• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Subscribers 83 subscribers
  • Views 1731 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.

TMS320F28377S: Choosing the capacitor at the input pin of C2000's ADC

Sheldon He
Expert 8585 points
Part Number: TMS320F28377S

Hi all,

My customer is evaluating F28377's ADC and now they have some question on choosing the capacitor.

Several years ago, they used to use a 30pF capacitor as CIN but didn't get a very good result, so now they want to use a 1nF capacitor as CIN. They want to know if there are any side effects using a capacitor of such value(sampling rate 40khz).

Thanks!

  • 0
    TI__Mastermind 33475 points
    Hi Sheldon,

    Without information about the type of input signal/frequency and the source impedance and how long the signal is sampled (ACQPS or SH), it is not straightforward to say what the effect of a higher CIN cap would do. Generally, using higher CIN caps would stabilize the signal and is safe to do this on DC or very low frequency signals however, if the signal being sampled is dynamic, in the order of a few kHz, having higher CIN might filter or attenuate the signal and converting it in ADC may not give the true digital representation.

    The different resistances (RIN, RSH) and capacitances (CIN, CSH) in the model you have above would eventually reduce to an RC network with an equivalent time constant. You would want to settle the sampled signal to within 0.5 of an LSB which is roughly 7.7 of the equivalent RC time constant.

    One way to determine if the 1nF CIN is effective is to do continuous conversion on a portion of the signal, especially if this is dynamic, and store the conversion results in an array while also capturing the input signal with a scope. If the ADC converted results do not resemble the captured input signal, then recommend using different value of CIN and SH.

    Regards,
    Joseph
  • 0 in reply to Joseph Casuga
    TI__Expert 8585 points
    Thanks for your reply Joseph, I will forward this link to my customer.
  • 0 in reply to Sheldon He
    TI__Mastermind 33475 points

    Hi Sheldon,

    Section 10.3.2 gives a good approximation of how to derive ACQPS if the impedances of the system are known. You are correct, the approximate number of time constants needed for 0.5LSB, as stated in formula #1 should be ~9.  Formula # 3 should have “T(S+H)” instead of “t”.  Likewise, there is an error in the example in that R should be = 550 ohms, not 550kohms.  We will have these corrected in the documentation.

                    For the impedances that you have provided, T is 9.01 , R = 925ohms and Ch = 14.5pF.  Using formula #3, T(S+H) = (9.01)*(925 * 14.5pF) = 120.8ns.  Not sure what the SYSCLK frequency is but assuming it is 200Mhz, ACQPS will be derived using formula # 4 ins section 10.3.2:

     

                    T(S+H) = (ACQPS+1) * (1/SYSCLK);

     

    ACQPS selected should be 23 (nearest whole number)

     

    Best regards,

    Joseph