• State TI Thinks Resolved
  • Locked Locked
  • Replies 12 replies
  • Answers 4 answers
  • Subscribers 47 subscribers
  • Views 2232 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.

ADS8861: Anti-aliasing filter for a sampling ADC SAR

Keith Keller
Expert 6240 points
Part Number: ADS8861
Other Parts Discussed in Thread: TINA-TI, OPA350, OPA320

Hello,

When designing an anti-aliasing filter for a sampling ADC SAR, is it preferred to have the final filter/amplifier stage directly drive the adc input or is it good/common practice to insert a buffer stage with either unity or more gain to handle the s/h input separate from the filter design?

I’ve done it both ways, and haven’t had issues to date with either, but do wonder what is the ‘best’ approach?

Thank you, Keith

  • 0
    TI__Genius 17737 points

    Keith,

    In general it is best to have a wide bandwidth buffer drive the SAR input. In this case the first stage can be optimized for offset, and noise.  The second stage is just optimized for ADC drive and settling.  Sometimes when using amplifiers in gain it can be challenging to make the ADC settle.  The main reason why you would use only one amplifier stage is for cost / size optimization.  ADC precision labs (link below) discusses driving a SAR and testing settling using TINA SPICE.  https://training.ti.com/ti-precision-labs-adcs

     

    Art

  • 0 in reply to Art Kay
    TI__Intellectual 1205 points
    Keith:

    Just to add to Art's comments, an anti-aliasing filter can be incorporated into the design of the first stage amplifier. (This would be my recommendation as a best-practice.) As Art mentions above, the second amplifier is a wide bandwidth buffer. Between the buffer amplifier and the SAR ADC is almost always an RC combination that should not be mistaken for an anti-aliasing filter. Typically, the frequency cutoff of this RC circuit is way too high for effective filtering. Instead, its purpose is to help the amplifier drive the switched capacitor load presented by the SAR input sampling circuits. More details on that subject can be found in the link that Art shared.
  • 0 in reply to Art Kay
    TI__Intellectual 2850 points

    Hi Art,

    Thanks for your reply.

    I have a question about resistance accuracy of ADS8681EVM. As you can see below, R5/ R3 just has 0.1% accuracy.

    I think this will make the total system equals less than 10 bit ADC? So could please tell me how to calculate the DC performance of this EVM.

    BTW, could please explain why do we need to add the R7/ R8 and how to design the value of them?

    Thanks so much.

    Marc

  • 0 in reply to Bryan McKay
    TI__Intellectual 2850 points
    Hi Bryan,

    Thanks for your reply. Could please share a reference circuit to show how combine an anti-aliasing filter with the first stage amplifier？
    It's great if you could share the design details about your circuit.
    Thanks so much.

    Marc
  • 0 in reply to Marc Liu
    TI__Intellectual 1205 points
    Marc:

    When we think of DC performance of an ADC, we think of parameters like Negative Full Scale Error (NFSE), Positive Full Scale Error (PFSE), Bipolar Zero Error (BPZE), Differential Nonlinearity (DNL) and Integral Nonlinearity (INL). All of these parameters, when measured, include the errors of the entire signal chain. So the performance of the fully differential amplifier (FDA) circuit you have shown above, and the ADC itself, are both important.

    The resistors R3, R4, R5, and R6 are the gain-setting resistors for the FDA, and their accuracy would have an effect on the NFSE and PFSE of the circuit. Their matching would have an effect on the BPZE and linearity. So you are correct, selection of these resistors is important for overall performance.

    Keep in mind that the tolerance tells you the maximum and minimum values of the resistors. A large sample of resistors would have values that are normally distributed, with the mean value of the distribution at the nominal value. The maximum and minimum values are at +/- 3 standard deviations from the mean.

    As for resistors R7 and R8, they provide part of the resistance for the output RC filter. Resistance outside of the feedback loop also adds to DC errors at the output of the FDA circuit. So part of the resistance is brought inside of the feedback loop to help the amplifier accurately drive the capacitive load. However, keep in mind that R7 and R8 are in series with the feedback resistors R5 and R6. That means they will have a large effect on the gain of the FDA circuit. In fact, it is more than 10x of the effect of the 0.1% resistor tolerance.

