• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 46 subscribers
  • Views 802 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

Use of ADS1248EVM

Mito
Prodigy 35 points
Other Parts Discussed in Thread: ADCPRO, ADS1248

Hi,

I am evaluating ADS 1248 ADC using MMB3 motherboard and ADS1248EVM board

to monitor temperature up to 0.001 degree Celsius accuracy. My question is how to

read the temperature using ADCpro?

I am using 4 wire RTD but I used only 3 of them. I connect the RTD wires to ADS1248EVM-board

without any resistors and the connection is AN0 (white), AN1 (red), and REF+ (red).

As I read the datasheet I found that you have to tie Rbias  and Rcom to the RTD too. Is it correct?

Thanks in advance

  • 0
    TI__Guru* 97515 points

    Mito,


    Here are a few slides on using an RTD and the ADS1248 to measure temperature. There are a few different setups used depending on the type of RTD, however, it is a good basic overview.

    While measuring the temperature to a 0.001C accuracy may be difficult measurement in any system, you can try one of the following setups. In many of these setups, you use a current source that feeds the RTD and Rbias. The voltage across Rbias is used to setup the reference. For the measurement, you need an exact measurement of the Rbias resistance. Then the output code can be used to calculate the RTD resistance as a percentage of the full scale.


    Joseph Wu

    ADS1248 Temperature Measurement.pdf
  • 0 in reply to Joseph Wu
    Prodigy 35 points

     

    Joseph,

    Thanks for the reply.

    Did you mean it's impossilbe to monitor the RTD temperature up to 0.001C accurracy with the ADC?

     Sainer

  • 0 in reply to Mito
    TI__Guru* 97515 points

    Sainer,


    Let's look at this from a system point of view. If you go to www.omega.com, and look up RTDs that are ultra-accurate and ultra-precise, the best ones are specified as Class 1/10DIN = ±1/10 (0.3 + 0.005 |t| )°C, From 0 to 100°C. This is just for the RTD itself and doesn't include any measurement errors caused by parasitics in the system. It probably includes some initial error, some drift error, and non-linearity in the element.

    If we are using a PT100 RTD that has a resistance of 100Ohms at 0C and a temperature coefficient of 0.385Ohms/C. That means that that we'd have to be able to measure a resolution of 0.000385Ohms for a change of 0.001C.

    Lets say that we find a PT1000 RTD with the same specifications so that the resolution would only need to be 0.00385Ohms for each 0.001C. With a 2mA current source coming from the ADS1248, you would be measuring a change of 7.7uV with each change of 0.001C. You would certainly have some extra noise coming from the current source, which might be mitigated by using the same currents source going into the reference resistor. However, you'd still need an accurate measurement for the reference resistor also.

    Also, the parasitic resistance in series with the current source could be reduced with a 4-wire RTD, you do still have some resistance from the input bias current. This error would depend on the resistance going to the input of the ADC, the gain, and the data rate (because the input current depends on the mode of operation). With a typical input current of 100pA, it would depend on the line resistance.

    While the noise might be close (I think the ADS1248 would be in the ball park), I think that the other errors would be a larger problem. Besides, what would you use to calibrate such a system?


    Joseph Wu

  • 0 in reply to Joseph Wu
    Prodigy 35 points

    Joseph,
     thanks for the slides that you sent to me. It was really useful, now I can measure the voltage
    and codes of the RTD using ADCPro.
     My project is about to measure temperature using RTD in range of -10C till 110C with 0.001C accuracy.
    I was thinking by adjusting the data rate and PGA I can get the accuracy.

    Following was my setting/conditions:
    Set the data rate around 10-160 sps. (my target is around 160SPS)
    Set the PGA as big as possible to make the RTD temperature range as big as the ADC full-scale voltage.
    (doing so because the RTD's voltage dynamic-range was so small compared to that of ADC).
     Just FYI:
    from RTD's resistor-temperature equation you can convert the RTD temperature (-10Ctill 110C) to resistor-value
    ,and at last you will get voltage-range 144.13mV till 213.44mV. Right now, I am using 820 Ohm bias resistor,
    which makes Vref=2.443[V]. This Vref range is very big compared to the above RTD's voltage-range, so to make the
      Vref range smaller and more useful, I decided to use smaller R bias  (around 80 Ohm).
    By doing so I wish maybe I can get  70% of ADC full-scale voltage used.

     But by doing so the resolution is getting down, so my next task is how to
    have at least pure 19-bits in my hand. If I could get it, I think the problem was solved.
    My problem now is  I only can get 16-useful bits.
    I was also thinking about using what we call programmable filter (using data sampling-average)
    to get the noise down, so I can get the resolution up.
    I don't know, but that's all for now.
    Thanks again for your help and wish me luck.

    Sainer

     

  • 0 in reply to Mito
    TI__Guru* 97515 points

    Sainier,


    When it comes to the noise (and extracting every possible clean bit you can get out of the system), I would concentrate on two things: the input and the reference. There may be other sources of noise (such as supply noise coupling into the measurement - particularly the digital supply), but getting the last few bits of performance requires careful attention to the inputs and reference.

    The noise in the ADS1248 is given in the datasheet in tables 5-10 come from actual measurements, but you need to be careful in making sure that you have a clean input source and reference. If you are using a ratiometric measurement based on a current source, it might have some noise immunity from using the same current as part of the source and reference. However, you still need to consider the contributions of other possible effects. You also need to be careful with any noise that can be coupled into the measurement from outside sources.

    Averaging can certainly help, but you do have a penalty in the speed in which the data is taken. By lowering the data rate, you should get back a similar amount of noise reduction.

    If you have any other questions, feel free to post again.


    Joseph Wu

  • 0 in reply to Joseph Wu
    TI__Guru* 97515 points

    One of my colleagues pointed out to me that you can use the Multichannel scope plug-in to display the temperature data in degrees C instead of the voltage.

    In this example, I'm using the on board temp sensor in the ADS1248. Using the multichannel scope, here's what you need.  

    1. Set the Units to Custom. This will bring up the equation block in the plug-in.

    2. Press Edit to bring up the Units Equation Editor.

    3. Write the equaton you want to use and then press Close.

    This will convert the data to what you want in the multichannel scope window. Here, I've used the system monitor for data, but you can convert using codes or voltage for any system you want with the EVM.


    Joseph Wu