RE: ADS114S08: Thermocouple and RTD signal transmitter Design Question

Hi Mahmoud,

I split the thread as we are now discussing a different device.  See my answers below.

Best regards,

Bob B

Mahmoud Zannerni62 said:

Hi Bob,

I have made a design to measure (2 RTDs + 1 TCs) or (3 TCs) or (1 RTD + 2 TCs), could you check the schematic and tell me what you think about it? Please ignore the values for now since I didn’t not finish the calculations yet. [Bob] I think your schematic is fine.

The schematic shows one channel only which could accept Thermocouple or RTD of any type, this will use 6 Analog input (2 IDAC , 2 Ain and 2 Ref), the other channel will use the dedicated reference inputs so only 4 Analog inputs needed. The last two input will be used for one TC only.

1.The I_DAC2 source will remain connected even if it is not needed but I will disable it from the internal ADC mux, I am not sure if that will cause any problem? [Bob] That is not a problem.

Is it better to have an external jumper to wire the I_DAC2 source when 3-wire RTD used? [Bob] That is up to you but it is not necessary.

2. I would like to ask about the ADS114S08 datasheet equations number 21 and 23, in equation 21 the lead resistance seems to be ignored while it was taken into account in equation 23. [Bob] Equation 21 is actually a rough calculation. Technically you are correct, but we don't necessarily know the exact lead resistance which could change slightly from RTD to RTD.  What we need to find is a resistor value that will bias the RTD regardless of the lead resistance.  In the end the voltage drop across the lead resistance will be quite small relative to the bias voltage.  If the lead resistance is 10 ohms, the total voltage across the leads will be 10mV for total of 1mA of IDAC current.  10mV is quite small compared to 1V.  Also, consider there is some tolerance related to the bias resistor which will also deviate slightly with respect to the calculation.

In datasheet equation 21 is:

V_AinN = (I_DAC1 + I_DAC2) R_BIAS

However, I think it should be:

V_AinN = I_DAC2*R_L2 + (I_DAC1 + I_DAC2) (R_L3 + R_BIAS)

For Equation 23 in data sheet:

V_AinP = I_DAC1(R_L1 + R_RTD) + V_AinN

[Bob] Actually it is not  V_AinN  but  V_{Rbias + the voltage drop across RL3}

I think should be:

V_AinP = I_DAC1(R_L1 + R_RTD) + (V_AinN / I_DAC2*R_L2)

[Bob] V_AinP = I_DAC1(R_L1 + R_RTD) + (I_DAC1 +I_DAC2) (R_L3 + R_BIAS)

Please correct me if my calculations are wrong.

 

3.Regarding to connecting a 2-wire or a 4-wire RTD, the data sheet didn’t show detailed calculations for that, and I didn’t find any TI document doing the calculations for high side reference. So I tried to do them myself.

2 Wire RTD:

V_AinP = I_DAC1(R_L1 + R_RTD + R_L2 + R_BIAS)

V_AinN = I_DAC1* R_BIAS

To use the same R_BIAS used for 3-wire RTD the IDAC1 current needs be doubled? [Bob] You could double the IDAC1 current or you could use IDAC2 the same as with the 3-wire if you make the same connections as shown in your schematic.

In this case the two lead resistors will get into the measurement, and I think I should subtract them out by software? [Bob] You can try to calibrate the lead resistance, but the lead resistance can change over temperature.  Generally using a 2-wire you are not going to get much accuracy in the end.

4 Wire RTD:

V_AinP = I_DAC1(R_RTD + R_L4 + R_BIAS)

V_AinN = I_DAC1* (R_BIAS + R_L4)

To use the same R_BIAS used 3-wire RTD the IDAC1 current needs be doubled. [Bob] Yes you could do that but this changes the reference voltage too.  I should have mentioned this with the 2-wire case.  You could alter the design slightly so that the bias resistor is larger in value so that you can reach a good compromise.  For example, you could try doubling the bias resistor value and see if you could still meet the requirements for the 3-wire case.

In this case only R_RTD is measured.

Please let me know if I should post this reply as a separate question.

Regards,

Mahmoud

Hi Bob,

Thank you for your detailed answer.

1. Regarding to using two current sources for 2-wire and 4 wire RTD, will that cause any degradation in the measurement accuracy?

