Re: [Discuss-gnuradio] Using the MIMO port for serial comm. to a Virtex5 board

[Matt Ettus]

On 12/01/2009 12:58 PM, Charles Irick wrote:
> Hello,
> I'm most likely having a clock issue because I cannot get any data
> transfers going between the USRP2 and GTPs on the V5. A few questions:
>
> - I'm not quite sure I understand the 10MHz reference clock idea, or
> its usage; I looked through the HDL and everything on the FPGA is
> running at 100MHz.

The 10 MHz reference input on the front is optional.  If you connect it 
and enable locking, the 100 MHz main oscillator will lock to it, 
otherwise it will free run.  Everything inside the FPGA runs at 100 MHz 
(DSP) or 50 MHz (processor).

There is also a 10 MHz reference clock sent out on the MIMO connector. 
When turned on, it will be the 100 MHz clock divided down by 10.

> - I assume I will not need to provide this reference clock if the
> USRP2 is the driver given that sd_gentest appears to do so in the main
> functions clocks_mimo_config. Either way I'm using the SAS to quad
> SATA cable so the clock isn't even transferred to the GTPs (literally,
> it's supposed to derive it from the data).

sd_gentest will generate packets and send them on the serdes on the MIMO 
cable.  You should be able to receive it without using the 10 MHz 
reference clock because, as you say, the clock is recovered from the data.

> - My only clocking options on the GTP's are 150 or 75MHz. I'm not sure
> if I should clock the USRP2's FPGA or the reference differently to
> match this.


The important thing is that you get the GTPs set up for the correct data 
rate.  On the USRP2, the main clock to the SERDES chip is 100 MHz.  2 
Bytes are sent on each clock, so the effective data rate is 1.6 Gbps. 
However, since 8b-10b coding is used, the on the wire bit rate is 2.0 
Gbps (20*100MHz).  So whatever reference clock you use, you need the GTP 
to be working at 2 Gbps.  I am by no means an expert on the GTPs, but I 
think the easiest way to do this is to have a reference clock of 100 MHz.

You could probably use the PLLs in the V5 to create a 100 MHz reference 
clock from one of your other clock rates.

Matt