How to avoid measurement drift

[REDJ]

I used the NI 9205 to measure the voltage signal of the ICP accelerometer. This voltage signal has been modulated by the modulator, but there is a bias voltage. The output interface of the modulator is the BNC port. The outer casing of the accelerometer is metal and is mounted on a metal table. The modulator housing is plastic and placed on a metal table. The metal table is insulated from the ground. I used a shielded cable to connect the BNC port, and connected the core to AI+, the shield to AI-, and set the sampling method to the differential in vi. But the voltage signal I collected drifts severely. I have consulted some technicians. They said that the drift may be caused by poor grounding. But I don't know how to solve this problem, I want to ask for some help.

[croohcifer]

The 9205 provides no ICP current excitation or AC coupling to remove the bias voltage.  NI makes many devices suitable for directly interfacing with IEPE/ICP sensors (see NI 9232, 9234, 9218, etc.) 

If you want to use the 9205, then you should contact your sensor's manufacturer to see if they sell an ICP signal conditioner that will handle the current excitation and AC coupling.  Most vendors do.  These conditioners will usually have a BNC output that has the AC signal content centered around 0 V so you can acquire it with a general purpose DAQ like the 9205.

[REDJ]

Thank for your reply. I used the NI 9205 to collect the voltage signal from the amplifier. The amplifier provides ICP current excitation to the accelerometer and modulates the signal. The signal output from the amplifier BNC is a standard ±5v voltage signal. This signal is not drifted when acquired by the NI 9234 or the oscilloscope. However, the dynamic performance of the NI 9234 does not meet the requirements(0.1~0.5Hz acceleration signal).

[croohcifer · Accepted solution]

Sorry, I wasn't aware you were already using a signal conditioner.  I assume that's what you mean by "modulator".  For clarity, there is no modulating going on.  Your signal conditioner box is likely just a charge amplifier, low pass filter, and AC coupler.  No modulators involved, so I would avoid that term in this case.

---

@REDJ wrote:
This voltage signal has been modulated by the modulator, but there is a bias voltage.

---

I've never come across an ICP signal conditioner that leaves a significant bias voltage, so that sounds suspect to me.  Are you sure there is a bias voltage, or are you guessing that some of the "drift" you're seeing could be DC bias from the signal conditioner?  I would check the spec sheet for the signal conditioner to see if it specifies a residual DC bias or something to that effect.  I'd be surprised if it were more than 10 mV or so.

---

@REDJ wrote:
The modulator housing is plastic and placed on a metal table.

---

This doesn't tell us much since the way the signal is grounded has to do with the electronics inside the signal conditioner.  Those electronics could easily be isolated from a metal housing, so the material really isn't important.  Again, the spec sheet should tell you how the output is grounded.  If the output is isolated from measurement ground, then you should measure the signal with a single ended configuration or measure differentially with bias resistors to ground.  This document should help you out.  Table 1 is a good starting point:  http://www.ni.com/product-documentation/3344/en/

[REDJ]

Thank for your reply.  From the measurement results of the oscilloscope, the voltage signal has an output bias of more than 100mv. It seems that I should check if the signal conditioner(Model: PCB 483C05 https://www.pcb.com/products?m=483c05) is broken or not set correctly.  Thank for your constructive suggestions.