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.

ADS1220: Problem in Thermocouple and RTD signal transmitter Design

Part Number: ADS1220
Other Parts Discussed in Thread: ADS122U04EVM, ADS122U04, ADS1247, ADS124S08EVM, ADS124S08, ADS124S06, ADS1120, TIDA-00018, TIDA-00189, ADS1118, ADS1118EVM, MUX36D04, ADS114S08

Hi all,

I am trying to combine the designs in page 49 and 52 in ADS1220 Datasheet, to make one module to handle 2 Thermocouples or 1 RTD on its inputs.

The design specs are:

1. Unipolar 3.3 Volts supply.

2. Internal reference used for thermocouple and external 1 KOhm resistor used as Rref used for RTD.

3. Fc-Diff = 51 Hz and Fc cm = 1085 Hz

The two thermocouple inputs are TC1(Ain0,Ain1), TC2(Ain2,Ain3) or 1 RTD input.

The thermocouple input needs to be biased and filtered in this case I will have 4 filters on all inputs and bias resistors, the problem is when using the module as 1 RTD input,

the bias resistors and filters will remain in the signal path, I am not sure what this will cause! I don't want to use analog switches to avoid making the design more complicated and adding some extra noise.

I am thinking of having a bipolar supply to take the bias resistors off the design then, only RC filter will remain in current path when using RTD. Will this filter badly degrade the RTD measurement?

I could not see any reference design from TI using bipolar supply, is there any disadvantage of using bipolar supply?

One last point, if the used thermocouple was grounded to the machine. will the unipolar supply with bias resistors work as it should?

Regards,

Mahmoud

  • Hi Mahmoud,

    Welcome to the forum!  I will make a few assumptions and you can correct me if I'm wrong.  You want to use two TCs or one 3-wire RTD using the same circuit with no jumpers or switches.  You really will find this difficult to achieve without some external jumpers or switches.  Notice Figure 77 on page 52 of the ADS1220 datasheet.  Note how the current sources connect directly to the RTD.  If you drive the current through a series resistor you will quickly run into issues with drift and compliance voltage for the IDAC sources as there will be a voltage drop across the input resistors connected to the current source outputs.

    Depending on the operational temperature range, you may also see some leakage on AIN2 and AIN3 that may degrade your accuracy when connecting a TC to those inputs.

    For your RTD you may want to consider using a high-side reference and calculating your lead resistance and then subtracting from the measurement.  This is demonstrated in the ADS122U04EVM user's guide in figure 16. The ADS122U04 is a similar device to the ADS1220 except the communications interface is UART instead of SPI and the leakage is also less on AIN3 as the low-side switch is removed.

    For your TC measurements you will need to compensate the cold-junction where the inputs connect to the PCB.  You can use the internal temperature sensor of the ADS1220, but routing becomes critical when trying to the place ADS1220 in the isolated thermal area for the inputs.

    You may want to take a look at the ADS124S08 (or ADS124S06).  The ADS124S08EVM user's guide also has some example connections on how to connect various inputs to the device.  Another device similar to the ADS1220 is the ADS1247.  The advantages of these devices are an included VBIAS for biasing a TC at mid-AVDD supply as well as self offset calibration.

    All of these devices can operate with bipolar supply with no degradation as compared to the unipolar supply.  In all cases you would want to use a linear supply such as an LDO as opposed to a switching power supply to avoid additional noise.  If a switcher must be used, then avoid switching frequencies that are multiples of the ADC modulator rate, and then use an LDO to regulate to the final voltage.

    If the TC is grounded to the machine, you must make sure that the TC output is within the operational range of the ADC input.  You cannot assume that the machine ground is the same voltage potential as the ADC ground.  For the ADS1220, if the grounds are the same potential you will either need to use the bipolar supply mode if you want to use a gain greater than 4 or use PGA bypass mode (maximum gain of 4) when using a unipolar supply.

    Best regards,

    Bob B

  • Hi Bob,

    Thanks for your detailed answer.

    Please find the attached file showing the design that I was thinking about, the red lines show the RTD signal path and the blue is the TC. Let’s assume that I will make one design for 2 TCs and another one for 1 RTD.

    In the RTD design, I think the TVS diodes, Schottky diodes and resistors still needed to protect the inputs of the ADC from wrong wiring, so only the capacitors will be removed from my design. Will that give a bad accuracy? I am looking for a 12-13 noise free bits, and I may use ADS1120 instead if ADS1220 since I don’t need a high resolution in my system.

