Isolation considerations for field wiring

[cktce]

I am planning on acquiring multiple 0 to 10 volt analog signals using SCC AI-03 modules.  I wanted to achieve channel to channel isolation, so I was planning on hooking up only 1 signal to each AI module.  However, searching through the NI website, it appears that no matter how I hook up my signals I will break the isolation barrier.  If I have floating signals, it is recommended to tie all the signals negative inputs to the AI Sense terminal in the SC-2345 carrier to eliminate bias currents.  If I have ground referenced signals, all the signals negative inputs will be tied together via the ground reference.

I've looked at other products such as the SCXI-1125 which specifically provides channel to channel isolation, but it's instruction manual also says the isolation barrier is bypassed when you hook up floating signals negative input leads to ground.  It does not specifically say the isolation barrier is breached if you have ground referenced signals, but I don't see how the isolation barrier is not breached considering all the signals negative leads are tied to ground.

Is there a way to hook up my signals in a way that does not breach the isolation barrier (either channel-channel or channel-ground)?

[Jason_D]

ETNC,

If performing differential connections on a channel to channel isolated board then you will not break isolation and everything will work fine.  If you instead decide to connect all of your inputs to an external common ground there is no way the module can isolate this as it is happening external to the card no matter if it is channel to channel isolated or not.  You can use external circuitry if this is what you would like to do.

[cktce]

Hi Jason, thanks for the input!  I'd appreciate if you could provide a little clarification...

So if I'm reading you right:

(1) Connecting any externally ground referenced signals to any ch-ch or ch-gnd isolated board will break the isolation barrier.

(2)  Since NI recommends wiring floating inputs of both Differential and NRSE type acquisition cards to a common ground/reference point internal to the acquisition card via bias resistors, this common reference point internal to the card will not break the ch-ch or ch-gnd isolation barrier.  I'm having a hard time seeing how tying all the inputs to a common point doesn't break ch-ch isolation, even if it is internal to the card.  Or are you saying only Differential type acquisition cards are not susceptible to bias currents and thus do not need bias resistors?  I realize that NRSE inputs share a common floating negative input between all channels on the card, so maybe you're saying the bias resistors are good enough to isolate Differential inputs from each other?

[Jason_D]

ETNC,

1)  No, connecting each externally grounded signal to a different outside ground will work just fine and this is what I would recommend.  If you connect each to the same external ground you are externally breaking the isolation barrier.

2)  We have 1 ground with which to use.  If you have multiple floating signals and need to ground each of these the isolation barrier will have to be broken.  There is no way to hook to the same ground and keep isolation.  We can minimize the effects that this has, but at the end of the day you must ground at least all but 1 of your signals externally or you will break isolation.  I would recommend that you give it a try and see if you are getting unacceptable results. 

[cktce]

Hi Jason,

Unfortunately for me, it's a long and difficult task just to purchase let alone test different configurations, so I'd like to make sure I understand what's happening with the inputs and isolation.

For simplicity sake, whenever I refer to the article, I'm referring to this:

http://zone.ni.com/devzone/cda/tut/p/id/3344

1) When you say "connecting each externally grounded signal to a different outside ground will work just fine", are you suggesting to reference each signal to its own, seperate, isolated outside ground, and not the building outside ground?  In this case, I would essentially consider each of the inputs as floating, from the viewpoint of the DAQ card, since they would be referrenced to different potentials.  The article states "A variety of phenomena—for example, the instrumentation amplifier input bias currents—can move the voltage level of the floating source out of the valid range of the input stage of a data acquisition device."  Am I missing some concept?

2) The article also states:

"Warning: Bias resistors must be provided when measuring floating signal sources in DIFF and NRSE configurations. Failure to do so will result in erratic or saturated (positive full-scale or negative full-scale) readings."

The attached picture is how the article recommends hooking up different types of inputs to different types of measurement systems also.  In all the cases either the floating signals are hooked to a common reference via biasing resistors, or the signal inputs are tied to the external building ground.

So when you say to give it a try and see if I get unacceptable results, are you saying that in some cases bias resistors are not needed to obtain accurate measurements on floating signals?  Are there criteria that could help define these cases?

[Jason_D]

ETNC,

1)  The 100k resistor will almost definately work for you especially when considering all of the other error possiblities in the devices you are connecting to.  From a theoretical standpoint yes in order to be TRUELY isolate and TRUELY grounded you will need individual building grounds that are tied directly into the earth.  As you can understand this is overkill. 

2)  If you want to get more difinitive answers I need more details.  You say " bias resistors are not needed to obtain accurate measurements on floating signals?" what is "accurate" measurements?  Are you refering to accuracy of +-1V or +-1uV, it can make a difference!  I cannot tell you how much error you will get without knowing many other things about your system for example source impedance of your DUT.  When I said give it a try I was more in general making the comment that you seem to not know the specifications you need thus I was suggesting that you find out what you are getting from a simple setup and see if this is acceptable before spending all this time looking into these other methods.

[K C]

Hi,

If you want to maintain your ch-ch isolation maybe it is easier to use external barriers. You can connect the outputs of these barriers to one DAQ board. It depends on the your signal and further specification, but 0-10V barriers are easy to find (I think)

[cktce]

Thanks for the input everyone.  I think I'll ultimately end up hooking the DAQ devices per the manual's recommendations, which sounds like it will break the isolation barrier.  That external barrier sounds like a good idea, I'll look into that.