Re: Format of captured data

GNU Radio
1

Hi,

On Wed, 5/27/09, Karthik [email protected] wrote:

If you use usrp_c() source, then the real part of the complex values are the I values and the imaginary parts are Q values. They are each 16 bits.

Karthik

That is not correct. If you use usrp_c source, then I & Q each are 32
bits (total complex is 64bits). However, if you use usrp_s, then each
are 16 bits (total complex is 32bits)…

Best Regards,

Firas

2

On Wed, May 27, 2009 at 9:18 PM, Firas A. [email protected]
wrote:

That is not correct. If you use usrp_c source, then I & Q each are 32 bits (total complex is 64bits). However, if you use usrp_s, then each are 16 bits (total complex is 32bits)…

Best Regards,

Firas

Apologies for the wrong information. I don’t mean to side track the
original query, but I have a couple of questions.

I found this in the FAQ.

“Regarding the bandwidth, we can sustain 32MB/sec across the USB. All
samples sent over the USB interface are in 16-bit signed integers in
IQ format, i.e. 16-bit I and 16-bit Q data (complex) which means 4
bytes per complex sample. This resulting in a (32MByte per sec/4Byte)
8Mega complex samples/sec across the USB.”

So I guess here we are assuming that we are using usrp_s() ?

Also, since the ADC only has 12bit precision, are the extra bits used
to retain precision as we go through the CIC and Halfband filters?

If I were to use the std_4rx_0tx.rbf and usrp_s() to receive 4
channels. Then am I correct in saying that I need to have a
deinterleaver with 8 channels after the usrp_s() from which I will
club channel 0 and 1 into a new complex stream (similarly for channel
2 and 3 etc) to get 4 complex streams?

Thanks,
Karthik

3

On Wed, May 27, 2009 at 11:35 PM, Karthik [email protected] wrote:

I found this in the FAQ.
to retain precision as we go through the CIC and Halfband filters?

If I were to use the std_4rx_0tx.rbf and usrp_s() to receive 4
channels. Then am I correct in saying that I need to have a
deinterleaver with 8 channels after the usrp_s() from which I will
club channel 0 and 1 into a new complex stream (similarly for channel
2 and 3 etc) to get 4 complex streams?

Thanks,
Karthik

After snooping about a bit more, I found this on “The USRP under a
1.5X Magnifying Lens” ,
http://gnuradio.org/trac/attachment/wiki/UsrpFAQ/USRP_Documentation.pdf.

"Q) ADC samples at 64MHz, and passes through both I and Q channels
over the 24-bit RX bus.
Internal to the FPGA, the CIC automatically decimates by a value of
at least 4. The halfband
decimating FIR filter internal to the FPGA decimates by a fixed value
of 2. This gives a minimum
decimation rate of 8, leaving 8Msps going over the USB of the USRP.
Is this correct?

A) Yes. Note that some FPGA builds don’t contain the half-band.
With 16-bit I & Q decim = 8 → 8MS/sec → 32MB/sec.
With 8-bit I & Q decim = 4 → 16MS/sec → 32MB/sec "

Does this mean that when we use usrp_c(), the data sent over USB is
16bit I and Q but the container used to hold it after it reaches the
computer is 32 bit each. And when we use usrp_s(), the data sent is
8bit I and Q and the container used is 16bit?

Karthik

4

So I guess here we are assuming that we are using usrp_s() ?

Everything is 16 bits. usrp_c just casts the interleaved shorts into
floats, whereas usrp_s leaves the interleaved shorts as they are. That
is why the min and max values are +/- 2**15 even in floating point.

Also, since the ADC only has 12bit precision, are the extra bits used
to retain precision as we go through the CIC and Halfband filters?

Also, being byte-aligned is a key issue here. Imagine what would happen
if the host had to shift every sample in memory.

If I were to use the std_4rx_0tx.rbf and usrp_s() to receive 4
channels. Then am I correct in saying that I need to have a
deinterleaver with 8 channels after the usrp_s() from which I will
club channel 0 and 1 into a new complex stream (similarly for channel
2 and 3 etc) to get 4 complex streams?

does the usrp_s/c answer clear this up for you?

-Josh

5

On Thu, May 28, 2009 at 12:42 AM, Josh B. [email protected] wrote:

8Mega complex samples/sec across the USB."

does the usrp_s/c answer clear this up for you?

-Josh

Thanks Josh,

That clears things up for me.

Regards,
Karthik