    My plane is to use the internal temperature sensor for the cold-junction compensation and I will keep it as close as possible to the terminal.

    Regarding the grounded TC, if the grounds the same potential, will Rbias keep the TC output in the acceptable range for the ADC and PGA(I am using 32 gain)? If not, is there any way to keep it within that range other than using bipolar supply!

    And if the machine ground and ADC ground are not at the same potential, how to solve this issue, I know the topic is getting much wider, but can you suggest me a solution or recommend something to read.

    Finally, I used to work on an FPGA based digital designs, but I am trying to get my hands on Analog designs too.

     

    Regards,

    Mahmoud

  • Hi Mahmoud,

    Let's discuss the RTD circuit first.  First you need to complete the current path so R7 (reference resistor) needs to connect to AGND.  To limit the error, you would need to remove R11 and R12 (TC bias resistors) and change the value of the resistors for R3 and R4.  Even though the current through R11, and R12 is small, they will add drift and leakage paths.  The reason R3 and R4 require a lower value is based on the total voltage drop for IDAC operational compliance.  The maximum voltage drop to AVSS from the IDAC output is AVDD - 0.9V which equals 2.4V for your circuit.  You will need to use at least the 500uA settings for the IDAC outputs which will allow a 1mA total current through R7 creating 1.2V reference voltage.  You cannot use a lower setting as this will not provide enough current for the minimum external reference voltage of 0.75V.  The worst case total voltage drop will be the IDAC supplied from AIN3 which will be 500uA(R4 + RLEAD1 + RTD1 + 2(RLEAD3 + R4) and this voltage drop will exceed 2.4V.

    I will assume that 2.94k input resistors are used to current limit an overvoltage condition which is part of your input protection circuit.  I have done quit a bit of analysis on input protection circuits in the past and need to write an application note at some point.  What you have in your schematic is actually very good, but consider the effects of leakage.  I mentioned in my previous post the potential error from leakage on AIN2 and AIN3.  This device leakage is demonstrated in the typical characteristics graphs in the ADS1220 (ADS1120) datasheet.  What I have discovered in my studies of input protection is that there are a limited number of devices with low leakage for both TVS and Schottky diodes.  Many of these diodes can leak as much as 10uA or more so you want to find low leakage diodes that will work in your circuit to reduce the amount of error caused by the leakage.  You will need to calculate the amount of error caused by leakage to determine if the leakage will be greater than your desired resolution.

    As for the bias resistors for the TC, these serve the purpose of both setting the common-mode to mid-AVDD supply as well as providing a simple means for open circuit detection without using the burn-out current sources.  The negative impact is the current through the TC will add some self-heating.  You will also need to consider the analog settling delay of the input capacitors relative to the time when the TC is connected.

    If the TC is grounded to the same ground potential of the ADC, then with a unipolar supply you must use PGA bypass mode which limits you to a gain of 4 to maintain the proper input range for the ADC.  This is where it may be helpful to use the ADS1220 to increase the resolution.  If the TC and ADC are at different potentials, then the biasing resistors will allow the correct common-mode for the ADC input.

    Regarding general information on TC topics, you can use your favorite search engine.  Some of the best information comes from TC manufacturers.  TI has some application notes such as SBAA134 and SBAA189.  You may also find the TI Designs TIDA-00018 and TIDA-00189 helpful for showing the input circuitry and possible layouts for TCs.

    Best regards,

    Bob B

  • Hi Bob,

    I will focus on the TC for now, my plan is to use isolated DC-DC converter to supply the ADC with unipolar supply(3.3V),

    and I will use two bias resistors to set the common-mode to mid AVDD.

    In this case the design should work with grounded and ungrounded TCs as well?

    Regarding TVS diodes I found one device with low current leakage and I am planning to use it (PTVS30VP1UP).

    Regards,

    Mahmoud

  • Hi Mahmoud,

    Isolating the power may not be enough as the digital communication path may present an issue if not isolated as well.  You might want to also take a look at the ADS122U04 which is very much like the ADS1220 except with a 2-wire UART interface for isolation applications.

    The TVS diode you discussed appears to be a reasonable choice.

    Best regards,

    Bob B

  • Hi Bob,

     

    Thanks for your help.

     

    Actually, I forgot to mention about the digital communication path isolation, I will use an isolator, so the ADC will not be sharing the same power supply with the uC.

