• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 82 subscribers
  • Views 543 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.

TMS320F28375D: ADC VREF Pin Connections..

Newman
Expert 6005 points
Part Number: TMS320F28375D
Other Parts Discussed in Thread: OPA350, OPA320

Hi Team C2000,

Reference (https://e2e.ti.com/support/amplifiers/f/14/p/761547/2814846#2814846 ).  The TMS320F28375D TRM (http://www.ti.com/lit/ug/spruhm8g/spruhm8g.pdf ) provides some advice WRT driving the ADC VREF pins.  The suggestion is to drive the 4x VREF pins from either:

1. A single reference connected to 4 separate OPA350 devices/packages.

2. A single reference connected to 2 separate OPA350 devices/packages connected to VREFA/C & VREFB/D

WRT getting the best/aligned performance form the 4X ADC/VREF peripherals..  Questions..

1. Would it be better/same/worse to drive the VREF pins from a single device/package w/ 4xOpAmps or 4 separate OpAmps in 4 separate devices/packages?

2. Would it be better/same/worse to drive the VREF pins from a single reference connected to a single OpAmp.. w/ 4x filtering paths (see https://e2e.ti.com/support/amplifiers/f/14/p/761547/2814846#2814846 )?

Thanks, Merril

  • 0
    TI__Guru 60865 points
    Hi Newman,

    One VREF per op-amp channel consistently has a small but noticeable performance increase for SNR and THD vs. two VREFs sharing an op-amp channel (per our characterization using OPA350 and OPA320). Because of that we'd expect all the VREFs driven by a single op-amp channel to have measurably worse performance than one or two VREFs per channel.

    It might make sense to use a single op-amp channel if you really care more about ADC-to-ADC gain error matching or you have significant cost/area constraints.

    Note that if you drive multiple VREFs per op-amp channel you will probably need to pick a slightly higher snubber resistor (Ra, Rb, Rc, Rd) to ensure op-amp stability vs. what is on the C2000 EVMs. This is because each additional filter circuit is an additional capacitive load on the op-amp output.
  • 0 in reply to Devin Cottier
    TI__Expert 6005 points
    Hi Devin,
    Thanks.. WRT “what is on the C2000 EVMs”.. which specific EVMs are you referring to?

    Is there a reason for the VREFA/C & VREFB/D pairing recommend shown in the TRM (www.ti.com/.../spruhm8g.pdf ) in figure 10-24 (as opposed to say a VREFA/B & VREFC/D pairing)?

    WRT your comment “One VREF per op-amp channel consistently has a small but noticeable performance increase for SNR and THD”.. what magnitude of “small but noticeable performance increase” do we expect to see?

    WRT “Because of that we'd expect all the VREFs driven by a single op-amp channel to have measurably worse performance than one or two VREFs per channel.” I assume this is WRT SNR and THD.. what if we are looking for aligned performance across the 4x ADCs? Seems a single VREF.. single OpAmp and 4x filters would be best for this?

    Thanks, Merril
  • 0 in reply to Newman
    TI__Guru 60865 points
    Hi Newman,

    The two possible EVMs are the F2837xD Launchpad and F2837xD ControlCard. Both should have similar example VREFHI driving circuits.

    The pairing choices are based on the pin placement on the devices; A and C are close together in the pinout, as are B and D.

    Typical SINAD increases by 1dB-2dB (in 16-bit resolution mode) above the datasheet by driving each VREFHI separately. This is, I think, mostly due to an SNR increase of about the same amount, since SINAD on this device is dominated by SNR performance.

    Yes, this is affecting the AC specifications (SNR, THD, etc.). You can directly determine the increased ADC-to-ADC gain error based on the Vos specification of the driving op-amps. If the Vos specification is lower than the device ADC-to-ADC gain error in the device datasheet then there will be no noticeable difference when using multiple op-amps.