Post Go back to editing

AD9081 Example of PFILT for only 1 of the 4 ADCs?

Category: Hardware
Product Number: AD9081
Software Version: ACE 1.25.3217.1403

I want to use the programmable filter (PFILT) on the AD9081, but I'm unable to load filter coefficients to 1 ADC only. I'd eventually like to upload four unique 96-real tap filters to use separately on each of the 4 ADCs but am seeing that ADCs get paired or the spectrum has many spurious tones (details below) when I try to upload only 1 set of filter coefficients.

I'd appreciate if someone could explain what I am seeing, suggest things to try, or if I could get a separate example engaging PFILT on 1 ADC only on the AD9081

  • I've read through the guide to PFILT for the AD9082 and tried to do a similar setup.
  • Here are the ACE settings I am using with the AD9081
  • Here is what I see before programming PFILT. I am inputting a 100 MHz input at -16.5 dBm into ADC0. ADC 1, 2, 3 are terminated to 50 Ohms
    • Channel 0 sees 100 MHz in (as expected) and Channels 1, 2, 3, don't have a signal (as expected) 
  • I programmed the PFILT with the following settings and an impulse response but see that Channel 0 and 2 are "linked" or the spectrum looks strange
    •   
      7fff
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      0000
      
    • I programmed an impuse filter, which I uploaded as text file (csv didn't upload well here)
    • After programming the impulse filter, there's an output on both Channel 0 and 2 even though there's only a 100 MHz signal on ADC0. ADC1, 2, 3 are terminated to 50 Ohms
    • The Channel 0 spectrum looks correct. I expected Channel 2 to look like Channels 1, 3 though (nothing on the output)
  • I reprogrammed the chip, resetting the PFILT. I then set the PFILT with the same impulse filter but set PFILT Control Page to "PFILT ADC Pair0"  rather than "PFILT ADC Pair All". Though the filter only gets applied to ADC0, there are may spurious tones
    • There is a 100 MHz signal in the ADC0 and ADC1, 2, 3 are terminated to 50 Ohms
    • Programming just "ADC Pair 0" leads to an output on Channel 0 only, but the output looks strange and there are unexpected tones replicated
    • Channel 1, 2, 3 (right) look correct as there's nothing on the output
  • After programming PFILT Control Page for "PFILT ADC Pair0" above, I programmed it for Pair1. So, without reprogramming the whole chip I setup PFILT for Pair 0 then Pair 1. 
    •  
    • There is a 100 MHz signal in the ADC0 and ADC1, 2, 3 are terminated to 50 Ohms
    • Now the output on Channel 0 looks like the same spectrum on Channel 2. This looks like when I set PFILT Control Page to "ADC Pair All".
    • The spectrum looks correct, but I don't know why Channel 0 and 2 are linked together.


Edit - Updated post category
[edited by: youngpines at 8:22 PM (GMT -4) on 3 May 2022]
  • Hi youngpines,

    I'm sorry for the slow response. I'll try to find some answers for you.

    Doug

  • Hi youngpines,

    Thanks for sharing your configuration. I set up my bench to match yours.

    1. With your initial setup (Pfilter ADC Pair All), mine matches yours. The test signal applied to ADC0 appears in the ACE FFTs for channels 0 and 2.
    2. With Pfilter ADC Pair0, I do not see what you observed. I see the same thing as in the first case.
    3. With Pfilter ADC Pair1, again I see the same thing, i.e. for all three settings I saw the same thing.

    When you repeat configuration 2, do you get the same result? Is there anything else you changed?

    I'm still checking about why Channel 0 and Channel 2 show the same signal.

    Doug

  • Hi Doug,

    Thanks for the update, and I appreciate you trying this out on the bench. Between your step (1) and (2), I reprogrammed the AD9081.

    I retried this on the bench today here's what I saw the following

    1. I did your step (1, programming "ADC Pair All") and confirmed what we've seen
      1. Ch0, 2 spectrum is correct but they are linked
      2. Ch1, 3 shows nothing
    2. I power cycled the ADS9+AD9081 and then reprogrammed the AD9081 with the same settings.
      1. I confirmed the spectrum looks like the pre-programming baseline
        1. 100 MHz, -16.5 dBm to ADC0. ADC1, 2, 3 are terminated to 50 Ohms
        2. Ch 0 has 100 MHz as expected and Ch1, 2, 3 have nothing as expected 
    3. I did your step (2, programming only "ADC Pair 0" with the impulse filter) and see spurs
      1. Ch0 has spurs, Ch 1, 2, 3 show nothing
      2. The same 100 MHz is going into ADC0 while ADC1,2,3 are terminated to 50 Ohms
    4. I did your step (3, programming only "ADC Pair 1" with the same impulse filter) and see spurs
      1. Ch0, 2 has spurs, Ch 1, 3 show nothing
      2. The same 100 MHz is going into ADC0 while ADC1,2,3 are terminated to 50 Ohms
    5. I did your step (2, programming only "ADC Pair 0") again and now the spectrum is correct but Channel 0, 2 are linked
      1. Ch0 shows the correct spectrum, Ch 1,3 show nothing (as expected). Ch2 is linked to Ch0 for some reason (unexpected)
      2. The same 100 MHz is going into ADC0 while ADC1,2,3 are terminated to 50 Ohms

    Thanks!