ADS1220 thermocouple measuring

Hi Fakhruddin,

When making temperature measurements with a thermocouple (TC), it is not just a single conversion from the voltage reading you get to a temperature. You must also consider the cold junction where the TC connection is made on your board and the TC that is formed at that junction.  Basic information for the TC application can be found in the ADS1220 datasheet starting on page 48.

For a good background on TC measurements, look at the ADS1118 application note:

http://www.ti.com/lit/pdf/sbaa189

Most everything in this application note will also apply to the ADS1220.  You may also find some of these TI Designs helpful as well:

http://www.ti.com/general/docs/refdesignsearchresults.tsp?keyword=ADS1220

Best regards,

Bob B

Hi Fakhruddin,

Your LSB equation is correct.  However it appears that you are not reading the data correctly.  You show a 32-bit value and the ADS1220 is a 24-bit device.  I'm guessing that you are not properly sign extending the result.  A value of all bytes high can mean that you are reading a code value of -1, or it may mean that the device is either not setup correctly or communication is not working.  CS must remain low throughout the entire communication transaction.  Both AVDD and DVDD supplies must be correctly applied for the device to function.

Can you send me your schematic?  I would also like to know the register settings that you are using, and if you have successfully read the configuration registers from the ADS1220 to verify that your communication is functioning.

Best regards,

Bob B

Hi Fakhruddin,

Are you able to verify that the device configuration is properly written?  Can you send me any scope or logic analyzer shots of the communication?  Is your circuit built on a PCB, or is it built using a protoboard?

One thing you can do is monitor the DRDY pin.  The default mode is single-shot, so if you have written the device for continuous conversion mode you should see that the DRDY pin pulses at the specified data rate of 20sps.

When you are attempting to read from the device are you reading directly (just sending 3 bytes of SCLK) or are you using the RDATA command?  One common mistake is CS is not held low throughout the entire communication transaction.  For example, you need to hold CS low for the entire length of the command and data when issuing the RDATA command.  Raising/lowering CS will cause the SPI bus to reset ending the communication.

If you do not see the DRDY pulsing but you believe you have correctly communicated to the device, make sure that the CLK pin has a good and solid connection to GND through the pull-down (you can tie the pin directly to GND).  If noise glitches on this pin, the internal oscillator will be shut down.  The noise will appear to the device that you are attempting to apply an external clock.

Best regards,

Bob B

Hi Fakhruddin,

Previously you had said that the output for DOUT is always high.  Did you ever read back the configuration registers to make sure that you did not have a communications issue by a possible bad connection on the DOUT line?  Can you send me scope/logic analyzer shots of the communication?

What are you using for an input?  Do you actually have a K type TC connected?

Here is a software link to a TI Design (TIDA-00189) that uses an ADS1220 to measure a TC.  You may find some of the code useful:

http://www.ti.com/lit/zip/tidc701

Best regards,

Bob B

Hi Fakhruddhin,

How are you measuring your cold junction?  It would be very helpful to see the PCB layout with respect to the cold junction, as it appears that there is some error in your measurement. 

Another question I have is with respect as how you are determining the 'standard' temperature.  How are you quantifying that you are measuring 5 deg C higher at room temperature?  I have the same question with respect to the water bath.  And how do you know that the water bath has even temperature throughout the bath?  To determine the effectiveness of your system, you should be using calibrated sources.  And you need to make sure that the cold junction is not the cause of the variations.

Best regards,

Bob B

Hi Fakhruddin,

Yes, you can use the ADS1220 for grounded K-type thermocouple.  A grounded TC is one that has the sheath connected to the junction of the TC wires.  Depending on how/what the sheath connects to there may be a ground loop or additional noise pickup that is created.

The cold junction measurement is very critical.  You may be reading a stable measurement by the internal temperature sensor of the ADS1220, but that does not mean you are reading the correct temperature at the cold junction, nor that the cold junction itself is stable.  The cold junction is the connection point of the TC and the PCB.  The QFN package will yield better results than the TSSOP version for the cold junction measurement due to the package construction and smaller package size.  Essentially you need to get the package itself within the cold junction.  If you take a look at the ADS1118EVM user's guide you can see where the optimal placement would be for the ADC in relation to the cold junction.  Unfortunately I do not have a good example for the ADS1220 at this time.

What I fear you are trying to do is to achieve a high precision system within a non-optimal prototype configuration.  You should first attempt to get a stable measurement.  This instability could be from external noise or from a varying cold junction.  You may also find it beneficial to add some gain once you get a more stable measurement.

Best regards,

Bob B

Hi Fakhruddin,

Yes, you can use the ADS1118 for the grounded TC measurement.  You can get more information on the web landing pages on TI.com.  Links are created automatically in this post for the ADS1118 and ADS1118EVM to get you to the appropriate pages, or you can search for the info using the web search tools.  There is also a TIPD109 reference design that goes through the design process.  There is also a Launchpad Booster Pack, 430BOOST-1118 that you may find interesting as well.

None of these designs specifically use a grounded TC.  The main point when connecting any sensor to an ADC is to maintain the proper common-mode.  For example, it is possible to have a TC (which is outputting a voltage in millivolts) to have a potential much greater than that if the sensor tip is grounded to a potential that is much different than the ADC ground.

So what is ground?  It is really a reference point.  When we hook two devices together we need to make sure that the reference point of both devices is at or near 0V potential where the common-mode voltage requirement is met.  When the grounds are common for the ADS1118, as well as the ADS1220 when the PGA is bypassed, you can use the grounded TC in unipolar supply for AVDD.  In the case of the ADS1220 when the PGA is not bypassed, the restrictions are a little greater due to the common-mode restrictions of the PGA.  This is where the bipolar supply operation is beneficial as the mid-supply point for AVDD is at analog ground.

Here is the real difficult part and that is you will still have the same issues with the ADS1118 as you do with the ADS1220.  Proper PCB layout is critical, a stable cold junction with proper measurement is critical and external noise (potential ground loop and EMI/RFI) can add to the problems.

Best regards,

Bob B

Hi Fakhruddin,

You do not need to use 0402 sized devices.  The layout showed that an efficient layout can be achieved within the cold junction area using these devices.  Depending on your design the layout and devices used could be radically different.  Obviously you can squeeze a lot more devices into a smaller area when using 0402 sized devices, but they are a lot harder to hand solder.

Best regards,

Bob B

Hi Fakhruddin,

TIPD109 shows the temperature sensor placement relative to the cold junction.  This board is very small so unfortunately it doesn't show the thermal isolation.  There is an ADS1118 BoosterPack (430boost-ads1118) that shows a thermally isolated area while using the ADS1118.

What is tricky is that you have two connectors and are trying to measure the cold junction for both TCs.  In the end you need to make sure that the cold junction area remains thermally stable across the connections.  Then the temperature sensor must be placed within the same area as the cold junction so that the TCs temperature is properly compensated.  You must take into account any airflow that may move across your PCB that can change the temperature across the connections of the TC.

Best regards,

Bob B