Acquiring Differential vs RSE

[tbob]

Let me confuse things a little.  We call it ground, but it isn't necessarily ground.  The best way to describe it is a return path to the power supply.  Whatever leaves the power supply must return to the power supply.  So any circuit receiving the power must have provisions to return the power.  We call it ground, but it doesn't have to be connected to earth ground.  A better term to use would be common.

The method of using identical resistors to provide a return path on both sides of a differential line is called "balancing the circuit" and is commonly used in differential circuits.  What I was referring to was the differential voltage drop across a shunt resistor, commonly used to measure the current going to a circuit.  For this setup, the circuit itself must have the proper common connections, but the shunt resistor does not need any balanced connections to ground (common).  For this type of differential measurement, one AI channel is connected to the side of the shunt resistor going to the power supply positive.  The other AI in the differential pair is connected to the other side of the shunt resistor, the one eventually going to ground (power supply negative) but through the circuit whose current drain is being measured.  What I'm trying to point out is that there are different types of differential measurements.

[gstanczak]

Lynn,

In other words, a multimeter directly compares (+) to (-).  In doing so, some current flows from terminal to the other.

The same differential measurement in say the NI-9205 compares (+) to COM -to be more accurate as tbob suggests- and compares (-) to COM.  In doing so, some current flows from between (+) and COM, and (-) COM.  For this reason the COM has to return back to the power supply.

If this is correct, how is it that I can connect either end of a battery (not a plug-in power supply) to ACH0 and ACH08 and get a good reading of the voltage.  Isn't there still a bias current, where is it going in this case?  How is the battery different from the plug-in power supply?

tbob and Lynn, if you can bear with me for a few more back and forths I will get it.

Thanks,

Greg

[tbob]

I'll take a stab at explaining, although I'm not too good with words at times.

A battery does not have to be plugged into an AC power outlet.  It is completely floating.  All circuits that tie to the battery are also floating above earth ground.

A power supply, on the other hand, has to be plugged into an AC outlet.  It is usually tied to earth ground through the 3rd prong on the AC plug.  There may be some potential between the power supply negative and earth ground.  It is not completely floating.  The DAQ device also has a ground terminal.  Its differential inputs are separated from this ground.  If you tie the DAQ ground to the power supply earth ground, you should be able to use the differential inputs to measure across the power supply directly.  If you make this ground connection, you may not need the balancing resistors.  Try it.  I'm not sure this will workthough (see next paragraph).

However, if making a measurement where one side of the differential inputs is connected to power supply common, this effectively turns the measurement into a single ended one, therefore you should use RSE.  It will look at one side with respect to common (OK lets call it ground to avoid terminology confusion).  The ground side is terminated to the power supply so all currents can return to the same source.  If you use differential, one side is terminated to the ground, and you have a differential amplifier input (internal to the DAQ) with a grounded input.  This reading can be affected by common mode voltages (difference between common and real earth ground).  Geeez, I understand it but somehow find it very difficult to explain!  Why?   An input to a differential op-amp should never be grounded.  Putting the balanced resistors keeps both differential inputs floating above ground, thereby eliminating any potential differences between ground and common.  Does this make any sense?

[johnsold]

Greg,

I do not know what the internal circuit of the NI-9205 has, so I will make some general comments.

An ideal amplifier (or voltmeter) has zero input current. The amplifiers in data acquisition devices are built from transistors.  Bipolar transistors require base current to flow.  Junction FETs in normal operation have a reverse biased diode (the junction) at the gate. Some current flows in that junction.  MOSFET transistors have an insulated gate which should, in principle, require no current flow.  However, there are always leakage currents and most MOSFETs have zener diodes for overvoltage protection at the gate so some current flows in them also.  Amplifiers used in measurement instruments are usually designed to have bias currents as small as possible while still meeting the other performance requirements.

Then you get to the tough questions.  Why does the battery appear to work and the power supply not?  I am not sure I can answer that adequately and correctly.  One possibility is that the bias current can flow from one input to the other and to common internally in the amplifiers.  The battery can conduct current in either direction (charge and discharge) while the power supply may not allow reverse current flow.  There may be leakage paths sufficient to allow the device to work when using the battery but the power supply may force the common mode voltage to a level where the bias currents cannot flow.  Also, the power supply always has capacitive and inductive coupling to ground and the power mains, so there will be an AC common mode voltage present when connected to the power supply.  If the peak instantaneous voltage exceeds the maximum rating of the device, erroneous readings, and possibly damage may occur.

Lynn 

[Seth_B.]

To confuse the issue even more, we need to remember that the 9205 is an isolated module.  As a result, the "front end" of the module, the part responsible for the A to D conversion is not directly connected to the "back end", the part that passes the digital signal to the chassis.  In order for the isolators to work correctly, the "front end" and "back end" have to be within a certain potential of each other.  If the COM terminal isn't connected, the front end of the 9205 will start to float, and much like a boat that isn't tied to a dock, can float very far away from the "back end".  The can lead to all manner of unexpected behavior.

With the 9205 on my desk, I can leave COM unconnected and read correctly for about 5 minutes before it starts to drift and get noisy.  Eventually, it drifts far enough that my readings are meaningless.  But, if I touch the connector and my cDAQ chassis, I suddenly bring them back to the same potential and everything is sunshine and rainbows again for another few minutes.

Battery powered DMMs don't have this concern, because the whole system is floating.  With a cRIO or cDAQ chassis, the chassis is tied to either the power supply or USB ground and thus can't drift with the front end of the 9205.

Isolation always makes the ground question complicated.

Regards,