[Discuss-gnuradio] RE: [USRP-users] Sampling at 100MS

[Per Zetterberg]

Gregor,

I am curious - what are you going to do with samples ? - store on in RAM or 
send them somewhere else ?

Unfortunately, I have no clue about the question.

BR/
Per
________________________________________
From: address@hidden address@hidden on behalf of Gregor Dschung address@hidden
Sent: Tuesday, December 14, 2010 3:03 PM
To: address@hidden; address@hidden
Subject: [USRP-users] Sampling at 100MS

Hello,

in dsp_core_rx.v, what would be the best place to tap the samples for
working at 100MS? Currently, I'm using i_cordic and q_cordic (rounded by
something like "round_reg #(.bits_in(24),.bits_out(16))
round_isamplefull
(.clk(clk),.in(i_cordic),.out(i_sample_full_scaled))"), but the block
I've designed doesn't work as simulated by Simulink with an input at
100MS (for the simulation, I've used an interpolated 20MS recording from
GNU Radio). So before I spend hours to find the problem inside the block
(or to redesign the block, so it's working on 25MS), I would like to
exclude the problem in the block's linkage.
I'm asking myself, if the scaling and rotation of the samples (from
adc_i/q * scale_i/q => prod_i/q => i/q_cordic) is depending on the set
decimation---hence a preparatory work that has to be done for the
CIC/halfband decimation filters---or not, meaning it's a general and
reasonable digital signal process. If so, should I rather tap on adc_i/q
than on i/q_cordic?
Any hint is appreciated.

Best Regards,
Gregor

PS: Can't decide which list I should write to (general dsp question or
ursp2-specific?), so this mail is going to both ;)