# Gain and offset adjustment with voltage reference?

I am building an instrument that relies on a precision sensor module. The sensor is factory-calibrated, and outputs 0-10Vdc. I believe the factory calibration to be better than anything I can attain, and thus do not want to create my own calibration curve for the sensor.

I want to read this output with a 16-bit ADC; however, most ADCs only accept 0-3Vdc input. I will likely need an offset/gain op amp circuit between the ADC and the sensor, but I want this to be referenced to a known voltage reference. For example, I also want to run a known precision voltage reference through the offset/gain circuit to known that a given reading from the ADC corresponds exactly to a given reading on the sensor independent of temperature and everything else that could affect the reading. I want to be sure that I can determine the exact voltage coming off the sensor.

Can someone provide a schematic for such a circuit? Or at least point me to such a resource.

Thank you,

Michael

EDIT:

A number of commenters have asked about the specifics of my application (sorry! it took me a bit to dig up the datasheet). I am interfacing with an INFICON Priani Capacitance Diaphragm Gauge (a vacuum sensor). The device has a signal output with an absolute maximum range of 0-10Vdc; however actual measurements stay within the 2.2-8.68Vdc range. The accuracy of the voltage output is not stated because the device operates on a logarithmic scale. Voltage is directly proportional to the logarithm of pressure. The data sheet states the pressure measurement (not voltage) is ±15% accurate at 50 mbar, ±5% at 1000 mbar, and ±2% accurate at atmospheric pressure. The conversion factor is P = 10^(U-5.5) where P is pressure in mbar, and U is voltage.

The logarithmic output of the sensor is why I am concerned about accuracy (not so much precision). Errors in voltage measurement are amplified exponentially by the logarithmic nature of the device.

• What temperature range are you expecting your device to be accurate over? How accurate/precise is good enough? 1%? 0.1%? What bandwidth do you need? Aug 12, 2015 at 23:16
• "everything else that could affect the reading" - sure you can run a precision voltage reference thru the ADC but that only gives you a comparison at around that voltage. The integral non-linearity of the ADC makes errors elsewhere. May I suggest that you start putting numbers of what is acceptable to your question because at the moment it is unanswerable. Aug 13, 2015 at 7:50

Off course you always bump into a solution immediately after posting a question. Below is a TI Application Report explaining how to do exactly what I asked.

http://www.ti.com/lit/an/sloa097/sloa097.pdf

Hopefully, this question will at least be useful to others.

Sorry!

Michael

• What you experienced is called "confessional debugging" among coders - a truly powerful technique. Aug 12, 2015 at 21:45
• After working through the Application Report a little more, I noticed that they do not have a application for situations in which the gain in between 0 and 1 (as it is with my case), so I guess the question is not yet "fully answered". Aug 12, 2015 at 22:10
• I believe the Application Report will work with gains of less than 1. The value of m will be less than 1 but that will affect the resistor values to allow such a gain. Aug 12, 2015 at 23:55
• I guess I misunderstood the question then. I thought you were asking specifically about how you might verify that your gain/offset circuit was working correctly; i.e., not adding additional error terms of its own to the sensor's measurement. The paper your cite does not address this aspect at all. Aug 13, 2015 at 0:56
• That document does nothing about temperature drift - this is the biggest worry. I'd say the document was useless. Aug 13, 2015 at 7:53

If your sensor's transfer function is well enough defined so that its output is more accurate than the granularity of the ADC, then all you'll need is a voltage divider between the sensor's output and the input of the ADC to transform the sensor's output from zero to ten volts to zero to three volts.

• But I then need to calibrate the voltage divider given that the best resistors offer 0.1% tolerance. Additionally, that circuit isn't referenced to a known precision voltage source. So I may be able to calculate that 2.5V on the input of the divder equals xV on the output, but that's not a measured value. Aug 13, 2015 at 14:07
• @MichaelMolter: Nonsense. 0.001% fixed resistors are routinely available, and voltage divider networks down to about 0.005% aren't hard to find. What you do have to be careful about, though, is the load the divider will put on the sensor and the load the ADC will put on the divider. Why don't you post some spec's and data sheets for the stuff you'll be using so we'll know what you're talking about? Like Andy aka asked for... Aug 13, 2015 at 16:55