2. I found using R_Ref = 4K and R_Bias = 2.3 K will be good to measure PT100 sensors and PT1000. In case of using PT100 the VI_DAC1 is very close to maximum allowed level, is that ok?

3. And when measuring PT1000, gain of one is used and the Reference is only 1 V. Can the low reference be bad for the system? I know using low reference means sampling more noise in the ADC, but also amplifying the signal will amplify the existing noise anyway, I am not sure in practice if using low reference is a good or bad idea?

I know a digital potentiometer could solve the biasing problem but I prefer not to add extra component if possible.

Regards,

Mahmoud

Hi Mahmoud,

See my answers below.

Best regards,

Bob B

Mahmoud Zannerni62 said:

Hi Bob,

Thank you for your detailed answer.

1. Regarding to using two current sources for 2-wire and 4 wire RTD, will that cause any degradation in the measurement accuracy? [Bob] This would depend on how the current source is applied.  In the 2-wire case this should not be an issue as long as you apply the current using the second current source at the AINN input as the current summing mode is at the connector.  In the case of the 4-wire, the current must be applied where the lead resistance is not a part of the current path.  For the 4-wire case the current would need to be injected at node where the 4th lead wire connects to Rbias.

2. I found using R_Ref = 4K and R_Bias = 2.3 K will be good to measure PT100 sensors and PT1000. In case of using PT100 the VI_DAC1 is very close to maximum allowed level, is that ok? [Bob] Remember to include all voltage drops (resistors, diodes, etc.) in your final calculations and also remember resistor values and tolerances.  For example a 4k resistor might actually be 4.02k and 2.3k might actually be 2.32k.  Also consider any variation and drift of the IDAC sources.

3. And when measuring PT1000, gain of one is used and the Reference is only 1 V. Can the low reference be bad for the system? I know using low reference means sampling more noise in the ADC, but also amplifying the signal will amplify the existing noise anyway, I am not sure in practice if using low reference is a good or bad idea? [Bob] There are always tradeoffs in the design when attempting to support a variety of sensors.  In the end it will depend on if you can achieve the desired temperature resolution.

I know a digital potentiometer could solve the biasing problem but I prefer not to add extra component if possible.

 

Regards,

Mahmoud

Hi Bob,

I will redo the calculations and consider the TVS, Schottky diodes and the current limit resistors on the I_DAC1 path.

1. I would like to ask about the AAF in front of the ADS114S08, I think the I_DAC current will not flow through that path, so this will not affect the previous calculations? The effect will be a voltage drop in the ADC measurement?

2. I am not sure if using a bipolar supply will make things easier in terms of biasing?

Regards,

Mahmoud

Hi Mahmoud,

The IDAC current should be routed in such a way that the anti-aliasing filter is not in the current path.  Take note of the ADS114S08 datasheet RTD example and see how the second IDAC is routed around the series resistance for the filter.  As far as using bipolar supplies, the IDACs are referenced to AVSS I'm not sure there is any benefit for RTDs.  For TC that may be referenced to analog ground (such as a grounded tip TC) there can be some advantage to the bipolar supplies.

Best regards,

Bob B 

Hi Bob,

I would like to ask about the external blocking diode voltage, it seems in the calculations in equation (24) "ADS114S08 datasheet" V_D is considered to be 0.025 V, I could not find any Schottky diode with such a low forward voltage, the best I could find is 0.2 V. Did I misunderstand something here?

Also, I would like to ask if the blocking diodes will provide enough protection for I_DAC pins? do I need to add TVS and series resistor to protect those pins in addition to blocking diodes?

Regards

Mahmoud

Hi Mahmoud,

I didn't try to rework the equation 24 result, but you are correct about the forward diode voltage drop being at least 200mV for the Schottky diode.  The result of the equation may be incorrect, but the equation itself for the total voltage is correct.  You just need to make sure that you add in the proper voltage drop for the diode you would use.

Regarding the protection diodes, you need to use a diode that has a minimum voltage rating relative to the maximum voltage applied.  The diode itself will only protect for applied reverse voltage, so some transients may cause an issue.  In this case a TVS diode to suppress the transient may be required.  However, you need to consider diode leakage in the path.  Also, series resistance is helpful but will potentially add new problems with the additional voltage drop.

