ni9225 to measure 600VDC signals

[LionelMRI]

Hi, I have a DAQmx unit with a NI9225 card. This card has three isolated inputs rated for 300Vrms each, but I need to measure a voltage bias of tens of millivolts from a line that can deliver high transient voltages (up to 600V during a couple of seconds). I am wondering if I could use two of the ports on the NI9225 to split the signal into two, so that the module could be used for the job. This is a risky operation so I would like to make sure this is a robust solution before I try it. Lionel

[Buechsenschuetz]

I do not really have experience with a set-up like you are trying to do. Obviously you want to feed the input signal to two inputs connected in series. The problem is that to get an even voltage distribution the input impedances have to be closely matched, too. Input impedances usually have rather large tolerances so there is risk that one input will receive more than one half of the 600V. 

I know that this will reduce the signal to be measured but I think you are more on the safe side using a voltage divider for the input signal. 

[LionelMRI]

Thanks for the answer. I did not think about the tolerances on the input impedance, that is a very good point. A voltage divider may be a solution, there should be room to halve the voltage and still get enough precision to measure the small signal I want.

[JamesBeams]

The 9225 isolation is rated at 300Vrms ch-gnd. So inputting a 600V signal to one channel would violate this safety spec. There is a 600V ch-ch spec, but this covers the case where one channel is 300V and a neighboring channel is -300V. You could consider using a 9244 instead.

[Henrik_Volkers]

If you want to measure some XX mV  and have HV spikes you don't need to measure: Build a input protection circuit!

Depending on the dynamic range of the mV range needed (Assuming nearly DC), I would think of something like: Input current limiter 2 or 3 resistors (they usually have a 200V rating, but see spec of the resistors choosen!) building a 100k Ohm. Followed by a pair of anti-parallel diodes to V_low/GND , that will limit the voltages to ~700mV  with a acceptable linear region up to ~ 200 mV (depending on the diodes 😉 ). If you don't want the V_low connected to GND: A neon bulb will limit the voltage to 90V (ingnition, later about 40-50V while glowing :D) an additional current liming resistor in the GND sens line isn't a bad idea.

The 100k input limiter  to the 1M input impedance will matter: so an individual calibration with a voltage source, (to get mV use a low resistor voltage divider) and a second 9225 channel without protection circuit is quickly done.

[SteveD123]

One final detail to Henrik's solution to your problem: Put a pair of fuses, one on each wire, as close to the point where your instrumentation circuit meets your 600V circuit. That way, if there is an accidental short circuit in your instrumentation wiring during a 600V event, you will not cause a fire.

[LionelMRI]

Very good point indeed, thanks!