    In designing the value of the output RC components, how much resistance to pull inside of the feedback loop, adjusting the values of the feedback resistors, and so forth: TINA-TI is your best friend.

    Also, for more information, check out Precision Labs for Op-Amps, at this link:
    training.ti.com/ti-precision-labs-op-amps

    For specific questions, as always, feel free to post here on E2E.

    --Bryan
  • 0 in reply to Marc Liu
    TI__Intellectual 1205 points

    Marc:

    For implementing an Anti-Alias filter, or any input filter for that matter, we would recommend Filter Designer (an online tool on TI.com) to design the correct filter for your application.  Following the filter, a buffer amplifier would be recommended, which would have an RC circuit on its output, feeding the input of the SAR ADC.  This RC circuit should not be confused with an anti-alias filter.  I am posting two very general TINA-TI circuit examples, to be used for a topology reference only.   One has an RC (anti-alias) filter at the input of the ADC driver amplifier.  The second has a second-order active filter on the input of the ADC driver amplifier.

    Have fun with these!

    --Bryan

    /cfs-file/__key/communityserver-discussions-components-files/73/ADS8568_5F00_LowPass_5F00_OPA350.TSC

    /cfs-file/__key/communityserver-discussions-components-files/73/ADS8568_5F00_OPA350LowPass_5F00_OPA350.TSC

  • 0 in reply to Bryan McKay
    TI__Intellectual 2850 points
    Hi Bryan,

    Thanks for so detailed reply. That's a great help to me.

    Best regards,

    Marc
  • 0 in reply to Marc Liu
    TI__Intellectual 1205 points
    Marc:

    I am glad it helped. If you have any more questions, we will be here waiting for you.

    --Bryan
  • 0 in reply to Bryan McKay
    TI__Expert 6240 points

    Hello,

    Could you please look at these modifications to your circuit using the OPA350 to drive SAR converter on a TMS320F28xxx DSP?

    We’ve added:
    1. a filter / decoupling cap on the non-inverting input
    2. bucket cap on output
    3. changed the feedback loop slightly

    Thank you very much, Keith

    SAR drive with OPA350.TSC

  • 0 in reply to Keith Keller
    TI__Intellectual 1205 points
    Keith:

    In the example circuits that I made for you above, the OPA350 is (accidentally) referenced in the file name, but if you notice, I used the OPA320 in the circuit. The OPA350 is a good amplifier, but you should be careful of the crossover distortion region when you design with that amplifier. The OPA320 does not have crossover distortion, so that would be the recommended amplifier if you need rail-to-rail linear performance. However, it has slightly lower gain bandwidth than the OPA350, so it is not exactly a drop-in replacement. I apologize for the confusion.

    Your circuit looks like an inverting attenuator, allowing the input to accept a signal in a range of about -8V to 8V. The output would then swing from about 3.3V to 0V. The capacitor in the feedback path makes the amplifier an active low-pass filter (with about 2MHz cutoff), and appears to be very stable.

    We will take a closer look at your circuit and get back to you shortly with a couple of recommendations. In the meantime, you might get some better advice from the Precision Op-Amp team and the DSP team.

    --Bryan
  • 0 in reply to Bryan McKay
    TI__Intellectual 1205 points

    Keith:

    I have modified your circuit slightly in the attached TSM file.  Your input model for the ADC looks right, but the LSB error output was not working for me, so I changed that to an analog meter.  Other than that, I used a different feedback cap and different "Charge Bucket" filter values, and changed to an OPA320.   It is lower cost and provides linear performance with enough bandwidth over the entire output range, and settles to well within 1/2 LSB of your converter.

    /cfs-file/__key/communityserver-discussions-components-files/73/SAR-drive-with-OPA320.TSC

    --Bryan

  • 0 in reply to Bryan McKay
    TI__Intellectual 1205 points

    Keith:

    In the attached TSC, you will find your circuit with a few recommendations.  This circuit uses some different values for the feedback cap and charge bucket filter, as well as replacing the OPA350 with an OPA320.  The OPA320 is lower cost, and settle the value on the input of your converter to well within 1/2 LSB's.

    /cfs-file/__key/communityserver-discussions-components-files/73/1033.SAR-drive-with-OPA320.TSC

    --Bryan