Looking for 8/10/16 Channel thermocouple to Digital Converter

Hi Lakshminarayana,

thank you for reaching out with your request and your interest in our ADC solutions for thermocouple measurements. We do actually have a broad portfolio of ADCs that were designed for temperature measurement applications, such as RTDs, thermocouples and thermistors.

Considering the number of thermocouples you want to measure I would probably start with the ADS124S08. This device could measure up to 6 thermocouples. You can certainly measure more by adding an external multiplexer.

You might find the following application note worthwhile which we released this year:
A Basic Guide to Thermocouple Measurements

Regards,

Hi Joachim Würker,

Thanks for your quick reply.

As I can see from the Datasheet of ADS124S08, It supports up to 12 Channels Does it indicate that I can connect 12 Thermocouples?

I don't see any specification of which type of Thermocouples(K/J/Other) it supports.

Thanks.

Hi Lakshminarayana,

unless you connect one end of all thermocouples to the same potential you will need one differential analog input pair per thermocouple. That is why I said the ADS124S08 can support up to 6 thermocouples.

If your application allows to connect one end of all thermocouples to the same potential then you could measure up to 12 thermocouples. In that case the common thermocouple leads would be connected to AINCOM for example. The other analog inputs would then measure differentially against this AINCOM input to read the voltage created by the respective thermocouple.

Our ADCs can be used for any thermocouple type.
Please remember that our ADCs do not integrate any algorithms that perform linearization and the cold-junction compensation calculation. This will have to be implemented in your processor.

Regards,

Hi,

Previously I had used this MAX6675 which performs Cold-junction compensation calculation. But am not aware how to do those and even about Linearization.

Could you please guide me to some of the example with this  ADS124S08 How it has to be done.

Also are there any other IC which includes Linearization and Cold-junction compensation calculation with Multi channel support?

What will be the difference

* If I connect all thermocouple one end of Cold wire to AINCOM and other Cold end to AIN ? AND 

* If I use separate AIN for each thermocouple (Like one end to AIN1 and other end to AIN2)?

Thanks.

Hi Lakshminarayana,

I see. I just looked at the datasheet of the MAX6675. It does seem to implement the cold-junction compensation algorithm, but it only seems to implement a straight-line approximation of the thermocouple output. As this device only offers 12-bit performance it might be acceptable for them to not implement a proper linearization algorithm using polynomials or look-up tables.

If you search the web you will find a lot of material that explains how to accurately measure thermocouples.
Maybe you start with the application note I referenced before: A Basic Guide to Thermocouple Measurements

Following is an example which I found useful that shows how the algorithm could be implemented:

If you don't need better accuracy than what MAX6675 offers then you could replace all the polynomial calculations with the simple straight-line approximation as done in MAX6675.

Following is a list of reference designs we have built at TI using ADS1220 to measure thermocouples:

• http://www.ti.com/tool/TIDA-00189
• http://www.ti.com/tool/TIDA-00168
• http://www.ti.com/tool/TIDA-00650
• http://www.ti.com/tool/TIDA-00018

Unfortunately we don't have any solutions in our portfolio as of today that integrate the linearization and cold-junction compensation algorithms and support multiple thermocouples.

I would say the main difference will be that you must not have any common-mode voltage differences between the different thermocouples in your system when tying them all together.

Regards,

Hi, Thanks for the references.

- Are there any Thermocouple can be of Single Channel with the integrated the linearization and cold-junction compensation algorithms.

- Can you please refer me to some of the examples of Multichannel Thermocouple using MUX with Single /Dual Channel Thermocouple to Digital Converter IC?

Hi Lakshminarayana,

unfortunately we don't have any single, dual or multi-channel devices that integrate the linearization and cold-junction compensation algorithms.
However we are working on according example code for some of our ADCs that will include those algorithms. But it will take a few more months till we will have that online.

The algorithms are not that complicated to implement. Especially if you would use a straight-line approximation as in MAX6675 this makes things quite easy.
If you have read through some of the literature we can work together to put a suitable solution together.

If you are fine with the performance you get from MAX6675 then we don't need to use a 24-bit ADC. We could use the ADS114S08 (16-bit) instead.
You will also have to decide how and where to measure the cold-junction temperature. There are many options: thermistor, 2-wire RTD, analog output temperature sensor IC, digital output temperature sensor IC, etc.
As you want to measure so many thermocouples, you will probably have to measure the cold-junction temperature at multiple points because there could be some temperature gradient across your thermal block/cold-junction.

Regards,

Okay.

Since this is the first time I'm designing the Thermocouple I'm not much aware of Algorithms and to implement.

But I would like to work on those Algorithms.

I have few questions on those Algorithm: * Are those Hardware based or Software based?

In my application mainly I need to use Thermocouple (K J Type) to measure the temperature of Heater and based on the temperature Controlling the O/Ps.

I came across this MAX31856 IC which has an integrated Algorithms and 19 bit ADC and got some reference design too:

I would like to hear your opinion on this.

I am looking forward to design using TI IC's and Learn more about those Algorithm.

