Part Number:ADC32RF45
I am attempting to debug an issue we are seeing with an ADC32RF45 device. We have a custom board that has two ADC32RF45 devices on it giving us a total of four channels. We have manufactured multiple boards without issue. On one particular board, I am seeing a problem on one of the four channels.
Problem description:
Channel A of the second ADC is exhibiting noise spurs when running in DDC mode. I have a screen capture of the FFT spectrum from channel A below. The configuration for this test was:
990MHz DDC center frequency
3GHz ADC sample clock
Decimate by 4, so 750MSPS output sample rate
Test tone at 990MHz
![]()
To determine whether the issue was related to the analogue input of the ADC, I configured the DDC of channel B to the same configuration as channel A above and then enabled the DDC mux so that the channel B DDC was operating on channel A data. I then captured the data from channel B. I have a screen capture of the FFT spectrum from channel B below:
![]()
As can be seen, the noise spurs are no longer present when I use the DDC in channel B to process the data from channel A.
To further verify this fact, I configured the ADC32RF45 into bypass mode (bypass DDC). In this mode, I am capturing raw ADC samples from the ADC32RF45. Both channel A and channel B operate normally and do not contain the spurs visible in the first screen capture. This confirms the following:
- the issue is not related to the channel A ADC input circuitry
- the issue is not related to the channel A ADC interleaving and offset correction
- the issue is not related to the channel A JESD204B interface
From this testing, it can be seen that the only difference between a normal FFT spectrum and the noisy FFT spectrum is the channel A DDC circuitry inside the ADC32RF45.
I am not sure how this is possible though.
I did more testing to try and understand the behaviour of why channel A DDC is introducing noise spurs.
1) I repeated the above testing at a 1000MHz DDC center frequency. The following screen capture of the FFT spectrum is with channel A DDC operating:
1000MHz DDC center frequency
3GHz ADC sample clock
Decimate by 4, so 750MSPS output sample rate
Test tone at 1000MHz
![]()
By enabling the DDC mux again, I used channel B DDC to process the same data and this is a screen capture of the FFT spectrum:
![]()
Again you can see that simply by using channel B DDC circuitry, the noise spurs are gone. It can also be seen that the location of the spurs changes depending on the DDC center frequency.
2) The noise spurs are not present if there is no input signal.
3) I tried testing with all four NCOs set to the same frequency and set to different frequencies. It didn't have an impact on the noise spurs.
I am completely stumped as to how only the channel A DDC circuitry could be introducing noise spurs.
a) Is there a specific input power pin of the ADC32RF45 ADC that powers only the channel A DDC circuitry?
b) Is it possible for the channel A NCO clock to be 'unstable' while the channel B NCO is operating as expected?
c) Is there any other possible explanation for why the channel A DDC could be introducing noise spurs while channel B DDC operates as expected?
Lastly, I don't believe this is a configuration issue because we have multiple other boards where channel A DDC operates normally.
I am hoping you will be able to provide some further insight!
Thank you.
