Using rtlsdr do spectrum analysis of modulating signal, seeing image signals?

[yu]

The hardware is rtlsdr, spectrum software is SDR#, sampling rate is
2M/s, the signal source link to rtlsdr directly.

The center frequency of the modulating signal is 1262MHz, its
bandwidth is 1M, the picture next is showing the normal status of the
spectrum. http://104.236.60.141/static/1262.jpg

When move the tuning center frequency to 1261MHz, there occurs a
signal in 1260MHz, and seems just the image of the normal signal. and
when tuning to 1263, the image also show in 1264. the pics:

http://104.236.60.141/static/0-1261.jpg
http://104.236.60.141/static/1263.png

At first, I think it may be a problem of the rtlsdr, but when I change
rtlsdr to hackrf, the situation is exactly the same; so maybe it is a
FFT problem? and how can I confirm whether it is a hardware or
software problem?

[Leif Asbrink]

This looks like aliasing.

Have a look at 2:15 and onwards in this video:
https://www.youtube.com/watch?v=14CD1ERdRn0

The R820T is mis-used because this vido is old and
setting the IF bandwidth for the R820T was not yet possible.

At 4:00 here:
https://www.youtube.com/watch?v=cVU5X1d2XYU
you can see how a sweepind signal that gets outside the
spectrum on the high frequency side aliases into the low
frequency side.

If you have an R820T tuner you can enable the IF filter
and set a bandwidth of something like 1.8 MHz to suppress
the aliasing. This video shows a little about the advantages
of using the IF filter:
https://www.youtube.com/watch?v=btDiWHuI730

I think setting IF filter is now included in the osmocom API,
but I do not know if there is already a dll file that supports it
for SDR#

Regards

Leif

> --
> You received this message because you are subscribed to the Google Groups "Ultra Cheap SDR" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to ultra-cheap-s...@googlegroups.com.
> To post to this group, send email to ultra-c...@googlegroups.com.
> Visit this group at http://groups.google.com/group/ultra-cheap-sdr.
> For more options, visit https://groups.google.com/d/optout.

[yu]

Hi, Leif

Thanks a lot for pointing out the direction!

I have tested the rtlsdr_set_tuner_bandwidth API, both in
dl8aau/librtlsdr and steve-m/librtlsdr. It seems the aliasing can be
suppressed, but not perfectly.

My testing hardware is rtlsdr, spectrum software is SDR#, tuner is
r820T, sampling rate is 2M/s, the signal source link to rtlsdr
directly, the modulating signal's center frequency is 88MHz, bandwidth
is 1M.

For dl8aau/librtlsdr, because its code like 'static const int
r82xx_bandwidth_table[]={ 300000, 400000, 550000,
700000,1000000,1200000,1300000,1600000,2200000};'

I test 1M 1.3M 1.6M and 2.2M bandwidth, the results as follow:
tuning center frequency 87MHz :
http://104.236.60.141/static/if_filter_leif_87.jpg
tuning center frequency 89MHz :
http://104.236.60.141/static/if_filter_leif_89.jpg

For steve-m/librtlsdr, its code like 'static const int
r82xx_if_low_pass_bw_table[] = { 1700000, 1600000, 1550000, 1450000,
1200000, 900000, 700000, 550000, 450000, 350000};'

I test 1.6M 1.55M 1.45M and 1.2M bandwidth, here the results:
tuning center frequency 87MHz :
http://104.236.60.141/static/if_filter_steve_87.jpg
tuning center frequency 89MHz :
http://104.236.60.141/static/if_filter_steve_89.jpg

All of the settings above can suppress the aliasing, but more or less
the aliasing still exists, and the better result seems the
steve-m/librtlsdr 1.45/1.55M.

Two questions:

1、For tuning 87MHz,After IF filter, should the signal curve keep
horizontal and then drop vertically at the point of 88MHz or just
drop slowly like showing in the picture?

2、The IF filter bandwidth of the two drivers are not all the same and
the setting value is fixed, can I set the bandwidth freely such as
2000000? Is there any document or datasheet about these?

Regards

yu

[Leif Asbrink]

Hello yu,

> For dl8aau/librtlsdr, because its code like 'static const int
> r82xx_bandwidth_table[]={ 300000, 400000, 550000,
> 700000,1000000,1200000,1300000,1600000,2200000};'
>
> I test 1M 1.3M 1.6M and 2.2M bandwidth, the results as follow:
> tuning center frequency 87MHz :
> http://104.236.60.141/static/if_filter_leif_87.jpg
> tuning center frequency 89MHz :
> http://104.236.60.141/static/if_filter_leif_89.jpg
>
> For steve-m/librtlsdr, its code like 'static const int
> r82xx_if_low_pass_bw_table[] = { 1700000, 1600000, 1550000, 1450000,
> 1200000, 900000, 700000, 550000, 450000, 350000};'
>
> I test 1.6M 1.55M 1.45M and 1.2M bandwidth, here the results:
> tuning center frequency 87MHz :
> http://104.236.60.141/static/if_filter_steve_87.jpg
> tuning center frequency 89MHz :
> http://104.236.60.141/static/if_filter_steve_89.jpg
>
> All of the settings above can suppress the aliasing, but more or less
> the aliasing still exists, and the better result seems the
> steve-m/librtlsdr 1.45/1.55M.
>
> Two questions:
>
> 1、For tuning 87MHz,After IF filter, should the signal curve keep
> horizontal and then drop vertically at the point of 88MHz or just
> drop slowly like showing in the picture?

It should be as you see in your pictures. When you apply an IF filter
with nominally 1.6 MHz bandwidth you should ignore everything
that is more than 800 kHz away from the center. The signal would be
attenuated and aliases may occur outside the valid frequency range.

> 2、The IF filter bandwidth of the two drivers are not all the same and
> the setting value is fixed, can I set the bandwidth freely such as
> 2000000? Is there any document or datasheet about these?

http://superkuh.com/gnuradio/R820T_datasheet-Non_R-20111130_unlocked.pdf
You would need the register map:
http://t160.mobile-osmocom-sdr.mobiletalk.us/re-r820t-register-map-available-t160.html


Regards

Leif