Multifunction DAQ

cancel
Showing results for 
Search instead for 
Did you mean: 

AI GND vs AI SENSE for NRSE inputs

My firm contracted with a electronic test equipment provider to design and build a PXI-based data collection and control system for a production test application.  As a part of this effort, the contractor designed a custom interface board to condition various analog signals to be measured by a PXI-6025E DAQ card.  These analog signals include five 4-20 mA transducers.  Shunt resistors are used to produce voltage signals from these current sources.  We are experiencing unexplained shifts in the measured voltage values.  In reviewing the schematic for the interface board, I discovered that the two AIGND pins on the DAQ connector are shorted together.  In addition, the AISENSE, AOGND, and DGND pins are shorted together on the board (but separate from the AI GND circuit) and serves as the interface board's ground reference.  Thus, the AIGND pins on the DAQ card do not appear to be connected to anything.  Sadly, the contractor cannot provide support for this board as the engineering group that produced it has been "downsized."
 
The lack of connection of the AIGND pins is consistent (I think) with NRSE connection of a non-floating signal source.  The shorting of AISENSE to the other ground pins on the DAQ (as well as the ground for the interface board itself) seems a bit fishy.  Is this connection scheme proper?  What changes should be made?
 
Regards,
Chris Yohpe
0 Kudos
Message 1 of 20
(19,767 Views)

Hello Chris,

 

How much is the voltage shift that you are measuring? Is it a +ve or a -ve voltage shift?

The way it looks right now, you might need a bias or a pull-up resistor to offset this shift.

Shorting of AI SENSE to ground is definitely not a good idea as you would not be able to connect NRSE signals any more. I have attached an image file that illustrates proper signal connection for various types of input sources.

 

Chetan K

Application Engineer

NI

0 Kudos
Message 2 of 20
(19,746 Views)
Chetan:
 
The value of the shift is only 20mV, but the full-scale signal is only 2VDC, thus the error in the measured value is about 1% for one critical channel.  It turns out that this error is about 10 times greater than we can tolerate.  The error seems to occur in the plus direction.  I verified this by measuring the voltage drop across the shunt resistor (100 ohms) with a 6-digit DVM and comparing the value as read by the DAQ.  During part of a test sequence, the values are in good agreement.  Then, suddenly, the DAQ value jumps by 20mV while the DVM value remains constant. 
 
I have attached a PDF version of the schematic for the interface board.  The two connectors on the left, J1 and J2, feed signals to the 6025E DIO connector.  Lines 1-50 on J1 connect to pins 1-50 on the DAQ.  Lines 1-50 on J2 connect to pins 51-100 on the DAQ.  The signal in question is TANK PRESSURE that employs shunt resistor R24 to produce a voltage on AI0.
 
Regards,
Chris Yohpe
0 Kudos
Message 3 of 20
(19,730 Views)
Hi Chris,

Just to clarify things a little, the AIGND, AOGND, and DGND on our E Series boards such as your PXI-6025E are all connected internally.  The following KnowledgeBase article here discusses this a little bit more in detail.  Chetan is correct in that having AISENSE shunted to these other grounds probably isn't a good idea, as this would mean that your NRSE mode measurements will be referencing its ground exactly the same as if it was a RSE mode measurement since AISENSE and AIGND are essentially connected.  What mode is your software configuring your analog input to be read on (Differential, RSE, NRSE)?  Is the tank pressure's shunt resistor being grounded into one of the grounds on the DAQ board, or is it an external ground?  The ground will tell us what mode we need to be acquiring in and if the current grounding scheme will be appropriate for this sensor.  Let us know, and we will do our best keep helping you!

Thaison V
Applications Engineer
National Instruments

0 Kudos
Message 4 of 20
(19,707 Views)

Thaison:

If you examine the interface board schematic (IF PCB.pdf), you will see that the tank pressure shunt resistor, R24, is connected to the PCB ground plane.  This ground plane is connected to DAQ ground via pins 19, 23, 24, and 33 on connector J1; these pins correspond to DAQ pins AISENSE, AOGND, DGND, and DGND respectively.  As you will observe, these pins are tied together on the PCB.  You will also note that the PCB ground is connected to the incoming 24VDC minus terminal.

The tester chassis is grounded via the incoming ground line on the 120VAC.  This ground reference is also connected to the minus terminals on the tester's DC power supplies, including the 24VDC supply that powers the tank pressure transducer (4-20mA output).  The safety ground lugs on the backs of the power supplies are also tied to this reference.  At the moment, the safety ground lug on the back of the PXI chassis is unconnected.

The channel definition for tank pressure shows the mode as NRSE.

The KnowledgeBase article you referenced explains how the various grounds inside the E-Series DAQ are connected to a single reference, the bus ground on the PXI backplane.  Is this bus ground internally connected to another ground reference?

Thanks for the help.

Chris Yohpe

0 Kudos
Message 5 of 20
(19,667 Views)
Hey Chris,

First off, this issue really should be supported by the maker of this product.  While our board is in there, the issue could be coming from any number of places.  I would definitely try contacting them again and be a bit more persistent.  Unfortunately if they really are having the cutbacks and the person left, then we may be back to ground zero. 