    In order to avoid the input current leakage in ADS1120, I plan use ADS1118 for 2 TC inputs, but the smallest input range is +/-256 mV and for most TCs the input range is no more +/-128 mV, I am not sure if adding an instrumental amp before the ADC is a good idea?

     

    Regards,

    Mahmoud Z

  • Hi Mahmoud,

    Adding an INA will only help if the added noise/error will increase the benefit of the slight gain increase.  What temperature resolution are you hoping to achieve?  Are you considering two INAs with the ADS1118?  You could also use the ADS1120 with a buffer in front of the AIN2/AIN3.  You need to consider external noise (you don't want to just gain the noise) and any error related to the cold junction measurement.  You may find that the effort isn't worthwhile in the end by adding the additional external components.

    Best regards,

    Bob B

  • Hi Bob,

    I am hoping to achieve 0.5 or 0.25 accuracy for K TC, the voltage output range for the K is approximately (–6.5 to +55) mV, and the full temperature range of the K which is (-270 to +1370) .

    If I am using the ADS1118 with +-256 mV gain setup, I need to get 15 bit noise free conversation result to achieve 0.5 .

    My calculations are as follows, please correct me if I am wrong:

    TC K total voltage output range 61.5 mV

    TC K Total temp range 1640

    The range of 15 bit is 0-32767

    Only 1/8 of the ADC input range is used <=> 512 mV/61.5 mV = 8.3

    So, the used range of the ADC conversation is 32767 / 8 = 4095.

    The used ADC range needs to cover the whole temperature range 4095 / 1640 ≈ 2.5

    Regarding to using a buffer in front of the ADS1120, will this eliminate the leakage current?

    Regards,

    Mahmoud

  • Hi Mahmoud,

    A 0.25 degree C accuracy is going to be very difficult and in fact 0.5 degree is a difficult challenge. You need to consider all of your sources of error which will include offset, gain, drift and cold-junction measurement error.  Also, there is accuracy error of the TC itself.  The TC error maybe 1.5 deg C or more.  Due to all of these sources of error you will need to perform some type of calibration to reduce the effect of these errors.

    There are a couple of TI Designs that you may find helpful.  One relates to the ADS1118 and the other to the ADS1220.  The ADS1220 TI Design has a lot of information regarding the specific sources of error and comparison related to calibrated and uncalibrated error.

    http://www.ti.com/lit/ug/slau509/slau509.pdf

    http://www.ti.com/lit/ug/tidua11a/tidua11a.pdf

    I believe one of the biggest challenges will be to accurately measure the cold-junction temperature.  This will depend greatly on how well the TC input is thermally isolated from other heat sources and how accurately the sensor can measure the cold-junction.

    Best regards,

    Bob B

  • Hi Bob,

    I built a prototype of the TC transmitter by using ADS1118EVM board with external uC board.
    I am using NIST 10th order polynomials to convert the mV to C° and C° to mV. And I used ADS1118 internal temperature sensor to do the CJC.

    I would like to ask about the calibration needed for the system, I believe it can be divided into two main types:
    1. Circuit related calibration.
    2. TC related calibration.

    I checked the tidua11a document (Page 10) for Error estimation and I could not figure out, where some values came from.
    In the attached screenshot, are the yellow coloured values taken from ADS1118 data sheet? I tried to check page ‘6’ on datasheet and it does not seem to match.



    I plan to cancel the offset and drift by chopping the ADC inputs (taking two readings for the inputs, one normal polarity and other with reversed polarity) then averaging the two readings.

    Could you suggest me what else I need to calibrate in the circuit?

    Is there any need to calibrate each produced circuit individually or the calibration need to be done only on one circuit, and the result should apply to all other with the same components?

    Finally, Regarding TC calibration, is it just taking several points of the whole temp range and adjust the output regarding those points?

    Regards,
    Mahmoud

  • Hi Mahmoud,

    As you are using the ADS1220 error information relative to the error calculations, you need to look specifically at the ADS1118.  The information will differ as the devices differ with respect to the reference and PGA.  The devices also differ with respect to the total number of bits.  Many of the specifications for the ADS1118 are in terms of LSBs.

    You may want to take a look at the graphs in the datasheet and extrapolate values from the graphs.  If you look at the total error graph in Figure 2, you will notice that the dominate error is gain error.  Based on experience I would say that gain error, and cold junction temperature error will be your dominating errors relative to the measurement system.

