Digital I/O

cancel
Showing results for 
Search instead for 
Did you mean: 

DGND

 Digital systems are immune to noisy signals in a certain range and will tolerate grounding schemes which are not optimized, but it is better to avoid trouble just from the beginning.

Here you mend digtial circuit component in fixture and CB68LP can slove ground problem and noise problem? Am my understanding correct?

What do you mean by better to avoid trouble just from the beginning? Could you pls describe?

BTW in analogue data acquisition systems you can ruin signal integrity completely by wrong grounding. It is never a fault to take as much care of grounding digital signals as you would do with analogue signals

Are you talking about analog and digital signal in one board as CB68LP? need to take care the ground of the analog signal

If so, how can I reslove the same ground issue?



0 Kudos
Message 11 of 63
(1,989 Views)
Here is  a garph to describe my case:
 
 First: the operation of my test fixture is like:
 1) press "enable" button using digital I/O in Labview
  The current will flow from my fixture to NI5112 ground through BNC cable. BNC cable is a hard ground. It means most or all currents will select to go to this ground.
   I would like to ask when i press enable button, the current will go from p.0.1 to DGND, then to my fixture ground. I have a hard ground. Do you think the current will then go to the BNC cable in this opeation. I am not sure in this situation. Could you please tell me? Then the AI+ and AI- will be measuing in CB68LP board. Do you think the analog part in CB68LP will not affect the DGND (pin18) on the CB68LP board. They are the same wire that can not seperated. Are there any noise affect each other or the board (CB68LP) itselt automatically stop digital part first then go to the analog part.. Could you tell me?
 
Download All
0 Kudos
Message 12 of 63
(1,726 Views)
>How come you don't afraid of the ground loop issue?

You did not mention any other GND connections between i/o board and test fixture, so I assume the one I suggested will be the only one. If there are other GND connections of course you should not establish a second one.

>Are you saying itis much better to setup like the diagram you verify in the previous mail? If so, what other methods could I do? I could not think of.

In the diagram I verified you connected the DGND to a random point of your test fixture. To avoid any noise generated by supply currents in general it is better to use the neg terminal of the power supply.

>You are talling about different ground (earth) here.

Usually designers avoid to connect earth and GND. Anyhow, in all personal computers I have encoutered there is such a connection. If your system happens to have a connection between GND and earth, too (I hope it does not), voltage drops along the mains supply may cause severe problems.


> How come the digital circuit components can not solve this problem? Even the fixture has no earth connected, only digital I/O lines connected to earth.

Of course if your test fixture does not have an earth connection to GND you don't have to fear any hazzle with that issue.

> If the the earth (computer) has a very large ground 4.5V and the fixture has 0.3V. How come the digital components impendance can not solve this problem

This is a very common mistake. The high level voltage (4.5V) will not be severely affected but the low level voltage (0.3V) may be affected by differential voltages between seperate earth connections. A colleague currently is involved into a severe data transmission problem in an old building with poor mains power lines. Some systems work OK with the setup, but a new system failed from time to time - very probably due to problems with earth connection. We are getting somewhat off-topic now but for long data transmission lines it is better to use either fiber-optical connections (with suitable converters) or a non-GND dependent differential transmission protocoll such as RS485.
0 Kudos
Message 13 of 63
(1,977 Views)
>Here you mend digtial circuit component in fixture and CB68LP can slove ground problem and noise problem? Am my understanding correct?
>What do you mean by better to avoid trouble just from the beginning? Could you pls describe?

I am not sure about your background with digital data transmission, but I have some experience in this field, and I would strongly recommend to use the DGND terminal of the i/o board as GND connection to your test fixture. Maybe this is NOT the very best method but I think most others will be worse.

>Are you talking about analog and digital signal in one board as CB68LP? need to take care the ground of the analog signal

Properly grounding analogue signals is even more difficult than with digital signals, and more problems arise if you have to make GND connections for both analogue AND digital signals in the same setup. Anyhow, this is an issue which cannot be easily discussed from the distance. Proper grounding analogue signals sometimes even involves analysis of your power supply and wiring setup etc. I am neither a consultant nor a free-lance engineer, in some cases such a person is needed for doing it right.
0 Kudos
Message 14 of 63
(1,979 Views)

Hi Buechsens:

I would like to make it simple. I have the following questions that need your direct answer, otherwise I don't understand and I can not finish the project. Would you please explain to me?

