0.01 accuracy 4 wire PT100 measurement

What temperature range do you want to measure the temperature over?  Do you prefer a digital output, or do you want an analog voltage?

Best regards,

Art

The range is from 0 to 100. Prefer analog voltage, if digital output can get 0.01 accuracy, also can.

Alex,

0.01C accuracy on 100C equates to 0.01%.  This is a very high accuracy requirement and you will have to take much care to make sure that you achieve this accuracy.  Below is a short list of suggestions / comments that should help you to get started.

1. The best pure analog RTD to Vout device we have is the XTR108.  However, I don’t think it’s a good choice to achieve 0.01% accuracy.  It was designed to achieve better then 0.1%.  If you are willing to make some compromise you may be able to use this device for your application.  I think 0.05% is probably more realistic for this device.
2. If you actually require 0.01% I recommend you use the ADS1248 (digital output only).  Attached is an App Note that should help to get you started.
   1. To achieve the accuracy level you require you will need to calibrate offset and gain error out on the ACS1248.  This can be done during the manufacturing process by connecting precision (0.001%) resistors in place of the RTD for 0C and 100C calibration points.
   2. You need to use a precision low drift resistor for Rbias (see figure 4 of the sbaa180 App Note).  Choose the lowest drift most precise resistor that will work for your product budget.  A 0.1% 10ppm resistor may work if you calibrate the system as described in a.
   3. If the ADS1248 temperature drifts substantially you may have to compensate for this.  The ADS1248 has and on-board temperature sensor.
   4. Standard RTD’s do not have 0.01C accuracy.  You will have to calibrate the RTD.  Many systems of this accuracy level have a chip included on the plug that connects to your electronics.  The chip has calibration constants that are downloaded at the start of calibration.  Note: this calibration is separate from the calibration described in “a”.  Both the electronics and RTD need two separate calibrations.

http://focus.ti.com/general/docs/litabsmultiplefilelist.tsp?literatureNumber=sbaa180

Best regards,

Art

Many thank.

We may use the ADS1248. Is there any low power cpu can work with it together.

Also is it possible to suggestion some chips can provide accurate current or voltage source, which will be used in 4 wire measurement.

For the bridge type measurement, we need a voltage source, can we use TLV1117.

Actually we more prefer current source, but cannot find fit one.

Hi Alex,

MSP430 is the low power MCU, it has been widely use in temp measure field. ADS1248 has two precision current source, you can check it in its datasheet.

Rgs,

Jianhui

The accuracy of the current source in ADS1248 is only 6%(initial). Does I need to measure the current by measure a resistor's voltage. Does the default circuit of ADS1248(4 wire) can get 0.01degree accuracy, or any improved circuit is needed, thanks.

Hi Alex,

Your questions are remarkably similar to those of Rui Li in the Precision Data Converter forum:

http://e2e.ti.com/support/data_converters/precision_data_converters/f/73/t/116052.aspx

Are you by chance working together?  If so, can we consolidate your questions over in the Data Converter forum?

Alex;

Before going too far in your circuit design, make sure that there are RTDs available on the market that will meet your accuracy requirements. Even a Class A RTD cannot meet your 0.01% accuracy goal. See : http://www.thermometricscorp.com/acstan.html

To all,

This is a scientific level accuracy requirement.  Any system with this level of accuracy will require calibration:

1. The RTD itself will need to be calibrated.  Typically the calibration constants are stored in a EEPROM dongle that is at the end of the RTD.
2. The measurement system (A/D + reference) will also need to be calibrated.  An offset + gain calibration is probably sufficient.
3. Look at temperature drift, and non-linearity as the limiting factors.
4. Also look at long term shift.

Art