• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Subscribers 46 subscribers
  • Views 3275 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.

ADS1232: ADC counts not stable

Navin Rana
Intellectual 325 points

Part Number: ADS1232

Dear Sir,

I am using ADC ADS1232 in my hardware for reading the load cell value in my program. I am taking 21 bits of ADC data and doing average of 16 values of these data. Then use these values for further calculation. My problem is that when I apply power supply my circuit board, I got some counts like 77563 approx in starting. But these counts will decrease automatically in some time intervals, like: 77560, 77558, 77555, 77551 etc. while there is no load apply on the load cell. Also in some times same functioning performed in increase manner like: 77565, 77568, 77569, 77571.......... and so on. The time interval in these readings is approx 20-30 seconds and also in minutes some cases.

ADC configuration I am using is as below:

Gain: 128 (Gain0 -> 5V, Gain1 -> 5V)

Speed: 10 SPS  (SPEED Pin -> Ground)

Temp: Low Signal to Temp Pin

External Crystal (4Mhz) Attached on pin 3 and 4 .

So please suggest me why this behavior is occurred. Is there anything missing in my circuit?

Thanks in advance for your valuable suggestions.....

With Regards..

Anuj Jain

  • 0
    TI__Guru** 113765 points

    Hi Anuj,

    It would be helpful for me to actually see your schematic.  This sounds to me like analog settling of either the input or the reference.  This measurement should be ratiometric where the excitation source for the reference is the same as the excitation of the load cell.  If the measurement is truly ratiometric then you should not see the settling of the reference, but if the analog input filter differs from the reference input filters, or if there is some other loading that causes instability then you may see the settling. 

    Sometimes you can see issues with drift due to noise if you are using both differential and common-mode capacitors on your input circuit.  To prevent difference signals at the analog inputs the differential cap must be at least 10 times larger in value compared to the common-mode caps.

    There can also be some self-heating of the load cell itself that may need to stabilize.

    To help in evaluating the performance, it would be good to see the raw codes from the ADS1232 without any averaging or other computations as 24-bit values during this startup period.  It is not clear to me what the actual value of 77565 truly means in relation to the voltage reference.  Is this value based on the 24-bit output code or has this been altered to 21-bit?

    Best regards,

    Bob B

  • 0 in reply to Bob Benjamin
    Intellectual 325 points

    Dear Sir, 

    Thanks a lot for reply.. I am attaching my schematic and ADC code file here. Please check these is there anything should change in these. 

    And as you are asking for 21 bit data, I am reading 24 bits first, then right shift these bits 3 times. now I process these 21 bits data. 

    ADC.HADC 1232 schematic.pdf

    Thanks & Regards 

    Anuj Jain

  • 0 in reply to Navin Rana
    TI__Guru** 113765 points

    Hi Anuj,

    The schematic shows that you have filtering on the analog inputs, but not the reference near the reference inputs to the ADS1232.  Also, C5, C6 and C8 are shown as the same value.  You need to make sure that C5 is at least 10 times larger in value than C6 and C8.

    The GAIN1 pin should be high, have you verified this is happening?

    Best regards,

    Bob B

  • 0 in reply to Bob Benjamin
    Intellectual 325 points
    Dear Sir,

    I had done the changes suggested by you.
    C5 --> 0.1uF
    C6 and C8 --> 0.01uF

    And GAIN1 pin is high, I verified it.

    But still problem is same..

    With Best Regards

    Anuj Jain
  • 0 in reply to Navin Rana
    TI__Guru** 113765 points

    Hi Anuj,

    I read through the thread again. Just to review, you are taking the 24-bit result and right shifting 3.  You are also averaging, so it is not clear how you are averaging as to a moving average or something different.  If we just consider the result that is right shifted to 21-bit, you will still see some noise.  The absolute best you can get from the ADS1232 with 5V AVDD is given in the datasheet as 18.4 bits noise free using a gain of 128 at 10sps.  So you will still see some code variation even with right shifting the data.  This variation is very close to the original numbers you posted as it will be possible to see 8 codes of variation even after right shifting the data.

    Averaging will help if the noise is Gaussian, but the system may pick up other noise sources as well as having the potential to drift.

    To really isolate the problems, I would use the raw data without shifting or averaging to determine how well my system is responding.  I would also remove the load cell and apply a short to the inputs biased at mid-AVDD supply.  You should be able to see the datasheet noise performance with the shorted input case.  If not, then you need to discover the source of the problem, which could be related to PCB layout.

    Once you are able to achieve the datasheet performance, go back to the load cell and see the results.  With no load, you should see very similar performance as the shorted input case.  If it is much different, then you need to find the source of the noise.  This could be EMI/RFI or power-line cycle noise.  These noise sources usually come through the cabling/connections.

    Best regards,

    Bob B