Reiterating what Chetan and Thaison said, they really shouldn't have tied the AISense to AIGnd unless you have a floating source for your measurements.  While this could be part of the issue at hand (very easily could since you could have multiple ground planes causing ground loops and therefore shifting voltage readings). 

Since the grounds may or may not be part of the issue, then I think we should take a quick step back and make sure that we are viewing the whole problem and not missing something.  First off, is this the first time that you've set up the system?  If not, were you already running the system for awhile before you found this problem?  If it is the first time, if there any verrificaiton that it worked properly before?  Next step is if it was working, then to check to see what could have possibly changed between the time when it was working and now.  If it was never working properly, then we'll need to know a bit more info about when and what the errors are.  Are the only problems on these 3 analog input lines?  If not, what other errors are there?  Can you measure a voltage difference between the ground pins and the chassis ground?  If so what is it?  Do you have any other test systems of this type that are working correctly?

Let's take the highest level approach and begin working from there, because I'm not totally convinced that it's just a signal connection issue.

Message Edited by Otis [DE] on 05-17-2006 05:01 PM

0 Kudos
Message 6 of 20
(19,642 Views)

Otis:

Thanks for the reminder about "the big picture" and basic troubleshooting techniques.  It is difficult to tell whether the problem existed since the system was set up, as this work was done at the vendor's facility.  The pressure signal problem was masked by an automatic pressure-regulation setup - any shift in the apparent pressure was nulled out by the regulator.  It is only when we put a mechanical, lab-grade gauge on the vacuum tank that the problem revealed itself.  By that time the members of the team that developed the tester had been dispersed across the country.  We could try to contact them for advice, but they are the ones who designed the system in the first place, if you take my meaning.

Since my last post, we have been running tests to determine the precise conditions that precipitate the shift in the measured pressure signal.  By operating the various tester devices manually, we were able to determine that switching a certain 120VAC solenoid has a nearly 100% correlation with the shift.  This solenoid is one of four AC units on the tester.  Each of these solenoids is supplied with 120VAC power on one lead and a PXI relay card switches the remaining lead to neutral.  The suspect coil measures 42kohm and should be more like 800ohm.  We are investigating whether faulty wiring or a failed coil, or both, are present.  It is interesting to note that the solenoid still operates its valve.  This tester is unique and thus we have no other for comparison.

After we replace the solenoid valves and inspect/repair the wiring that controls them, we will perform further tests.  While I certainly appreciate NI's help in this matter, I realize that your assistance should be focused on the NI equipment and its operation.  That being said, I would like some help on determining the proper connection of the various grounds when using an interface board such as ours.  As shown in the PCB schematic (IF PCB.pdf attached to an earlier post in this thread), analog inputs, analog outputs, and digital signals are all handled (and conditioned) by the interface board.  Should the grounds (AI SENSE, AI GND, AO GND and DGND) each have their own ground planes on the board?   Which ones should be tied together?   As you can tell, I am not a EE.

Regards,

Chris Yohpe

0 Kudos
Message 7 of 20
(19,627 Views)

Let me know how replacing that solenoid works out.

Regarding the grounds.  AIGND, AOGND, and DGND should be separated when coming into the board just to try and make sure that any ground noise is as system isolated as possible.  However, ultimately all three of those ground planes will be referenced to the same PCI ground plane, so the three grounds should all be at the same potential.  AISense should not really be tied to ground unless you are using a floating voltage source, then you should connect something called a bias resistor (10k-100kOhms) between AISense and AIGND.  This allows electrons to flow, but keeps the signal stable.  If you are measuring a grounded source then nothing should be connected to AIGND, period.

I hope this clears up some of the grounding questions.

0 Kudos
Message 8 of 20
(19,604 Views)

Otis:

Thanks, your message does clear some of the fog.  You mentioned that all three grounds (AI, AO, D) are tied together internally on the DAQ card.  I assume that this ground is connected to a PXI chassis ground in some way.  How is this internal PXI ground connected to an external ground reference?  Is it through the grounding terminal on the AC power plug?  I cannot find this stated explicitly in any of the NI documentation.  Perhaps it is so obvious that no mention was deemed necessary.  Please forgive my ignorance.

I mention this because the PXI chassis is powered via a UPS.  We measured continuity between the grounding terminals on the output of the UPS and the AC power plug that supplies the UPS, so we believe that a path to our tester's chassis ground exists.  Would it be preferable to drop the grounding via the AC plug (using one of those 2-to-3 prong adapter plugs) and connect the PXI chassis ground lug directly to our tester chassis ground?

Regards,

Chris Yohpe

0 Kudos
Message 9 of 20
(19,589 Views)
Hi Chris,

Sung to the tune of "Dem Dry Bones":
Your AI Ground's connected to your AO Ground. 
Your AO Ground's connected to your D Ground. 
Your D Ground's connected to the your AI Ground.
And they're all connected to the backplane.

Your backplane's connected to your power supply.
Your power supply's connected to your earth ground.
Your earth ground's connected to your working place.
So don't connect two earth grounds.
Regards,


Message 10 of 20
(19,567 Views)