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.

ADS131M02-Q1: Question about ADS131M02-Q1 sample error

Part Number: ADS131M02-Q1

1.When  i write CH1_CFG regs MUX1 [1:0] is 01b,i get a channel value is FFA417. Then, I write this value into CHI_OCAL regs, i I didn't get an expectation. So, I want to know how to get offset error to write into CHI_OCAL regs.

2.I want to know how to get gain error to write into CHI_GCAL regs, I know how the error is translated into the register, but do not know how to get gain error.

3.Whether the noise error can be eliminated.

4.How to use  this DC Block Filter ,I can't calculate a COEFFICIENT.

5.I would like to know if the minimum accuracy of this adc can reach 0.2uv. Because we are used to measure the current of 100uΩ shunt resistance, 0.1uV is 1mA. Whether this accuracy will be disturbed by noise.

Looking forward to your reply, thank you

  • One more question.

    When i use different gain values, whether these errors are also different.

  • Hi kaixin,

    I will look into your questions and get them back to you tomorrow.

    Regards,

    Dale

  • Hi kaixin,

    1. When you set the MUX bits = 01b to short the ADC inputs together, it will only calibrate the ADC offset, it will not calibrate your system offset, so if there is any additional offset error in your system, it will not be accounted in the OCAL procedure for the calibration, but it will show up in your final measurements.

    2. Please see the details in the Precision Labs ADC series:  Understanding and Calibrating the Offset and Gain for ADC Systems

    3. You can minimize the impact of the noise in your system, for example: use higher OSR which leads to a lower data rate to get a lower noise. Precision Labs ADC series: ADC noise can help you understand the ADC noise.

    4. You do not need to calculate a COEFFICIENT. You need to determine the -3-dB high-pass corner frequency of the DC Block filter which scales directly with the data rate. Data rate also scales with CLKIN, so we could simplify the DC Block filter description in the data sheet and only mention the data rate. see a previous query on the forum.

    5. The accuracy in your system is affected by the errors from each component including ADC, power supply and layout etc. For example, you can calibrate out the offset error and gain error of the ADC, however you will have to consider the impact from the drift error. You can use Analog Engineer Calculator tool to predict the TUE error.

    6. Can you clarify the test you did? Did you test  ADS131M02EVM or your own circuit board? what's your gain and which did you see?

    Regards,

    Dale

  • Hi Dale

    I have read your reply and have the following questions.

    1.So can I connect P/N to ground when I want to calibrate my system offset.

    2.I see a previous query on the forum. .

    My fdata is 0.250ksps, but I don't know what "4 kSPS x" is.

    3.I still don't know how to set the DCBLOCK[3:0] bits . Can you explain it in detail. 

    4.I short-circuited P and N, always get a output around ±5*10^-5 mv.

    Kind Regards

  • 5. When i use different gain values, whether I need to rewrite OCAL and GCAL.

  • Hi kaixin,

    1. You can short the ADC's input and it will be same as short input internally. If you want to do a system level calibration, you can short the input on the system's input so the system offset error including your front-end circuit can be calibrated out.

    2. As I mentioned, the -3-dB high-pass corner frequency of the DC Block filter scales directly with the data rate, the data in the table 8-4 of the datasheet is provided for an 8.192-MHz CLKIN frequency and a 4-kSPS data rate, this is where the "4-kSPSx" comes from.

    3. You have to determine your -3dB cutoff frequency (fcutoff) of the filter first, then you can get the coefficient/DCBLOCK[3:0] setting from the equation: fcorner=(fcutoff/fdata)*4kSPS,  where fdata is data rate and fcorner is -3dB corner frequency in the table 8-4.

    4. 5*10^-5 mv is small enough. 

    Regards,

    Dale

  • Hi Dale

    5*10^-5 mv is small enough.

    I'm sorry I misspoke here,it's 4. 5*10^-5 V.

    Ads131 is used by me to measure shunt resistance, Rshunt = 100uΩ, so the voltage of 1uV is equal to the current of 10mA.

    I need 2mA accuracy,equivalently 0.2 uV accuracy.

    Kind Regards

  • Hi Dale

    The main problem now is that the continuous sampling mode I used, Clockin = 8Mhz, OSR = 16384, fdata = 0.250ksps, gain = 1.

    When Vin = 0.3185V, COAL_Regs and GCALRegs do not write values, Vout=0.3083V;

    When Vin = 0.6367V, COAL_Regs and GCALRegs do not write values, Vout=0.6021V.

    When I do not calculate COAL_Regs and GCALRegs, there will be a big difference between Vin and Vout, is this normal? This is no longer an error, it is the main problem that bothers me, and the questions I asked you before are based on this.

    We need to evaluate whether ads131 can be used to measure shunt currents.

    Kind Regards

  • Hi Keixin,

    4. 5*10^-5 V is also small and it is reasonable too because typical offset on ADS131M02-Q1 is 175uV when global-chop mode is turned off, it's typical 35uV when global-chop mode is turned on, your test result meets the specification in the datasheet.

    Your challenge is not calibration. According to your latest information, you need 2mA accuracy equivalently 0.2 uV accuracy, this is not realizable. Regardless other errors, the ADC noise for your 16384 OSR, 0.25ksps data rate and gain=1 is 1.90uVrms according to the noise table 7-1 in the ADS131M02-Q1 datasheet, so the current measurement resolution is 1.90uVrms/100uohm=19mArms (100uohm is your shunt resistor value), the peak-peak value is 19mArms*6.6=125mApp. You are expecting much lower noise and errors from the ADC for your high requirement. post-processing (e.g. FFT) in the software on your microcontroller may help but it is still a big challenge.

    Regards,

    Dale

  • Hi Dale

    Now here's the problem, when I'm not doing any calibration, the measurements are very different from the expected values,

    When Vin = 0.3185V, COAL_Regs and GCALRegs do not write values, Vout=0.3083V;

    When Vin = 0.6367V, COAL_Regs and GCALRegs do not write values, Vout=0.6021V.

    Is that reasonable?

    If calibration must be done first, whether to calculate offset or gain error first.

    Kind Regards.

  • HI kaixin,

    For offset calibration, did you short input on the input pin of ADC, use internal short or short your system's input? what's the code your got?

    For gain calibration, how did you do the calibration and what's data you got and programmed to GCALRegs?

    As I said, you will still not able to achieve the accuracy you wanted even after calibrate out the offset and gain error.

    BR,

    Dale