RFSoC ADC and DAC full scale levels 6dB low

[mmatusov]

Spoiler

Hello,

Using the RF Analyzer we are finding on a ZU28 that the ADC goes over-range when the received digital signal is at about -6 dBFS and it becomes non-linear if we try to increase signal any further. We are also finding that at 0 dBFS the DAC output seems to be about 6 dB below the expected level. 

Here is our configuration:

We tried enabling QMC gain but while it increases the absolute level we don't seem to gain any of the lost dynamic range. Are we missing something?

Thanks,
/Mikhail

[zhendon]

Hi @mmatusov 

What kind of signal is tested here? If it is a bandwidth signal, -6dBFS would only be a signal power on one bin and it might go over-range due to the PAPR. 

[mmatusov]

Hi @zhendon ,

The signal is a simple CW.

/Mikhail

[zhendon]

Hi,

If you are monitoring the DAC output with XM500 daughter board and looping DAC output back to ADC, you need to consider the board insertion loss + signal frequency roll-off which might be 6dB (depending on your signal frequency).

[mmatusov]

Hi @zhendon ,

We are using a 3rd party board with the analog input/output circuits based on Xilinx reference designs. But I don't think this is relevant for our discussion, at least not with regards to the ADC side. Here's a capture of the input CW at a level of 0.1 dB below of the ADC Over Range interrupt threshold. As you can see the amplitude values are barely above 16000. I would expect them to approach 32000 for a 16-bit signal close to the full scale. 

Thanks,
/Mikhail

[zhendon]

Hi @mmatusov 

I can see from your screen shot that you are enabling mixer so there is a 3db attenuation included in mixer output. You can find a detailed description of this part in PG269. 

[mmatusov]

Hi @zhendon ,

Yes, the mixer is enabled but we are in the R2C mode, so according to PG269 the mixer output attenuation is 0 dB. Am I wrong?

Thanks,
/Mikhail

[ctelfer]

Hi @zhendon,

Another test we performed was to disable the fine mixer, we used the coarse mixer only, and we still see the same issue.  This suggests the 3db attenuation you pointed out is not causing the low level.  Do you have other ideas about what could be causing this?

Thanks, Colin