• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Subscribers 46 subscribers
  • Views 670 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

minimum input signal frequency to ADS5404 ADC

Amos Agmon
Prodigy 30 points
Other Parts Discussed in Thread: ADS5404

Hi,

I assume that ADC model ADS5404 (and also other similar ADCs) has some DC-block frequency-response applied to the analog input signal, and as a consequence its frequency response in the DC area has the form of a notch filter.

a) Is this assumption correct?

b) If it is correct I am interested in the spectral rejection are, i.e. the 3dB bandwidth of the attenuated frequency band around DC.

Thanks,

Amos

  • 0
    TI__Mastermind 29220 points

    Hi,

    The ADC itself does not have any AC-coupling or DC-blocking internal to the ADC.  The ADC is good all the way down to DC.  (but there may be some DC offset to the signal that shows up in bin 0 of the FFT.)

    The EVM for this ADC and nearly all of our high speed ADCs will have AC-coupling (and hence DC-blocking) in the input circuit of the EVM before the ADC.  Our default configuration for an EVM is to have transformer coupling on the input signal so as to be able to take a single ended signal from a piece of test equipment and convert that to a differential signal that can then be biased up to around the desired common mode voltage for the analog inputs.  And amp device could be used to do that, but then the noise figure of the amplifier would influence the observed performance and you would only be able to see the combined performance of amp plus ADC adn not the performance of the ADC itself.

    Sometimes a balun is used instead of a transformer, and baluns do not provide AC coupling and so there are AC coupling caps after the balun on the ADS5404 EVM.  The lower limit for getting an input frequency into the ADS5404 EVM will be the lower end bandwidth of the ETC1-1-13 balun. http://www.macomtech.com/datasheets/ETC1-1-13.pdf  Keep in mind that the EVM has two such baluns in series, so the 3db point of the bandwidth will reflect two of these baluns in series.

    If you want to see the performace of the ADC EVM around DC, then the EVM would have to be modified to bypass the baluns, install the second SMA connector for the other side of the differential input, adn then bring a differential signal straight into the board.  There are provisions in the EVM layout to allow you to do exactly this.  But then you would be responsible for getting the signal biased to the desired VCM level and making sure the phase and amplitude is balanced between the two sides of hte differential signal.

    Regards,

    Richard P.

  • 0 in reply to Richard Prentice
    Prodigy 30 points

    Hi,

    Thank you Richard, for the informative and detailed answer!

    Addressing your note that:

    " there may be some DC offset to the signal that shows up in bin 0 of the FFT "

    Is this offset expected to be constant over time? or at least will it have some very-slow wondering pattern?

    For example: if I receive an OFDM (Orthogonal Frequency Division Multiplexing) signal with information on the DC bin of the FFT, will I be able to estimate this constant (or slowly varying offset, reduce it from the DC bin value and reconstruct the true DC value?

    Thanks,

    Amos

  • 0 in reply to Richard Prentice
    Prodigy 30 points

    And yet anothr related question:

    If using a differential input signal is not an option in my system, Then to minimize the damage to the signal what do you think about modifying the EVM by replacing the ETC1-1-13 by would you recommend to replace the existing balun with a different one, supporting lower frequencies:

    E.g.:

    - ETC1-1T-2TR (ETC1-1T-2TR , my input signal is limited to ~200 MHz)

    - MABA-009126-ET11SM (https://www.macomtech.com/datasheets/MABA-009126-ET11SM_V2.pdf)

    Thank you,

    Amos

  • 0 in reply to Amos Agmon
    TI__Mastermind 29220 points

    Hi,

    The DC offset will be related to the typ and max offset spec listed in the AC parameters section of the datasheet.  There will be device to device variation, so one device might have very little DC offset and show nearly nothing in bin 0 of the FFT while another might reflect the datasheet max DC offset.  And in general, there are many little things in the design of the device that contribute to offset error, so there can be drift of the offset over time - especially as temp of the device drifts over time.

    For this particular device, each channel of the ADC is itself really two interleaved channels with one clocked on rising edge and the other on faling edge.  If two interleaved channels have *different* offset from each other or different gain then there can be an interleaving spur inthe resulting FFT. For that reason there is digital logic to detect and correct for DC offset over time.  So this device has two specs for offset - one with autocorrection on and one with autocorrection off.  (Page 7 - offset error)

     Regards,

    Richard P.

  • 0 in reply to Amos Agmon
    TI__Mastermind 29220 points

    Hi,

    there is not one part number of transformer or balun that is perfect for all possible applications, so for an EVM we try to select the magnetics to reach the broadest range of customers and then allow for mods to reach the rest.  it is very reasonable to replace the baluns with something more suitable.  We often use the ETC1-1-13 when there is the possibility of higher input frequencies.  We often us a Coilcraft WBC1-1 when we don't expect the input to be above 300M or so, adn the Coilcraft is good for down below 1M.   There are other choices that can get down to 100's of KHz but we don't have EVMs that would have this as their first choice.

    If you look at the solder pads for the transformers/baluns on the EVM, you will see a funny structure that looks like 6 pads (two primary two secondary two centertap) that are spaced larger and within that the same 6 pads spaced closer.  This lets us accept magnetics from the ETC line of devices, magnetics from the Coilcraft WBC line, and some larger Minicircuits ADT or JTX lines.

    But there is no choice of magnetics that have bandwidth down to DC.  There will be a lower cutoff frequency somewhere.  You just get to choose where that is by your choice of available magnetics.

    Regards,

    Richard P.