scanning with an rtlsdr

[Ben Hiett]

if you were wishing to use an rtlsdr to scan and identify signals across
its full tunable range, how long do you think it might take?

It looks like the instantaneous bandwidth is about 2.4MHz, tunable range
say 50-1750MHz

so you would have to look at the whole spectrum in about 700 chunks

i've seen a wide range of times as to how long the stick takes to retune,
ranging from about 500 times a second (2ms) to 20 times a second (50ms)

so it would take from 1.4s - 35s just to retune the 700 times

I have no idea how long you would need to sample at each freq to be able to
have enough data to decide if there were a signal present within that
'chunk' of spectrum

I assume you would probably capture x s of IQ samples, run them through an
fft and then just look for peaks, i.e energy in a bin or bins exceeding
some suitable threshold. If I could get a feel for what x s would roughly
be then I could get a handle on my original question

I guess the whole thing could be speeded up with some a priori knowledge of
where signals are likely to be, e.g. you may be able to skip parts of that
spectrum that you know won't have anything of interest in them

but i would still be very interested to know how long it would take to look
at the 'whole-thing' as it were.

apologies for the vagueness of the question, but i'm only really after a
vague answer for now.. i.e. ballpark in seconds say, or are we talking
minutes, hours or even days!

I did see some of the rtl_power plots that seem to stretch to 24 hrs so
maybe it takes longer than i thought!

[Oliver Jowett]

if you were wishing to use an rtlsdr to scan and identify signals across its
full tunable range, how long do you think it might take?

It looks like you're reinventing rtl_power.
Simplest thing is to run rtl_power and see, surely?
You can work backwards from the CSV output, which has one line per
frequency hop and a counter for the number of times the frequency was
sampled during the integration period.

The retuning speed is quite dependent on the exact version of the
library you're using.
The number of hops you need is dependent on how much bandwidth per hop
is usable (I find that you start running into the edge of the filters
and get uneven results if you try to grab the full 2.4MHz)

apologies for the vagueness of the question, but i'm only really after a
vague answer for now.. i.e. ballpark in seconds say, or are we talking
minutes, hours or even days!

Less than a minute assuming you don't want to dwell on each frequency
hop, just grab some power data and move on.

Oliver

[SDR Guru]

rtl_power, 2Mhz step, 25-1825Mhz, 900 steps/53 (step/sec) = 17 sec full sweep.

rtl_fm 7,7 step/sec

if you were wishing to use an rtlsdr to scan and identify signals across its full tunable range, how long do you think it might take?
It looks like the instantaneous bandwidth is about 2.4MHz, tunable range say 50-1750MHz
so you would have to look at the whole spectrum in about 700 chunks
i've seen a wide range of times as to how long the stick takes to retune, ranging from about 500 times a second (2ms) to 20 times a second (50ms)
so it would take from 1.4s - 35s just to retune the 700 times
I have no idea how long you would need to sample at each freq to be able to have enough data to decide if there were a signal present within that 'chunk' of spectrum
I assume you would probably capture x s of IQ samples, run them through an fft and then just look for peaks, i.e energy in a bin or bins exceeding some suitable threshold. If I could get a feel for what x s would roughly be then I could get a handle on my original question
I guess the whole thing could be speeded up with some a priori knowledge of where signals are likely to be, e.g. you may be able to skip parts of that spectrum that you know won't have anything of interest in them
but i would still be very interested to know how long it would take to look at the 'whole-thing' as it were.
apologies for the vagueness of the question, but i'm only really after a vague answer for now.. i.e. ballpark in seconds say, or are we talking minutes, hours or even days!
I did see some of the rtl_power plots that seem to stretch to 24 hrs so maybe it takes longer than i thought!