Sounding signal output power and spectrum

GNU Radio
1

Hi all,

I have been doing some channel sounding measurements with the USRP,
and there is something that I am not sure about.

  1. The gr-sounder default transmit amplitude is 4096, is there a
    particular reason why this number is chosen? My guess is because a 12-
    bit DAC is used.
  2. All my measurements were conducted with RFX 2400 d’board, and the
    d’board is connected directly to a spectrum analyzer, the measured
    output power levels are listed below:

128: -28 dBm
256: -23 dBm
512: -17 dBm
1024: -11 dBm
2048: -6 dBm
4096: 1 dBm
so far so good(roughly 6dB step when I double the tx amplitude), and
all signals cover 32MHz bandwidth, but when I change --tx-amplitude
to 8192, it gave me a spike at the center frequency with very narrow
bandwidth, the power measured is about 11 dBm, any tx amplitude value
greater than 8192 yields a small power (-61dBm) with 32MHz bandwidth.
I am sure this has something to do with the DAC and PA, but not sure
exactly how to explain it.

  1. I did a indoor measurement with TX-RX separation of 50 meters (w/
    LOS), the received channel impulse response has 5 chunks of CIRs
    instead of one, and the number of samples between each chunk is
    always 800 chips, I am sure those CIRs are not multipath delays since
    in an indoor environment the corresponding delays can’t be 800-chips
    apart from each other(way too long). Am I missing something here?

Thanks in advance for any help.

Qi Chen

2

On Sat, Nov 29, 2008 at 9:55 AM, Qi Chen [email protected] wrote:

  1. The gr-sounder default transmit amplitude is 4096, is there a particular
    reason why this number is chosen? My guess is because a 12-bit DAC is used.

You are correct. This app uses a custom FPGA image for transmission,
and unlike most GNU Radio applications, the amplitude here is directly
converted to the DAC output values. This is the maximum amplitude.

  1. I did a indoor measurement with TX-RX separation of 50 meters (w/ LOS),
    the received channel impulse response has 5 chunks of CIRs instead of one,
    and the number of samples between each chunk is always 800 chips, I am sure
    those CIRs are not multipath delays since in an indoor environment the
    corresponding delays can’t be 800-chips apart from each other(way too long).
    Am I missing something here?

You need to post your command line parameters for the transmitter and
receiver.

-Johnathan

3

On Dec 1, 2008, at 2:24 PM, Johnathan C. wrote:

That makes sense. Do those measured output power values make sense?

Am I missing something here?

You need to post your command line parameters for the transmitter
and receiver.

I am using the default parameters for both tx and rx.

My command lines are as follows:

Tx: sudo ./usrp_sounder.py -t -f 2.44G -v -D
Rx: sudo ./usrp_sounder.py -t -f 2.44G -v -D -F output.dat

I use read_complex_binary.m to read the log file. The resulting
channel impulse response looks like this:
Figure 1: http://www.ittc.ku.edu/~chenqi/cir1.jpg
Figure 2: http://www.ittc.ku.edu/~chenqi/cir2.jpg

Figure 1 is the first cycle of the recorded impulse response.
Figure 2 shows five chunks of CIRs the first and second group of CIRs
are 800 chips apart.

Any clue on that?

4

On Dec 1, 2008, at 4:04 PM, Qi Chen wrote:

You are correct. This app uses a custom FPGA image for transmission,

instead of one,

I am using the default parameters for both tx and rx.

My command lines are as follows:

Tx: sudo ./usrp_sounder.py -t -f 2.44G -v -D
Rx: sudo ./usrp_sounder.py -t -f 2.44G -v -D -F output.dat

Oops, correction: Rx: sudo ./usrp_sounder.py -r -f 2.44G -v -D -F
output.dat.
Simple copy and past typo.