Signal Conditioning of Pressure Transducer

[tuyenhoang7]

I am using a pressure transducer by Honeywell and it's working well. Link to the datasheet: http://www.farnell.com/datasheets/1283210.pdf

I am using a 3K load resistor between the output pin and ground to measure the voltage. I've been using a multimeter (Fluke 187) and am able to very precisely set ambient pressure to 1.0000V and any change in pressure will result in a fairly accurate change in voltage. 

However, when inputting the output signal (voltage) to my DAQ (USB-6211), the measured output is very jumpy. It jumps between approximately 0.996V to 1.004V, instead staying at 1.000V. What kind of signal conditioning can I use (preferably within LabVIEW) to get a more stable output? Filtering, averaging, excitation, amplification? What is the best approach?

Thank you very much,

Tuyen

[pallen]

Seeing some amount of noise on a signal is completely normal. 

NI has great tutorial on signal conditioning here:  http://www.ni.com/pdf/features/sigcontut.pdf

Some specific info on pressure here:

http://www.ni.com/pressure/

In this case excitation and amplification seem to already be taken care of by your hardware. 

If you can pin your signal noise down to a specific source, ( example: 60Hz line noise ) then a dedicated filter might be the way to go.  But my first suggestion would be to try averaging.  

https://decibel.ni.com/content/docs/DOC-34469

[NIquist]

You're probably seeing the limits of digitizing an analog signal with a digital ADC.  If you're using full range (+/- 10V) that's 20 Volts / 16-bits.  20/65536 = .0003 Volts/ bit.  Limiting the range will enable some NI DAQ cards to put an analog amp in the chain which will allow you to realize better dynamic range but you will have to make sure your input signal doesn't exceed that range or it will clip.  See the manual for your 6211.  All of this (and more) is explained in the Analog Input section (ch 4).

EDIT:  As far as filtering to make the signal at least appear smoother I would suggest you collect data at a highr rate than you need, take every 10 (or so) measurements, average them and output the average.  This is called a Moving-Average Filter:  http://en.wikipedia.org/wiki/Moving_average