We examined the spectrum of the received signal of AD9279, found that there are frequency peaks on the spectrum, and peak frequency varies with different sampling rate, for example(The inputs of AD9279 are floating)
- ADC sampling rate = 20MHz, peaks around 5~10MHz
- ADC sampling rate = 50MHz, obvious peaks at 10 and 20 MHz
- ADC sampling rate = 70MHz, obvious peaks at 27 and 31 MHz, also on 8 and 23MHz
- ADC sampling rate = 80MHz, obvious peaks at 19 and 39 MHz
To find out where the frequency peaks come from, I have measured the ADC input(and all the possible nets of AD9279), but I didn’t find these peaks on scope, please see below picture. (there are peaks at 20 and 60MHz, but they belongs to EMI interference, actually the two peaks present even if scope probe does not measure anything)
I had used the EVB board of AD9279 to confirm this problem, and concluded that this problem is not present on EVB board.
I had also suspect that we do not set AD9279 well, so I tried these register settings but no finding till now
- 0x02 : CHIP_GRADE
- 0x0F : FLEX_CHANNEL_INPUT (these peaks can be reduced by anti-aliasing filter setting but still there)
- 0x2B : FLEX_FILTER, enable automatic low-pass tuning
I have no idea on where these frequency peaks come from for a long time, maybe it is caused by bad PCB layout, but I am kind of feel like these peak comes from AD9279 itself, mainly because I cannot measure these peaks on scope, it would be appreciated if you can give me any instruction about this.
The attached file is our schematic of AD9279.