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.

ADS1278: How to eliminate ringing in the step response of the ADS1278 by shaping the input signal?

Part Number: ADS1278
Other Parts Discussed in Thread: ADS131M08, ADS131E08

I use the ADS1278 in an application where the input signal changes in steps.

The ADC input signal is a response of a measurement device to an excitation signal created with a DAC.

The excitation signal can be changed.

I’m interested in reducing or eliminating the ringing in the step response of the ADS1278 by ‘rounding off’

the corners of the excitation signal.  A simple first order low pass filter eliminates ringing after the edge

but some ringing before the edge remains.

 

What I’d like to know is how to ‘slow down’ the excitation signal so the edge times are as short as possible

but are not causing ringing in the response of the ADS1278.

 

  • Hello,

    Our ADS1278 expert is out of the office until Wednesday and will be available to support you then.  Until he returns, could you provide more details about your application including screen shots of the input signals you're describing and also the captured data from the ADS1278?

  • Sampling rate of the ADS1278 in the following examples was 10kSps, the duration of each step in the test signal 5ms.

    This is the step response to the unfiltered step signal fed into ADS1278 input.

    This is the response to the low pass (1st order) filtered step signal. The 3dB cutoff frequency of the low pass filter

    is about 1.4kHz

    Most of the ringing after the step change was eliminated but there still is ringing before the step change

    This picture shows the response of the ADS1278 to same filtered signal, the vertical scale is different.

  • Hello,

    I feel this is very likely the digital filter response to the step input you apply to the inputs.  Figure 66 in the datasheet shows the ringing that occurs at both the beginning and the end of the step input.  The conversion response starts to 'ring' immediately after the step input (conversion 1 in the below picture), but is very small at first and not easily observable when plotted from 0 to 100%.  After about 35 conversions, you can start to see the response changing, and by 40 conversions the response is within 5% of the final settled value.

    This ringing is proportional to the rate of change on the input signal.  The RC filter response changes slowly as it approaches the final value, but the initial change is much more abrupt (slower with the filter than without, but still enough to cause significant ringing).

    I am not aware of a simple analog circuit to shape the response on the leading edge, but you did mention using a DAC in the system.  If you slowly change the output over the first few microseconds before ramping up to the final value, this should help with the ringing.  If you can do this on both the beginning and end of the 'ramp', you should be able to reduce the ringing on both edges without adding additional components.

    Regards,
    Keith Nicholas
    Precision ADC Applications

  • Hello,

    I would also like to add that the digital filter requires 76 conversion cycles to fully settle (78 if using high resolution mode).  You mentioned you are sampling at 10ksps, so it will take 7.8mS to fully settle.  Since each step is only 5mS wide, you will only get around 1% settling.

    Regards,
    Keith

  • Thank you for your replies.

    I tried slowing down the edges in a few different ways. The only way I could eliminate

    the ringing after the edge was by using a low pass filter of a specific cutoff frequency.

    This has very little affect on the beginning of the edge though. I can apply the same method

    to the signal generation (DAC) to round off the beginning of the edge as well and minimize the ringing

    before the edge. Although I was hoping that based on the ADS1278 filter characteristic

    you would be able to tell what the optimum (fastest) shape of the fastest step signal that doesn't produce

    ringing is.

    Without knowing the filter characteristic, it's not a trivial task. I tried slowing down the edge

    and rounding off the corners in a few ways with no good effects. 

    There may be a faster signal than the low pass filtered step that doesn't cause significant ringing.

  • Hello,

    I am checking to see if we can provide a little more guidance on the rounding of the edges.  I do not expect to have an update until early next week.

    Typically, for step inputs, I would recommend a different ADC with a SINC filter option that settles much quicker than the wideband digital filter implemented in the ADS1278.  Are you open to looking at a different ADC, or do you need the higher sample rate of the ADS1278?

    Regards,
    Keith

  • I am looking at a couple of alternative ADCs but I'm not giving up on the ADS1278 yet.

    It has certain advantages. Since we have control of the input signals, there is a chance

    we could avoid the overshooting.

  • Hello,

    I have not found any additional information regarding the details on the step response.  If I do find something, I will reply back to this thread.

    Thanks,
    Keith

  • Would you be able to consult the people who designed the digital filters in the ADS1278?

    They would probably know that.

    BTW, is there a similar ADC to the ADS1278 that's using the Sinc filters?(other than the ADS131 family)

  • Hello,

    I am working on this and will let you know.

    Thanks,

    Keith

  • Hello,

    I checked with the filter designer, and he is not aware of any simple mathematical relationship that would predict the magnitude of overshoot vs the rate of change on the step input.  This may have been figured out by someone in the industry, but this is not something that we have available.

    In addition to the ADS131 family, we also have the ADS129x family; 24b delta-sigma, sinc filter options, with data rates up to 32ksps.

    Regards,
    Keith

  • Were you able to determine if it's possible to shape a step signal to it doesn't cause ringing

    in the ADS1278 response? 1st order low pass filtered signal is the best fit I could find experimentally

    but it slows the signal significantly. I wonder if there is a better method.

    Could you recommend an alternative ADC with similar features to the ADS1278 but with Sinc filters?

    I could not find one made by TI. The closest fit missed an external reference input.

  • Hello,

    I was not able to dig up any additional information on input shaping for the ADS1278 digital filter.

    Both ADS131M08 and ADS131E08 are 24b, 8 channel devices, and support internal or external voltage references.  Both use Sinc3 filters for fast settling.  ADS131M08 supports data rates for each channel up to 32ksps, and ADS131E08 supports data rates for each channel up to 64ksps.

    Regards,
    Keith