Are analog inputs to multifunction Daq boards grounded to computer?

[timothyjb]

I'm considering getting one or more 622X boards, but I need to determine whether or not I should have the analog input signals conditioned by using isolation amplifiers. I'll be acquiring the analog outputs from several gauge readouts (which I've been told by the manufacturer I cannot tie their grounds together) and I don't want to create ground loops between the PC and the gauges. If I take the analog output and its ground from each unit and connect them (unconditioned) as a differential input pair to a 622X board, is that input pair isolated from ground enough to avoid ground loops?

Thanks,

Tim

[Thaison_V]

Hi Tim,

You are right on with using a differential mode analog input acquisition to measure your gauges.  If each of your signal sources are referenced to their own grounds, measuring them each with a differential pair will keep each gauges' grounds isolated from each other.  Also, when you connect a signal to our data acquisition cards in differential mode, the measurements are not referenced to the card's ground either.  Take a look at the following Developer Zone article that discusses wiring/grounding issues: http://zone.ni.com/devzone%5Cconceptd.nsf/webmain/01F147E156A1BE15862568650057DF15?opendocument .  Particularly, the section on measuring grounded signal sources may be helpful.

Thaison V

[timothyjb]

Thanks Thaison, That was very helpful. Now I have a followup question. I also plan to use the digital I/O on the same 622X board, and I wonder if I'll have the same problem with that. My gauge readout has some digital output lines that indicate which scale its on. They use the same ground as the analog signals. How do you recommend I handle those?

Thanks,

Tim

[Thaison_V]

Hi Tim,

All ground pins on the M-Series boards are all tied together, as mentioned in the following KB: http://digital.ni.com/public.nsf/websearch/02A451C0A88A3EAE8625702F0059C98E?OpenDocument .  All of the digital lines on the 6220 are referenced to the same ground, so there is no way to directly connect multiple digital grounds to the board.  You may need to find some way to get all of your sensors to reference their digital output signals to the same ground, if it's possible, and then connect this one ground into one of the D GND pins.

Thaison V

[timothyjb]

Thanks Thaison,

Now I have a followup question related back to my original analog input situation, this time having to do with signal cable shielding.

I have several gauge readout units and for each one I will have an individual twisted pair going to its analog output signal and ground (the other end will be connected to differential inputs of a PCI-6225). The shells on the DB9 output connector of the gauges don't seem to be connected to either the signal ground or the case ground. Actually, neither the shell nor the signal ground nor the instrument case seem to be connected electrically. So I'm suspecting I could connect these shells together, unlike the analog signal grounds. When a cable is connected to an output jack like this, usually the shell of the jack would get electrically connected to the shielding of the cable. I plan to have a multi-conductor cable bundle, from which one twisted pair from the bundle will go to each unit. I can either have one overall shield for the whole cable bundle and connect it to the DB9 shell of each gauge unit (thereby connecting all the shells together), or I can use a cable bundle that has individual shields for each pair of wires, and then connect each individual shield separately to each shell of every gauge readout, thereby avoiding connecting any shells together. So my question is, is it necessary to avoid connecting the shells together, requiring me to use individual shields, or can I just use an overall shield and connect the shells together to the shield? Since the shells seem to be floating, that would mean any shield connected to it would also be floating, and that isn't good according to the article you referenced. That would also mean there wouldn't be any advantage to using individual shields. Perhaps I should avoid using the shells altogether and just use an overall shield and connect it to the case of each gauge. If there is another recommended way to do this, I'd also like to know that.

Thanks,

Tim

[Thaison_V]

Hi Tim,

I'm curious about why the shells need to be used in this case, since you should be able to ground the signal end of the individual shields to some kind of ground (possibly the chassis ground?).  As mentioned in the article, to avoid capacitive coupling, you will not want to leave the shielding floating (which would be the case if you have your shielding connecting to the shell), and this can increase capacitively coupled noise.  I think your best bet would to avoid the shells (unless there is a particular reason you are using them), and ground your shields on one end.

Thaison V

[timothyjb]

Thaison,

I anticipate using a cable bundle that splits at the end into 9 twisted pairs, each pair being soldered to a male DB9 connector that mates with the DB9 female connector on a gauge readout unit. The shell of the DB9 output connecter on each gauge readout unit would become connected to the shell of the connector for the twisted pair when it gets plugged in. I guess the shells don't need to be connected - I could make sure the cable is constructed so that if I use individual shields, none of them gets tied to any of the connector shells. I could then connect a short lead from the shield to the gauge's chassis. But if the gauge's shell is really floating, then it wouldn't matter if it is connected to the shield, as long as the shield gets connected to the chassis to ground it (or them). Now, since all the gauge readouts are all mounted to the same strip of steel, if I do connect the shield to the gauge's chassis, it shouldn't matter if I use one overall shield or if I use individual shields. I think I should go with a single overall shield and then connected that shield to one of the gauge readout's chassis and not worry about the shells on either the cable or the gauge readouts. Does that sound reasonable to you?

Thanks,

Tim

[Thaison_V]

Tim,

What you have posted makes sense.  With individual shields though, you may not have to worry about the individual signal wires interfering with one another.  I'm not sure if this will be a big concern for you, with the type of sensors you are using.  Otherwise, using one big shield would be the same as using individual shields since they all are grounded to the same chassis.

Thaison V