On the Original setting you simply connected the digital input you wanted to test to BOTH dig output AND DGND. This, of course, will short the output (what do you mean by shorting the output? Becasue I know the picture of this setup, I can not get what you meant. Could you please describe to me or draw a simple picture) , and no test signal is generated. (I believe if you expalin to me about shorting the output, I may see why you said the test signal is not generated)


The 'Setting is right' seems to be correct. However, I would suggest to connect DGND directly to the neg terminal of the 9VDC power supply. This will avoid noise generated by excessive long and winding GND connections. (Did you mean the lead cable wire from the groud of the push buttons in the fixture to the DGND will contain noise?  How come the wire put in the negative terminal of the 9V DC using the same wire will have no noise?  Would you pls explain?

The third setting will only work (and probably not very good) if there is parasitic GND connection via the mains supply earth (here you meant computer ground connected to the earth, right?). Similar to a supply circuit (supply circuit means a signal source with a power supply and a resistor in a close loop circuit, right) , a signal circuit must always be closed, i.e. both need a return path. In this setting the return path is open (you meant if the DGND is not connected to the ground of the fixture, it is connected to the computer earth. It is not a return path. Why? Becaure the ground of the fixture and the DGND is connected right now, why computer ground can replace for a reture path in ths case. Would you pleaase explain? I don't understand.and you very probably will never get proper test signals ( I think you have the experience because I am exactly having this appearance.)

0 Kudos
Message 15 of 63
(1,712 Views)
>(what do you mean by shorting the output? Becasue I know the picture of this setup, I can not get what you meant. Could you please describe to me or draw a simple picture)

Ooops, I had another look. There are TWO lines going to the CN2-1 terminal, one from the dig i/o and one from the DGND but I think the latter was drawn accidentally. If you ACTUALLY connect CN2-1 to both dig i/o and DGND the signal will be shortened, but I think you are wise enough NOT to do so.


>(Did you mean the lead cable wire from the groud of the push buttons in the fixture to the DGND will contain noise?  How come the wire put in the negative terminal of the 9V DC using the same wire will have no noise?  Would you pls explain?

There is some risk that supply current flows through the internal connection between the GND of the IC in question and a randomly selected GND terminal point. You can reduce this risk by using the reference point of the power supply. Anyhow, these are general statements, and selection of grounding method depends on actual setup, including subtleties as pc board layout etc. If you find a solution which works correctly, without any noise signals, you are on the right path.

>(you meant if the DGND is not connected to the ground of the fixture, it is connected to the computer earth. It is not a return path. Why?

As I understand, your test fixture does NOT have a connection between GND (neg supply) and earth. So there is no connection between DGND/earth of the pc and GND of the fixture, and the loop is open.


> Becaure the ground of the fixture and the DGND is connected right now, why computer ground can replace for a reture path in ths case. Would you pleaase explain? I don't understand.

I do not quite understand this question. As mentioned your test fixture does not have a connection GND-earth (this should be avoided in most cases, and most designers do so). So you have to connect GND of test fixture to a GND terminal of the PC. The best location is the DGND terminal of the i/o board.

>( I think you have the experience because I am exactly having this appearance.)

Thanks. I hope some day my boss will have a similar impression... and print it on the wages bill.
0 Kudos
Message 16 of 63
(1,706 Views)

Hi Buechsens:



>(Did you mean the lead cable wire from the groud of the push buttons in the fixture to the DGND will contain noise?  How come the wire put in the negative terminal of the 9V DC using the same wire will have no noise?  Would you pls explain?

There is some risk that supply current flows through the internal connection between the GND of the IC in question and a randomly selected GND terminal point. You can reduce this risk by using the reference point of the power supply. Anyhow, these are general statements, and selection of grounding method depends on actual setup, including subtleties as pc board layout etc. If you find a solution which works correctly, without any noise signals, you are on the right path.

In that case, you meant the current flows through the internal connection between GND of the IC and a randonly selected GND terminal point are in questions. That will generated noise? You said noise is caused by a current changed. According to what you said, when the current flow throung the test fixure, the current is in question and the current is changed, is that correct?



>(you meant if the DGND is not connected to the ground of the fixture, it is connected to the computer earth. It is not a return path. Why?