The datasheet 3-wire application for the ADS114S08 shows a high-side reference and two current sources with the protection diodes.  It is not necessary to use 2 current sources in this configuration, nor do you have to use the diodes.  The high-side reference already has a series resistance that can protect the IDAC source, and the diode is just additional protection.  The lead resistance can be calculated by making 2 measurements if the 3rd lead is also connected to an input (such as by moving the AIN3 connection from lead 2 to lead 3.  In this way you can measure the lead resistance and subtract out the lead voltage drop from the result.  The AIN3 input would have the same filtering as AIN1 and AIN2 protecting the AIN3 input with a series resistance.

Best regards,

Bob B

Hi Bob,

I am thinking to use a digital rheostat as a bias resistor, in this case a software change will allow to bias different RTDs.

All digital rheostats I have found don't have a separated power pins(Analog & Digital). I wonder about the result of powering it from the Analog power supply! will that introduce a noise to the instrumentation system? The digital rheostat is controlled by SPI or I2C.

The PSU that I am designing consists of a DC-DC and the two LDOs one for Analog and the other for Digital power.

Regards,

Mahmoud

Hi Mahmoud,

There are two considerations.  One is related to having an analog supply connected to a digital device.  The second is having digital signals in the analog domain.  The second in probably more important than the first as it relates to routing digital signals (primarily clocks) in a potentially sensitive analog areas.

You might consider using a digital isolator (ISO type products such as ISO7741 for SPI) where you can isolate the power supplies and disable the output which will pretty much take care of both issues mentioned above.  Basically the digital signals are dynamic, but your use case in primarily static, so if you can isolate the digital pot from the rest of your digital signals this should be of benefit.  The second issue remaining is keeping the dynamic signals (SPI communication for example) away from the analog.  If you choose an isolator with an enable, you can turn off the dynamic signals to the digital pot once configured.  Of course you will need pull-up or pull-down resistors to keep the inputs of the digital pot from floating.  You also need to be careful with PCB routing keeping in mind return currents of digital signals and preventing them from getting into the analog.

Best regards,

Bob B

Hi Bob,

Regarding the R_Bias I will make different boards with different R_Bias values to avoid using the digital pot.

I may to use zero ohm links in my design so the board can be modified during production to support different sensors (Loadcell ,RTD ,TC). Will using zero Ohm resistor in the signal path introduce a noise or it will not be a problem as long as I am using a ratiometric measurement system?

As an example, I will use a zero Ohm resistor to connect the Loadcell negative excitation voltage to ground.

Also, I would like to ask about self and system calibration:

1. Is there any advantage of using on chip calibration registers instead of doing the calculation in the microcontroller? it seems like all done digitally so there is a tiny loose in the dynamic range of the ADC anyway?

2. How often do I need to run the calibration for self and system?

Is it a good idea to run the system calibrations on the production phase and store the values in the microcontroller memory. So every time a new configuration is used the values will be written to ADS114s08 without running the calibration?

Regards,

Mahmoud

Hi Joseph,

Thanks for your reply.

I will post my schematic when it is ready for a review.

I would like to ask about ADS114s08 internal temperature sensor accuracy, I couldn’t find detailed information in the datasheet, and is it calibrated or it needs calibration?

I noticed that the Evaluation board has a NTC for cold junction compensation, I tried to use the predefined scripts to measure the internal temp, on board NTC (RT2) and External PT100. The NTC and RTD giving almost the same temp value but the on-chip temp diode always gives one degree less than the other two.

One more thing I would like to ask about, is there any way to make a single ended measurement without connecting any Analog input to ground? Can one side of PGA be connected internally to AVss!

Regards,

Mahmoud

Mahmoud,


It's true that we don't give much information about the internal temperature sensor accuracy. It does not have a guaranteed specification, and I wouldn't use it for cold junction compensation.

I would guess that the error would be somewhere near ±6°C for the range of operation, but that is a guess based on some characterization data with this device and similar other devices. Again, we don't give a min/max specifiaction and it shouldn't be used for any CJC.

On your other question, single-ended measurements should be made with a connection to an AINx pin (as AINCOM may be used). There is not other internal connection to AVSS. Note that the PGA must be disabled to make measurements with respect to AVSS. The PGA input range does not extend all the way to the supply rails and must be bypassed.


Joseph Wu