Hello Nicolas,
sorry for the delayed reaction!
So, the signal chain looks roughly like this:
RX:
ADC (@Master Clock Rate (MCR) --> Rotator (@MCR) --> decimator --> Host
(@user-requested sampling rate)
TX:
Host (@user sampling rate) --> interpolator --> Rotator (@MCR) --> DAC
(@MCR)
The two rotators here are the CORDICs you're referring to.
UHD just calculates by how much of the master clock rate you need to
rotate per sample – e.g. if you tune digitally by 1 MHz, and your MCR is
200 MHz, then each sample needs to be phase-shifted by
$\frac{2\pi}{\frac{\text{MCR}}{f_\text{offset}}}$; ie. you need a
multiplicator that takes the input sample, and multiplies it with a
complex factor $e^{j\varphi}$; and after each sample, that $\varphi$
needs to be incremented by $\frac{2\pi}{200}$.
Thus, what UHD calculates is exactly that factor of 2π/200 and sends it
to the CORDIC unit on the FPGA, which then uses that value as phase
increase.
Best regards,
Marcus
On 06/09/2017 11:04 AM, Nicolas GALLAND wrote:
> Hello everyone,
>
> I am using a USRP X310 with basic RX and basic TX daughterboards and I
> would like to know the exact relation between the number I enter as a
> parameter for the central frequency in the UHD USRP sink/source block
> (which is a real number) and the actual phase per each sample.
>
> As a matter of facts, if the DDC/DUC chains were using a DDS, the
> relation would be pretty straightforward (knowing the control word of
> the DDS and the number of bits of the phase accumulator...). However,
> as I understand, the FPGA design uses a CORDIC for the DDC/DUC chain,
> for which the exact correspondence between sample index and phase is
> not as straight forward to derive. I am sorry if this is textbook
> material for CORDIC specialists (which I am not...), but I am
> wondering if someone could point me to the exact formula of phase vs
> sample index that I could use to very accurately know what is going on
> in the DDC/DUC chain. The bottom line is that I need to be sure of
> what is happening at the nrad level for MHz-ish input/output and
> mod/demod frequency, and the intricacies of the CORDIC behavior is -
> to my best understanding - playing a major role at this level.
>
> Any help would be greatly appreciated.
>
> Best regards,
>
> Nicolas
>
>
> _______________________________________________
> Discuss-gnuradio mailing list
> Discuss-gnuradio@gnu.org
> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
_______________________________________________
Discuss-gnuradio mailing list
Discuss-gnuradio@gnu.org
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
