Re: freqz Matlab Incompatibility

[siko1056]

Thomas D. Dean-2 wrote
> From a matlab example, I get an error - Matlab incompatibility??
> 
> http://www.aaronscher.com/wireless_com_SDR/RTL_SDR_AM_spectrum_demod.html
> 
> rtl_sdr -f 100500000 -g 40 -s 2500000 -n 25000000 /tmp/FMcapture1.dat
> 
> read 100.5 MHz gain 40db 2.5e6 samples/sec 10 seconds output to file
> 
> fid = fopen('/tmp/FMcapture1.dat','rb');
> y = fread(fid,'uint8=>double');
> # y is 8-bit i and q data from an rtlsdr,
> y = fread(fid,'uint8=>double');
> y = y-127.5;
> y = y(1:2:end) + i*y(2:2:end);
> 
> plot([1:5000], y(1:5000)); ## looks reasonable
> 
> freqz(y(1:5000),1,[-4E6:0.01e6:4E6],2.5E6);
> error: __plt2vv__: vector lengths must match
> error: called from
>      __plt__>__plt2vv__ at line 489 column 5
>      __plt__>__plt2__ at line 248 column 14
>      __plt__ at line 113 column 17
>      plot at line 223 column 10
>      freqz_plot at line 55 column 3
>      freqz at line 194 column 5
> 
> Tom Dean

Dear Tom,

In your example code and [1] you should avoid the redundant line and
consider using "close (fid);" after "fread".

The problem you are facing is that the vectors to be plotted differ by one
in their length in "freqz". This is due to a change two years ago [2] to fix
bug #43895 [3]. Assuming your calculations are right (I don't have a
background on these), if I comment out that line [2] in "freqz", your
example [1] works for me and the tests of "freqz" still all pass?!

Maybe one should open bug #43895 [3] again for further discussions?

HTH, Kai

[1]:
http://www.aaronscher.com/wireless_com_SDR/RTL_SDR_AM_spectrum_demod.html
[2]: https://hg.savannah.gnu.org/hgweb/octave/rev/8724af455b11#l1.73
[3]: https://savannah.gnu.org/bugs/?43895



--
View this message in context: 
http://octave.1599824.n4.nabble.com/freqz-Matlab-Incompatibility-tp4684530p4684531.html
Sent from the Octave - General mailing list archive at Nabble.com.