Voltage measurement with differential amplifier

[loamobn1]

I'm trying to measure the resistance of a part whose resistance varies by about 1 ohm, plus or minus (the average is about 5 ohms). So I have a voltage divider with the part in it (lets say the part is R1), and a reference voltage divider that has R1=5ohms. I'm trying to find the difference in voltage between the two signals, and in that way measure the resistance of the part. So I feed both voltages to an instrumentation amplifier http://en.wikipedia.org/wiki/Instrumentation_amplifier

This hasn't worked for me. I set all resistances to 1k, except the gain resistor (set to 500), and it doesnt work at all like it is supposed to. Does anyone have any ideas?

 

[shrtrnd]

One Ohm is a very low resistance to read accurately.
My input would be that the tolerances of your resistors is critical.
The accuracy of your meter is critical.
Yageo/Phycomp and Susumu make resistors in the .5% tolerance accuracy range.
Your meter must be calibrated, and is hopefully in the accuracy category of a
Hewlett Packard 3458A or Agilent 34401A.
I don't know if this is a work project or home brew, but you're going to need super
accuracy in your circuit and your measurement device to BE accurate.

 

[loamobn1]

As I matter of fact, I am using an Agilent 34401A . And the resistors I'm using are potentiometers that I've set by hand, using said multimeter. The thing is, I'm feeding the voltage into a (n equally accurate) DAQ system, so I need a voltage that varies linearly with resistance. So how do I map a change in resistance of +/- one ohm to a change in voltage of +/- 5V?

 

[duke37]

If you set up a vitual earth op-amp, feed it with a constant current and use your variable resistance for the feed back, then the output voltage will be proportional to the resistance.

With only 1 ohm you will not get much output voltage, so will need to include a post amplifier.

Your original post seemed as if you are using a Wheatstone bridge. How do you manage 1k resistors? You would be better with a 1 ohm reference resistor.

 

[loamobn1]

duke:
why make it more complex than it needs to be? Thanks for the suggestion, I made a negative feedback amp work just fine. Duh!
Thanks