As I understand, your test fixture does NOT have a connection between GND (neg supply) and earth. So there is no connection between DGND/earth of the pc and GND of the fixture, and the loop is open.

I am sorry. I meant why it MUST take DGND/earth connected to GND of the fixture to complete the path, How can not use take DGND/earth.

I don't understand. They both seems like a ground.

>

0 Kudos
Message 17 of 63
(1,706 Views)
>In that case, you meant the current flows through the internal connection between GND of the IC and a randonly selected GND terminal point are in questions. That will generated noise? You said noise is caused by a current changed. According to what you said, when the current flow throung the test fixure, the current is in question and the current is changed, is that correct?

As mentioned, it is hard to tell from the distance where to locate the optimum grounding point. Any IC will have variations in supply current when any of its states (input or output) changes, so the supply current will never be constant. Also, do NOT consider copper tracks, wires etc as a pure ohmic resistance. When there are fast changes in supply current the parasitic inductance comes into play - and this may cause much higher voltage spikes than expected.

>I am sorry. I meant why it MUST take DGND/earth connected to GND of the fixture to complete the path, How can not use take DGND/earth I don't understand. They both seems like a ground.

Yes, this is a very common mistake. You never know where supply current(s) flow in a pc system (it is difficult enough to tell in a self designed system). So you never know whether there will be any voltage shifts between earth and DGND of the PC. And you never know what voltage shifts occur between earth terminals of two different mains outlets.

Just an experiment: Take a DMM, set it to the lowest voltage range and connect it to the DGND terminal of the dig i/o board. Now take the other test tip and wander around the pc metal case (with the pc switched on, of course). You probably will get slightly different readings at any point you touch. Do you still think that DGND and earth of the PC are identical???

Believe me, there is more to grounding and earthing than meets the eye...

These differences in voltage will add (or subtract, in case they are negative) to the low level voltage of the digital IC. As soon as you have more than 0.8V at the low state you are in an illegal region between low and high, and you never know what your system does in this case. (You can avoid quite a bit of trouble by using ICs with Schmitt trigger inputs - many of the boards I designed are cluttered with Schmitt trigger driver IC or at least driver ICs with input hysteresis. Of course you still have to respect proper grounding - once you have noise on a ground line there is no way to filter it since all filters need a ground terminal....)
0 Kudos
Message 18 of 63
(1,703 Views)

Hi Buechsens:

 

 As mentioned, it is hard to tell from the distance where to locate the optimum grounding point. Any IC will have variations in supply current when any of its states (input or output) changes, so the supply current will never be constant. Also, do NOT consider copper tracks, wires etc as a pure ohmic resistance. When there are fast changes in supply current the parasitic inductance comes into play - and this may cause much higher voltage spikes than expected.

So, in conclusion, it is better to put a DGND (earth) to the negative terminal of the power supply, since wire has a noise, IC have variations that cause noise, in queston to serch ground. Once you put it in the negative terminal of the power supply, it is optimum to reduce troubles, am I correct?

 

I am sorry. I meant why it MUST take DGND/earth connected to GND of the fixture to complete the path, How can I not just use DGND to computer ground without wiring GND of the fixture  I don't understand. They both seems like a ground. I chnage it here.

Yes, this is a very common mistake. You never know where supply current(s) flow in a pc system (it is difficult enough to tell in a self designed system). So you never know whether there will be any voltage shifts between earth and DGND of the PC. And you never know what voltage shifts occur between earth terminals of two different mains outlets. (I am confuse here because DGND and the computer ground, don't you know they are the same if you use DMM, why you said different here, I don't understand , could you pls explain?  

0 Kudos
Message 19 of 63
(1,699 Views)
Just an experiment: Take a DMM, set it to the lowest voltage range and connect it to the DGND terminal of the dig i/o board. Now take the other test tip and wander around the pc metal case (with the pc switched on, of course). You probably will get slightly different readings at any point you touch. Do you still think that DGND and earth of the PC are identical??? But they go beep in diode mode. What does it mean? COuld you pls explain? I don't understand

(i don't want to out of my topic, my topic is if I put wire from DGND to computer ground, I mean not connect DGND to test fixture ground, why can't it set this up to complete the circuit, could you please explain?)

 

Please give your answer directly. It is much easier

0 Kudos
Message 20 of 63
(1,700 Views)