MIMO Question

A while ago (Nov. 14, 2008) Matt talked about using a USRP2 with
two antennas and two external radio receivers driving the basicRx.
My question is, is it possible to do that if the external radio IF and
bandwidth are too high to sample at 25 MHz without mixing the
signal to a lower freqeuncy first? I’m concerned that the reference
frequency into the mixer is probably complex, and the mixing process
will wind up putting energy from both formerly real signals onto each
of the mixer outputs. Is that true, and if so is there a way around it
that is already implemented? What gnuRadio modules would I use?

Thanks
Bruce

On Fri, Aug 14, 2009 at 04:55:56PM -0400, Bruce McGuffin wrote:

Thanks
Bruce

Bruce,

All of this depends on the center of your IF and how wide it is.
If the numbers are right, bandpass sampling definitely works.
The USRP2 samples at 100 MS/s.

What’s your IF freq?
How wide is it?

Eric

Eric B. wrote:

that is already implemented? What gnuRadio modules would I use?

What’s your IF freq?
How wide is it?

Eric
Eric

My IF center frequency is 21.4 MHz, and my bandwidth is about 20 MHz.
The good news is I don’t need the full IF bandwidth. I will probably
stick an analog filter in front of the USRP2 to reduce the aliased power
of the image.
But, for the sake of flexibility (I might want to use wider band signals
in the future)
and to keep costs down, I don’t want to use an analog filter that is too
narrow.
The numbers can be made to work.

I’m trying to avoid extensive (well, really, any if I can) FPGA
programming, and
wonder if versions of the USRP2 firmware exist that can either bandpass
filter, then
down-sample, or mix the signal to a lower IF with a sinusiod rather than
an
exponential, and low-pass filter. I prefer the direct down sampling
approach.

Bruce

On Mon, Aug 17, 2009 at 11:23:42AM -0400, Bruce McGuffin wrote:

of the mixer outputs. Is that true, and if so is there a way around it
The USRP2 samples at 100 MS/s.
of the image.
But, for the sake of flexibility (I might want to use wider band signals
in the future)
and to keep costs down, I don’t want to use an analog filter that is too
narrow.
The numbers can be made to work.

I’m trying to avoid extensive (well, really, any if I can) FPGA
programming, and wonder if versions of the USRP2 firmware exist that
can either bandpass filter, then down-sample, or mix the signal to a
lower IF with a sinusiod rather than an exponential, and low-pass
filter. I prefer the direct down sampling approach.

Bruce,

The standard firmware and fpga image should work fine for the single
channel case.

In the single channel case this will capture the entire IF and
translate it to baseband:

u.set_decim(4) # 25MS/s complex across ethernet
(100MB/s)
u.set_center_freq(21.4e6) # tune to the center of the IF

There’s currently no support in the standard code (host/firmware/fpga)
for multiple channels. This will change when we switch to over to the
VRT interface.

Eric