Thanks.
Lakshminarayana K N.

Hi Lakshminarayana,

I know those TC-to-digital converters from Maxim and Microchip. I think they are interesting solutions for customers who don't want to spend the time developing the SW algorithms. They are more-or-less plug&play solutions. However they are limited in flexibility (e.g. only one RTD can be measured, internal temperature sensor has to be used for cold-junction compensation) and also in achievable accuracy and resolution.

From the hardware features alone our ADS1120 (16 bit) and ADS1220 (24 bit) are very comparable to those devices, but offer additional flexibility. Once we have the algorithms for those devices released on ti.com as example code then our solutions are hopefully as easy to use for thermocouple measurements as those competitive solutions.

The algorithms are implemented in SW in the MCU. They are just some calculations based on the measurement of the thermocouple voltage and the cold-junction temperature.

On the hardware side you should think about the following things first:

• How should the thermocouples be biased so that the differential output signal falls within the input voltage range of the PGA integrated in our ADCs.
  If you tie one side of all thermocouple to the same potential, you could for example leverage the REFOUT pin on ADS114S08 to bias the thermocouples to 2.5V.
• Do you require any open-wire detection for the individual thermocouples?
  Open-wire detection can be implemented in multiple ways.
• How do you want to measure the cold-junction temperature?
  As mentioned before there are multiple temperature sensor options available. I cannot tell you which one works best for you. The easiest would probably be to use one of our analog or digital output temperature sensors.
  And you might potentially have to measure the cold-junction temperature at multiple points considering how many thermocouples you need to measure.

Following is high level how the algorithm could look like:

1. Configure the ADC to measure the differential output voltage of the thermocouple.
2. Take one or multiple readings.
3. Convert the ADC code into an according voltage, V_TC.
   With ADS114S08 the conversion reads:
   V_TC = (Code/2^15) x (VREF/Gain)
   where VREF = 2.5V and Gain is the gain setting of the PGA.
   The gain setting will depend on the temperature range you want to measure with the thermocouple.
   Note: You might have to implement additional offset and gain calibration.
4. In case a thermistor, 2-wire RTD or analog output temperature sensor is connected to the ADC to measure the cold-junction temperature, then configure the ADC to measure it accordingly.
5. Take one or multiple readings.
6. Convert the measured ADC code into an according temperature, T_CJ.
   How the conversion looks like depends highly on the temperature sensor being used.
   With a thermistor or RTD you might have to implement some linearization algorithm depending on the accuracy you want to achieve.
7. Convert the measured cold junction temperature, T_CJ, into an equivalent thermoelectric voltage, V_CJ, using the tables or equations provided by NIST.
   This is either a table look-up or a polynomial calculation. A table look-up is in general less math intense.
8. Add V_TC and V_CJ and translate the summation back into a thermocouple temperature using the NIST tables or equations.
   This is again a table look-up or a polynomial calculation.
   The result is a linearized and cold-junction compensated temperature reading of the thermocouple.

If you don't need highest accuracy then you might be fine with just a straight-line approximation as done in the Maxim device. This would make the whole algorithm very easy.

Regards,

Hi Joachim Würker,

Thanks for your reply.

By when can I expect those algorithms on ADS1220.

Can't the Internal Precision Temperature Sensor of ADS1220 used for Cold-Junction Compensation?

"Do you require any open-wire detection for the individual thermocouples?" : I'm okay to use the MUX based solution rather than Multichannel IC as in the example.

Does TI offer any MUX?

"As mentioned before there are multiple temperature sensor options available. I cannot tell you which one works best for you. The easiest would probably be to use one of our analog or digital output temperature sensors." : Do you mean to use the External Analog/Digital Temperature sensor for Compensation?

Hi Lakshminarayana,

unfortunately I cannot provide any timeline for this project (example code for ADS1220) at this moment. It might take a few more weeks/months.

However one of the reference designs I mentioned previously () seems to provide some firmware example code (TIDA-00168 Firmware).
This design was developed for K-type thermocouple measurements and shows how an external RTD or the ADS1220 internal temperature sensor can be used for the cold-junction temperature measurement. I have not reviewed the design and firmware and therefore don't know any more details about it. But I think it would be worthwhile to study.

The challenge in using the ADS1220 internal temperature sensor for a multi-channel thermocouple measurement system is that you might not be able to place the ADC close enough to the thermocouple connectors, i.e. the thermal block/cold junction. Therefore the temperature the ADS1220 measures internally might be a few °C different than the actual cold-junction temperature. This temperature difference will directly affect the accuracy of your thermocouple temperature measurement.
If you can make sure that you have good thermal connection between the cold-junction and the ADC, then nothing speaks against using the ADS1220 internal temperature sensor. But I could see it as a challenge.
Otherwise it might be better to use an external temperature sensor and place it as close to the thermocouple connectors as possible.

TI offers a large portfolio of multiplexers. Unfortunately I am not a MUX expert. The requirements for the MUX shouldn't be very demanding. You don't need lowest R_ON in this case. I often suggest the CD405x series, but I know that we have much newer and better MUXes in our portfolio now.

