04-20-2020 01:17 AM
Hello everyone,
I saw that in PG269 the AXI4-Stream Interface in which the number of AXI4-stream and the width of each bus depends on the bandwidth required by the mode of operation. But I am still confuse about this problem:
Single RF-DAC Input - 32 bits
As we can see in this figure, at 2 ns (~500 MHz) we have 2 samples Q0 and I0, therefore we have just only 1 GSPS (instead of 4 GPSP in the figure). May be I don't understand something!
Also, I don't know why we have s00_asix_tdata = 32 bits!
Can you please explain for me.
Thanks with best regards
04-20-2020 08:52 AM
For real data, one sample is 16bits.
But for I/Q data, one sample includes both I and Q. So one sample is 32bits.
And the RFDC can be configured for different data widths.
So for example:
If dwidth=32bits, (8x interpolation), 4Gsps, then the axis clk has to be 500MHz.
If you change RFDC dac axis to 64bits, then you could reduce clk to 250MHz.
04-20-2020 08:52 AM
For real data, one sample is 16bits.
But for I/Q data, one sample includes both I and Q. So one sample is 32bits.
And the RFDC can be configured for different data widths.
So for example:
If dwidth=32bits, (8x interpolation), 4Gsps, then the axis clk has to be 500MHz.
If you change RFDC dac axis to 64bits, then you could reduce clk to 250MHz.
08-18-2020 06:54 AM
Thanks @johnmcd for your correct answer,
But I am still confuse about this figure as well as the figure 79 of PG269 V2.1.
According to the document:
PLdatarate = (DACdatarate * IQMode) / InterpolationRate = 4000*2/8 = 1000 MHz
PLFclock* PLnumword = PLDatarate
=> PLFclock = PLDatarate/PLnumword = 1000/1 = 1000 MHz
Where IQmode = 2 if I/Q data and =1 if Real Data
So the 1000 MHz must be correct result
I verified with Figure 79 by using DC Tool in Vivado and I got the same result like above equation.
Please tell me where I am wrong
Thank you