This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

[FAQ] PGA900: Questions Regarding Calibration for the PGA900, PGA300, PGA302 and PGA305

Part Number: PGA900
Other Parts Discussed in Thread: PGA302, PGA300, , PGA305, PGA308, PGA309

Tool/software:

I have some general questions regarding calibration relating to the PGA900, PGA300, PGA302 and PGA305.  The main questions are as follows:

  • What is the general process of calibration and which combination of pressure/temperature is best?
  • Is there a production multical hardware/software available?
  • How do I calibrate without using the EVM GUI?
  • What is batch modeling and how do I make use of it?
  • How long should I soak the device when changing temperature points?
  • How do I calibrate if I’m not using the DAC?
  • See the following sections for the answers to the questions:

    What is the general process of calibration and which combination of pressure/temperature is best?

    The process of calibration is to take known points of pressure at specific points of known temperature.  The measured temperature and pressure points become the two independent variables for a desired, or dependent, DAC outcome.  If the sensor runs only at room temperature, then the calibration could be as simple as two pressure points at a single temperature point (2P1T).  In this case the DAC value is determined by the line slope formula y = mx + b, where in the polynomial equation only the h0 (offset and relates to ‘b’) and g0 (gain and relates to ‘m’) are the coefficients used.  If the sensor ranges through a variety of temperatures a better 3rd order polynomial fit can be used in a four pressure and four temperature (4P4T) calibration.  Here you would take ADC measurements for both the pressure and temperature and determine the desired DAC output for each of the cases.  In this use case (4P4T), 16 coefficients will be generated by linear regression to completely cover the 3rd order polynomial.

    Determining which method to use in practice is somewhat difficult to answer as this will be determined based on the system requirements and the accuracy the sensor needs to remain over the course of a range of pressures and temperatures.  Also, the more points taken the longer the calibration process will take.

    Is there a production multical hardware/software available?

    There have been a number of requests for a calibration hardware/software system for calibrating multiple PGA devices at the same time.  Unfortunately, TI does not have an available system for these devices that is similar to the analog PGA308/PGA309.  This means that the end user will need to develop their own system for calibrating multiple devices in a production environment.

    How do I calibrate without using the EVM GUI?

    This will depend somewhat on the PGA device being used and the communication interfaces available.  The PGA device to be calibrated requires communication to the PGA to be able to measure and determine the actual DAC values.  There must also be a sensor connected to establish a known pressure and the ability to read the results from the PGA.  Depending on the calibration type (3P1T, 2P2T, etc.) there may be a need to measure the temperature accurately and to determine the temperature ADC result by reading the PGA registers.  So, the big key is to have some ability to read and write to the PGA device and to set specific known parameters.  Once the data are collected, the linear regression process can take place and the coefficients computed and then stored to the EEPROM registers.  All of this will require some form of software developed to read/write register data and also be able to collect data as a part of the calibration process.  This is followed by the requirement to write the coefficient values to the device EEPROM.

    What is batch modeling and how do I make use of it?

    If you have a sensor that is tightly controlled in manufacturing so that over a batch build of the sensor the operating characteristics are similar from device to device, it is possible to calibrate a single device, and then program those same values of coefficients and operating parameters to every device used from the same batch.

    How long should I soak the device when changing temperature points?

    You need to soak the PGA device long enough so that the temperature has stabilized when reading temperature from the device.

    How do I calibrate if I’m not using the DAC?

    You would calibrate it in a similar way, even if you don’t use the DAC as a part of the result.  As the linear regression is 2 independent variables and 1 dependent variable, the linearization needs some measurable outcome.  For the PGA3xx devices the outcome is 2^14 bit DAC output and can be thought of in terms of codes instead of voltage or current output.  For the PGA900, you can program the firmware to deliver whatever outcome you would like digitally.