Wheatstone bridge sensor amplifier

[Rachie5272]

I'm trying to build an interface for Wheatstone bridge pressure sensor.  I want to output a signal of 1 mV/1 mmHg.  I think this would be as simple as a single op-amp, but I'm not entirely sure how to set the gain.  The sensor datasheet is rather sparse, and I can't quite figure out what voltage corresponds to what pressure.  Can anyone help me with this circuit?

Here are the specs on the pressure sensor:

Operating Pressure Range: -50 to + 300 mmHg
Sensitivity: 5?V/V/mmHg, ± 1%
Zero Offset: 0 ± 25 mmHg
Non-linearity and Hysteresis:
0 to 100 mmHg : ± 1 mmHg
101 to 200 mmHg : ± 1.5%
201 to 300 mmHg : ± 2.0%

Zero Temperature Coefficient: 0.3 mmHg per degree C
Zero Drift: <2 mmHg; 8 hour drift after warmup
Defibrillation Withstands: 5 x 360 Joules
Leakage Current: <10 ?A at 120 Vac RMS 60 Hz
Volume Displacement: < .03 mm at 100 mmHg
Natural Frequency: >200 Hz in 0.9% saline
Over pressure Protection: -400 to 6000 mmHg
Operating Temperature: 15oC to 40oC
Output Impedance: 300 ? ± 10%
Input Impedance:
1800 to 4500 ? (pressure sensor only)
350 ? ± 10% with MER300 series cable

Excitation Voltage and Frequency: 2.5 to 10 Vdc or Vac RMS to 5 KHz
Light Sensitivity: < 5 mmHg
Phase Shift: < 5o (0 to 5 Khz)
Storage Temperature: -25oC to 70oC
Cal Factor: 100 ± 2.5% mmHg
Maximum Shock Acceleration: Withstands shock of 5 falls from 2 meters onto a concrete floor

[Kleinstein]

One could use a simgle OP amplifier however in this case the amplification will depend on the sensor output impedance (or introduce extra noise due to an extra series resisistor). Usually is would be better to use an instrumentation amplifier IC. Amplification is usually set with an extrernal resistor in this case.

Careful to much Hg - think RoHs.

[Andy Watson]

... but I'm not entirely sure how to set the gain.

Sensitivity: 5?V/V/mmHg, ± 1%

Assuming that question mark is this forum's method of representing a mu:
That reads as the sensitivity of the bridge is 5 microvolts per Volt of bridge excitation for every  mm of mercury pressure. i.e. to get 1mV per mmHg you would need a gain of 200 with a bridge excitation of 1V, or a gain of 40 with 5V bridge excitation etc.. Use the maximum bridge excitation possible within the bounds of not releasing the magic smoke or causing thermal drift problems.

[Rachie5272]

I think I have the basic idea now.  I'm not quite sure about a couple things though.  First is the reference pin of the amplifier.  Is a simple 50% voltage divider from the power rail good enough, or do I need a separate precision reference IC?

Next is input protection.  It's probably not best to have the power rail tied directly to the bridge excitation voltage, but what would be a good way to limit the current?  It seems like a series resistor would throw off the measurements.

Lastly, what about capacitive decoupling?  I took care of this for the power rail, but what about the inputs and outputs?  How do I know when I need extra capacitors?

[babysitter]

Sounds like a pressure transducer of the medical kind, for extracorporal blood pressure measurement acc. DIN EN ISO 60601-2-34 or AAMI BP22:1994 for those in FDA land?

[Rachie5272]

Yep, that's exactly what it is.  Unfortunately, those standards are behind heavy paywalls.  They can't have just anyone making things safe, can they?

[babysitter]

I would advise to get a instrumentation amp like AD623 or AD620 or equivalent from your favourite supplier, attach a gain setting resistor that matches your 1 mV/mmHg requirement (the rule 5µV/mmHg per volt excitation voltage is right) and, uhm, prepare offset injection as there is both a manufacture-related offset (0mmHg output different from 0µV) and also some temperature coefficient, and you also see a different zero signal depending on the column of water above your sensor. Patient monitor manufacturers in FDA land are required to give or take 150 mmHg to compensate for offset iirc.

Please use a rather stable excitation voltage source capable of some mA, and remember your signal is proportional to your excitation voltage, that means, ratiometric. You should at least be able to tell what excitation voltage is there if you want to tell the actual pressure.