Is there a way to isolate the input channel grounds on PXI 5112?

[bosozoku]

I have two PXI-5112 Oscopes in the same chassis (PXI-1042Q) that it would be really helpful to isolate the input ground from the chassis ground.  Is there a method for doing this?  We are testing two separate circuits at the same time and would like them to not be grounded to each other.  Also can it be done for isolating a 3rd party scope connected to the GPIB card used in the same chassis?

[Matt_E.]

The BNC ground on the 5112 is directly connected to the front panel and the board's ground plane so there is no way to isolate it from the rest of the chassis ground.  This is also true of all of our other PXI and PCI digitizers with the exception of the 5102.  The 5102 is connected in a pseudo-differential configuration which basically means that the BNC ground is not directly connected to chassis ground but it is not isolated either.  The BNC ground on the 5102 is connected to chassis ground through a capacitor and diodes.  This is because the purpose of the pseudo-differential configuration is to break DC ground loops caused by small differences in the ground potential of the digitizer and the measurement source.  If the ground potentials get more than a diode drop apart you can damage the digitizer so this configuration is not at all the same as an isolated ground.

NI does not currently sell any isolated high speed digitizer products with specs comparable to the 5112.  It is possible to use the USB-5132 or USB-5133 with an external USB isolator.  The resolution and sample rate of the 5133 is the same as the 5112 but the analog bandwidth is 50MHz instead of 100MHz.  If you would like to give more details on your requirements we might be able to suggest alternative setups.

Regards,

-Matt

Message Edited by Matt E. on 10-10-2008 05:16 PM

[DFGray]

A possible method for doing this, provided you only need to see one signal at once, is to connect the signal input of the source to channel 0 and the ground to the input of channel 1, then subtract them in software.  This effectively gives you an instrumentation amplifier at the cost of half the useable channels.  This may or may not work for you, given the noisiness of the 5112 input.  There are a number of other hardware solutions (e.g. isolation transformer, isolation amplifier).

[Matt_E.]

You have to be a little careful with where you are making the connections for the method described above.  It is possible to use the second channel to compensate for DC ground offsets but you need to be sure to connect the second probe to the ground point at the signal source.  Connecting the probe at ground on the digitizer side will only measure the value of the digitizer's local ground which is the same ground that the input amplifier is referencing anyway so it will not actually contribute any useful information.  Using 2 inputs with subtraction also does not provide any common mode rejection in the analog signal chain as a true differential amplifier would.  Any variations in the gain/offset/bandwidth etc. of the two channels will appear as part of your measured signal.  This wouldn't be a big deal if all you are measuring with the second probe is a DC offset but limits the usefulness of the setup when trying to measure differential signals.  The biggest thing though is that like the psuedo-differential scheme of the 5102 this method does not provide isolation at all so if the goal is to provide isolation for safety or to allow the board's ground to float on a DC offset to allow measurement of a signal that is outside the normal measurement range of the board then this is not a valid option. 

Isolation transformers can be used if you don't need to measure DC or low frequencies.  Isolation amplifiers are certainly a possibility but can be very costly and will not provide the same bandwidth of the 5112.  Overall since there are many different reasons people use isolation it is important to know what you are trying to accomplish by adding the isolation.  Wide bandwidth isolation is usually an expensive proposition and there can be many tradeoffs involved so with some more details about what you are hoping to accomplish it is possible there could be a simpler/cheaper solution that would meet your requirements.

-Matt