    A thermocouple has two major issues.  One is there is a significant amount of error relative to the sensor itself relative to temperature.  The temperature accuracy can easily vary 1.5 deg C or more.  The second issue is the thermocouple is a non-linear device.  When calibrating you want to have a stable cold junction (or as stable as possible) and enough measurements to recreate a piece-wise linear correction factor.

    Best regards,

    Bob B

  • Hi Bob,

     

    I am sorry, I was checking the ADS1118 datasheet and the tidua11a is about different IC.

     

    I prefer to use ADS1120 instead of ADS1118 for 2 thermocouples input, but I am worried about the current leakage in the AIN2 and AIN3 channels.

    The reason why I prefer ADS1120 because it has more flexible gain and mux, also it can be used in future for RTD or Load cell transmitter.

     

    I am trying to calculate the error caused by the current leakage on the ADS1120 Ain2-Ain3 input, could you correct me if my calculations are wrong.

     

    if the input filter uses 2.94 K resistor, input voltage is 2 V, and the temperature is 85 C

    The Approx.. leakage on AIN2 = -15 nA and AIN3 = -30 nA (ADS1120 datasheet Figure 15)

    Voltage drop is:

                    Vain2 = IR = 15nA * 2.94K = 44.1 uV

                    Vain3 = IR = 30nA * 2.94K = 88.2 uV

     

    The LSB(ADS1120) = 2.048 / ((2^16)-1) = = 31.25 uV

     

    In this case I am losing 3 LSB on AIN3?

     

    1.Do I need to consider the input impedance of the ADC? Or these calculations good enough?

     

    2. Can I use a software calibration for the leakage current? Since I have internal temp sensor in the ADC so I could use the temp reading to add or sub the voltage drop value caused by leakage current?

     

    Regards,

    Mahmoud

  • Hi Mahmoud,

    If you bias the TC with a pull up and pull down resistor you essentially have a current divider at the TC input connections.  If you could apply an external short, you could calculate the offset created by the leakage.  When attempting to calibrate, each device will behave differently and the leakage effect is non-linear over temperature.  It may be possible to use the internal temperature sensor to help correct offset and gain drift errors (something I've wanted to try but have never had the time) but the leakage is more difficult as there are both absolute leakage as well as differential leakage differences that are dynamic with voltage and temperature.

    As you initially calculated, the error is substantial using resistors of 2.94k.  The actual number of codes of error will depend on the gain being applied as gain directly relates to the full-scale range of the ADS1120.

    The ADS122U04 is nearly an identical device (except it is only currently available as a 24-bit ADC) but with a UART interface instead of SPI.  The low-side switch has been removed and some other leakage improvements were also made.  This may be a much more suitable device.

    Best regards,

    Bob B

  • Hi Bob,

    The ADS122U04 is slightly over the budget and slightly over the accuracy I need.

    I am thinking to use ADS1120 with Mux on its input, such as MUX36D04(8 to 2), so I can use CH0 & CH1 of the ADC for all measurements, and CH2,3 will be used as current source.

    In this case I could connect 2 TCs and 2 RTDs, please find the attached document, could you let me know your comments about its principle?

    I checked the Mux data sheet and it has a very low leakage current which is good but the on-resistance could be a problem, because it is changing with input voltage and the temperature.

    If I used this Mux is it possible to achieve 0.1% accuracy in my measurement?

    Regards,

    Mahmoud

  • Hi Mahmoud,

    The MUX36D04 is a good low leakage mux.  It does require a minimum of +/-5V bipolar or a 10V unipolar for the supply.  This must be considered as a part of the cost of the design.  I think a better possibility would be to use the ADS114S08 which will offer the same flexibility of the external mux without the mux issues.  The combined cost of the external mux and ADS1120 will be comparable to the ADS114S08 (the ADS114S08 will most likely be less and will not require the higher voltage source needed for the external mux).  It is much lower leakage and has better gain and offset specs as well as self-offset calibration as compared to the ADS1120. You will not need to worry about the additional resistance or settling of the external mux either.

    So when you say you wish to achieve 0.1% accuracy what are you referring to exactly? With respect to temperature measurement, noise is a very large contributor preventing temperature accuracy over the range of temperature being measured.  In overall specifications the ADS114S08 will be much better than the ADS1120.

    Best regards,

    Bob B

  • Hi Bob,

    Thank you for suggesting ADS114S08, it seems a perfect fit for my application.

    I will get a development board to have a look and do some tests.

    Regards,

    Mahmoud