DGND and earth, Or, more precisely (since DGND is connected to earth in your pc and on the same potential), there will be no compensation current flowing into (or out of) the DGND terminal.

I used DMM to measure DGND and earth of the pc. It said 1mV, not what you said above

Is this a baseline shift? If so, why cause base line shilt?

Not really... it depends what you want to measure. Of course a 'base line shift' of 0,9mV very probably will not affect operation of digital circuits.

Since on your test fixture there is only one GND (common GND for digital AND analogue signal - they are fed by the same power supply), installing one connection between test fixture GND - dig i/o board DGND (as required for digital signal return line) and a second one between test fixture GND - AGND of i/o board (or other board with non-isolated AGND terminal) it is possible that there will be a ground loop via the shield of the BNC cable.

A ground loop is probably occurred via the shield of the BNC cable. In fact, could you pls explain what ground is here? I don’t quite get it and can we use DMM to measure.

The only way to avoid this completely is to use galvanic isolation (via opto-isolators) of either the digital or analogue signal(s). Isolation of digital signals is much easier.

Anyhow only a test will reveal whether this is necessary. It depends on the absolute value of your analogue signal and on the accuracy required.

When you said “Anyhow only a test will reveal whether this is necessary. It depends on the absolute value of your analogue signal and on the accuracy required.”

Did you mean the result I got from NI5112 board measurement? Could you please give directly answer here?