Regards,

Hi Joachim Würker,

Yeah I agree that the accuracy of the Thermocouple will vary as I keep MUX in between the Cold Junction and ADC.

But the Internal Temperature sensor of the ADC will measure the surrounding/Ambient Temperature right?(which will be almost same as the Cold-Junction Temp if am not wrong).

I'll have a look into the design.Thanks for the reference.

I have a question on using an external Temperature sensor to monitor the cold junction Temperature. If so we have to follow the Cold-Junction compensation and Linearization Algorithm right? Won't there be few second(to be precise millisecond) delay while reading the actual Temperature?

Yeah It seems that Accuracy while using Multi-channel Thermocouple is a challenge. How one can get more accuracy other than placing the external Temperature sensor near the Cold-Junction?(Is it the only way to go with separate ADC for each thermocouple?).

I would like know your opinion on this.

Again thanks for referring for the MUX.

Regards,

Lakshminarayana K N.

Hi Lakshminarayana,

the internal temperature sensor of ADS1220 measures the internal die temperature. If the package has good thermal connection to the PCB then the die temperature will track the ambient temperature very closely.

You will be surprised, but depending on your PCB layout and placing of components there can easily be a difference of a few °C between the cold-junction temperature and the point where the ADS1220 is located. The closer you can bring the ADS1220 to the cold-junction measurement point the better it will be. This also assumes that you have good thermal connection between the ADS1220 and cold-junction measurement point. So good thermal layout is of the essence.

For an accurate thermocouple measurement you will always have to implement the linearization and cold-junction compensation algorithm. This is independent of the way you measure the cold-junction temperature.

Just to clarify, I am not concerned about the time difference between the thermocouple and cold-junction measurement. Temperatures fortunately do not change that quickly. Therefore it isn't usually critical to measure the thermocouple and cold-junction temperature at the same time. I am more concerned about the distance between the cold junction and the point where you measure the ambient temperature on the PCB.
What most customers do is to take multiple readings of the thermocouple, and then every now and then take a reading of the cold-junction temperature. If you measure multiple thermocouples you could for example take one reading from every thermocouple and then take one reading of the cold-junction temperature.

If your accuracy requirements are not really high then measuring the cold-junction temperature at a single point using the internal temperature sensor of ADS1220 might be sufficient.
If you are targeting best temperature measurement accuracy then it might be adviseable to use at least two external temperature sensors and place them at either end of the thermal block. If you need to connect 16 thermocouples to the PCB I would imagine the thermal block being pretty wide. With the help of the two temperature sensors you will then be able to determine if you have any temperature gradient across the thermal block. It could for example be that one side of the thermal block is a few °C hotter than the other side. By using a linear interpolation you could then calculate what the approximate temperature at every single thermocouple connection point is. This is pretty advanced and high-end stuff. I am not sure if you need to go to that extreme.

Regards,

Hi Joachim Wuerker,

I'm not worried about the time delay when using external sensors for the Cold-Junction Temperature measurement  I was just curious to know will there be any time delay in seconds.

"I am more concerned about the distance between the cold junction and the point where you measure the ambient temperature on the PCB.": Coming to the point of Layout, My idea is to keep the Mux in the bottom Layer of PCB so that I can keep the ADC(ADS1120/Other) near to the Cold-Junction of Thermocouple.

Any opinions on this?

Thanks and Regards,

Lakshminarayana.

Hi Lakshminarayana,

the time delay between the measurements depends on the data rate that you are selecting for the ADC. With ADS1120 you can read one channel within ~0.52ms when using the fastest data rate. Means you could in principle measure one thermocouple and one temperature sensor for the cold-junction compensation within ~1.04ms.

Putting the Mux on the bottom side is definitely an option you could try to make the layout more compact and to move the ADC closer to the cold junction.

Regards,

Okay Thanks.

Once I complete the Layout Can I send it for the review?

Thanks.

Lakshminarayana.

Yes, certainly Lakshminarayana.

Regards,

Thank you.

Hi Lakshminarayana,

it might actually be a good idea to review your schematic before you go into layout.

Regards,

Sure.

Thank you.

Do you recommend anything to read the values over RS-485?

Thanks.

Hi Lakshminarayana,

do you mean you want to communicate directly through RS-485 with the ADC?
I am afraid this will not be possible. You will need an MCU that connects through SPI to the ADC to communicate with the ADS1120.

We have ADS112U04 (16-bit) and ADS122U04 (24-bit) available which use a UART interface. Those ADCs are very similar to ADS1120 and ADS1220 and could in principle be connected directly to the serial port of a PC.

Regards,

No, I didn't mean directly reading from the ADC. Yes, I would need a microcontroller to convert the digital value into the temperature and feed to the PLC through the RS-485.

Is there a reference example design for this?

Thanks.

Hi Lakshminarayana,

I am sorry, but this seems to be a question outside my area of expertise. That might be a better question to ask in our microcontroller or interface  forums I would guess.

Regards,

okay.