• State TI Thinks Resolved
  • Locked Locked
  • Replies 4 replies
  • Answers 1 answer
  • Subscribers 47 subscribers
  • Views 193 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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: Observed error due to rate of change of temperature

jonathan smith
Prodigy 10 points
Part Number: ADS1232

Tool/software:

Hi,

We have a weighscale system using the ADS1232, and have observed large spikes in adc value when rapid changes in temperature occur. We have managed to isolate this issue to the PCB, by having our loadcell outside of a thermal chamber, and we still observed this behaviour. See below for an example:

In this graph, Blue is raw adc, Green is the temperature taken from an NTC connected to the PCB and Orange is the temperature of an external data logger within the thermal chamber.

In this graph, Blue is raw adc, and Green is a simple approximation of the derivative of temperature, taken from the external datalogger. 

We are confident this is not an offset problem as the error we see varies heavily with the amount of load on the cell. This data is taken at 60kg, and the error produced is 1-2kg. With a load of 10-20kg this problem is much smaller, maybe 20-40g of error. 

Our device needs to be lower power, so we are cutting power to the ADS1232 and load cell between measurements, instead of using one of the existing power modes. Will this affect our reading? Is there some important memory on the chip that we would be erasing?

Do you know what could be causing this dependancy on rate of change of temperature?

We are also running some tests using the devkit, and preliminary testing is indicating that the devkit also is affected by the rate of change of temperature.

Thanks,

Jonny

  • 0
    TI__Mastermind 40950 points

    Hello Jonny,

    Welcome to the TI E2E community.

    First of all, what 'error' is concern?  Are we talking about the initial transient that then settles back , or is the error a steady state error after the readings have fully settled?

    There are a few things that could cause the spikes in measurement error.

    1.  External RC filter temperature sensitivity.  This could cause a long settling tail, although I would not expect it to be more than a few seconds depending on the values that you use.  If possible, use a good 6.5 digit DMM external to your temperature chamber to measure the voltage at the ADC inputs to see if the input voltage is stable during the temperature transients.  Please provide an image of the ADS1232 schematic, including the input filters and all ADC pin connections.  Also provide the ADC configuration, SPEED, GAIN, and the typical input voltage used.

    2.  This could be related to power-cycling the ADC.  For debug purposes, I suggest keeping the ADC on and rerun your tests to see if the problem persists.  When you do power-cycle the ADS1232, it is very important that you follow the power-up sequence shown in Figure 8-15 of the datasheet.  If the /PWDN pin is not cycled as recommended, you can see gain error shifts in the ADC readings.

    Regards,
    Keith Nicholas
    Precision ADC Applications

  • 0 in reply to Keith Nicholas
    Prodigy 10 points

    Hi Keith,

    Thanks for getting back to us.

    The error we are concerned with is the initial transient error that then settles back. 

    We are using a gain of 128, slow speed and using 3.3 volts as a supply and reference. 

    Here is an image of our ADS1232 schematic:

    We have managed to compare our design against the devkit, both show the error, but the magnitude of the error is a lot smaller on the devkit.

    blue: raw adc from our board

    green: raw adc from devkit

    orange: temperature from DL during our board test

    red : temperature from DL during devkit test

    In response to your point 1, we have removed C27, C29, C33, C34 from a board and ran the beginning of the same test:

    blue: raw adc, orange: temp

    It appears that this has reduced the magnitude of the error, although it is worth noticing this is on a different pcb (of the same design).

    As for your point 2, since I drafted that message yesterday we tried keeping the ADC awake and it didn't appear to make any difference. Also the power up sequence we perform is the same as the one specified by the datasheet. 

    Thanks,

    Jonny

  • 0 in reply to jonathan smith
    TI__Mastermind 41861 points

    Hi jonathan smith,

    Keith is out of the office, he will respond to your question by the middle of next week. Thanks for your patience

    -Bryan

  • 0 in reply to jonathan smith
    TI__Mastermind 40950 points

    Hi Jonny,

    The capacitance of the EMI filters is quite large; 1uF.  They probably have high dielectric adsorption, which will increase settling time.

    It appears most of this shift is due to the external components.  You may want to increase the differential capacitance for the reference inputs, C30, to 100nF.  Also, increasing the power supply capacitance, C12, may also help stabilize the circuit.  Although I do not think the ferrite L4 is likely causing any issues, you may want to replace this with a 0ohm jumper as well.  The CAP pin capacitor has a recommended value of 100nF as well; you are using 10nF.

    I am not really certain what else to recommend, other than trying to isolate the main source of the shift.

    Regards,
    Keith