Working at 20 MS/s

Hi,

I was interested in working with detectors for OFDM 802.11b in my USRP2.
I’ve read on the internet about this and it seems that there are people
able to use sampling rates of 20 MS/s.

However, when I try to send an OFDM signal with 20 MHz of bandwidth, at
the USRP2 receiver board the FFT plot shows the signal blinking.

This effect only disappears for sampling rates as low as ~2 MHz.

This will affect my detectors performance

What is the main cause for this? Can it be solved?

Regards,

Francisco


View this message in context:
http://old.nabble.com/Working-at-20-MS-s-tp33763341p33763341.html
Sent from the GnuRadio mailing list archive at Nabble.com.

On 05/01/2012 08:55 AM, frankist wrote:

Hi,

I was interested in working with detectors for OFDM 802.11b in my USRP2.
I’ve read on the internet about this and it seems that there are people
able to use sampling rates of 20 MS/s.

However, when I try to send an OFDM signal with 20 MHz of bandwidth, at
the USRP2 receiver board the FFT plot shows the signal blinking.

I think you are seeing the result of an FFT over a sample discontinuity.
This discontinuity is due to overflow, in other words, your computer is
not able to process at this rate.

http://files.ettus.com/uhd_docs/manual/html/general.html#overflow-underflow-notes
Are you seeing overflow (O’s are being printed)?

Are there any warnings printed by the driver?
(Answering some of those warnings can help with performance)

Usually running an FFT display on RX data is not a big performance hit.
Any information about your PC that you can share? CPU? OS?

-Josh

Able to use sample rates of 20 MHz doesn’t necessarily include doing
something as slow and killing of performance as display of an FFT.

Bob

Hi,

In fact I have underruns in the transmitter which means according to the
website you gave me, my transmitter is not producing data fast enough. I
guess I need a better computer for that.

Just one more question:
Are sampling rates of 20 and 25 MS/s too much for devices, more
especifically, for cognitive radio?

Josh B.-3 wrote:

[email protected]
Discuss-gnuradio Info Page


View this message in context:
http://old.nabble.com/Working-at-20-MS-s-tp33763341p33779586.html
Sent from the GnuRadio mailing list archive at Nabble.com.

On 05/01/2012 06:58 PM, Robert McGwier wrote:

Able to use sample rates of 20 MHz doesn’t necessarily include doing
something as slow and killing of performance as display of an FFT.

Bob

The Gnu Radio FFT display widgets do fairly-heavy decimation in
vector-sized chunks, so they’re actually fairly non-costly. You lose
sensitivity
that way, but can display spectral estimates even for fairly high
sample rates, since you’re only doing an FFT over a small fraction of
the incoming data–the rest is discarded.