Re: [Discuss-gnuradio] Overflows in benchmark_rx.py

[Tom Rondeau]

On Fri, Feb 15, 2013 at 9:46 PM, Manu T S <address@hidden> wrote:

Hi,

I was running the benchmark scripts with the following set of parametrs

-f 915M -r 5000000 -m psk -8

with 30 dB gain set on both Tx and Rx. I was getting cluttered overflows and packet received during that time are in error.. Like I'm receiving about a 1000 packet without error and then then system runs into overflow and about then quite a large number of packets are received in error.

Output is posted

Demodulator:

bits per symbol:     3

RRC roll-off factor: 0.35

FLL bandwidth:       6.28e-02

Timing bandwidth:    6.28e-02

Phase bandwidth:     6.28e-02

-- Opening a USRP2/N-Series device...

-- Current recv frame size: 1472 bytes

-- Current send frame size: 1472 bytes

UHD Receiver:

UHD Args:    

Freq:        915MHz

Gain:        30.000000 dB

Sample Rate: 3.33333Msps

Antenna:     None

Spec:        None

Demodulator:

bits per symbol:     3

RRC roll-off factor: 0.35

FLL bandwidth:       6.28e-02

Timing bandwidth:    6.28e-02

Phase bandwidth:     6.28e-02

Receive Path:

modulation:      psk_demod

bitrate:         5Mb/s

samples/symbol:  2.0000

Differential:    True

Using Volk machine: avx_64_mmx

ok =  True  pktno =    1  n_rcvd =    1  n_right =    1

ok =  True  pktno =    2  n_rcvd =    2  n_right =    2

ok =  True  pktno =    3  n_rcvd =    3  n_right =    3

.

.

.

ok =  True  pktno = 2953  n_rcvd = 2953  n_right = 2953

ok =  True  pktno = 2954  n_rcvd = 2954  n_right = 2954

ok =  True  pktno = 2955  n_rcvd = 2955  n_right = 2955

ok =  True  pktno = 2956  n_rcvd = 2956  n_right = 2956

ok = False  pktno = 2957  n_rcvd = 2957  n_right = 2956

ok = False  pktno = 2959  n_rcvd = 2958  n_right = 2956

ok = False  pktno = 2961  n_rcvd = 2959  n_right = 2956

ok = False  pktno = 2963  n_rcvd = 2960  n_right = 2956

ok = False  pktno = 2965  n_rcvd = 2961  n_right = 2956

ok = False  pktno = 2967  n_rcvd = 2962  n_right = 2956

.

.

.

ok = False  pktno = 3838  n_rcvd = 3369  n_right = 2956

ok = False  pktno = 3841  n_rcvd = 3370  n_right = 2956

ok = False  pktno = 3843  n_rcvd = 3371  n_right = 2956

ok =  True  pktno = 3845  n_rcvd = 3372  n_right = 2957

ok =  True  pktno = 3846  n_rcvd = 3373  n_right = 2958

ok =  True  pktno = 3847  n_rcvd = 3374  n_right = 2959

ok =  True  pktno = 3848  n_rcvd = 3375  n_right = 2960

.

.

.

Any idea how to avoid these overflows???

--
Manu T S

This is usually a problem due to the receiver not being able to keep up with the flow of samples from the UHD device. Try running at a lower bitrate. If you're seeing overflows temporarily and then things are correcting themselves, it looks like you might be right on the edge of your system's capabilities. In this case, it's like some other process/task is taking some time away from your app, causing it to slow down just enough to cause problems, but then things look like they come back. You could also try to run with realtime privileges enabled to make sure your process gets priority.

On the other hand, the output you show here doesn't have any overflow indication. Are you sure that this is what's happening? You're running at 915 MHz, the center of the 900 MHz ISM band (depending on where you are). It's likely to have other signals in it. Maybe this is just a case of interference?

Tom