Scaling a +/-5V Signal to 200mV for a Torque Sensor

[zekester]

i am taking a +/-5 voltage signal is supposed coming from a torque sensor. this is supposed to be equal to 100 N*m. However, -5.04 volts is actually equal to 100 N*m and 4.85 volts is actually equal to 100 N*m. I need to scale this signal to a maximum input of 200 mV for display on a panel meter. I was just going to use a voltage divider going into a noninverting op-amp, but I am not sure what to do now that I know the negative and positive voltage signals don't match, I was told a differential amplifier would work for this, however, I am not quite sure of how to accomplish this, any help would be appreciated.

 

[waht]

A differential amp could work but is not required. Depending how much current you have available from the sensor and where you are connecting it, a simple two resistor voltage divider would suffice.

If you want to use an op-amp, then a non-inverting amp configuration could follow the voltage divider network. But you would need a positive and a negative power supply connection for the IC.

 

[Integral]

Many panel display units have built in scalling features. I am familiar with the Omron http://www.omron247.com/Industrial-Automation/Store/OmronFamilyView_10051_10051_-1_11921" [Broken] It allows for a 2 point linear scaling. This may be a more documentable solution.

 

[antoker]

Most of the sensors provide an output that is offseted by Vcc/2. Is your sensor based on the same method?

Measuring real life data is not an easy task to accomplish. I've been dealing with this kind of circuits for a couple of years now, so here is my advice for best results:

1. Good power supply with stable output voltage! (output voltage variations are allways caused by unstable powersupply) Hense, use some stabilizing circuits.

2. DECOUPLING DECOUPLING DECOUPLING ! ! ! Place decoupling capacitors as close to the sensor IC as possible, avoid ground layer around the IC etc. Short distances between the circuits = GOOD! Long distances = BAD (remember transmission line theory) parasitics etc. etc.

3. When dealing with unmatched output (i.e 5,04 & 4,85V) you can either offset (by applying the dc to the output) the input (so the output will swing between $$\frac{5,04+4,85}{2}$$) or you can use microcontroller or some other means to do so. Just be shure that those unmatched swings are not caused by the way you've configured the IC.

4. Use 1% resistors, if your circuit is modeled correctly and psu variations has nothing to do with unmatched output, then problem can be that you're using >=5% resistors. When modelling sensor-circuits allways use 1% tolerance resistors, such as 1206 or lower series.

Now, for for scaling the output voltage down, I would suggest a divider network followed by a low-distortion buffer amplifier so the circuits after buffer would not load the divider, hence not affecting your read-outs.