odd results working at 250ksp/s

jdow jdow at earthlink.net
Tue Oct 22 10:05:24 UTC 2013

Sample rate is 28.8 MHz. That is your basic bandwidth starting point. With
the E4000 delivering I and Q data for the sampler you have a starting
bandwidth of 28.8 MHz. The R820T converts to around 3 MHz I only. That
means the bandwidth is 0 to 14.4 MHz with the dongle slicing out up to
2.4 Msps of I/Q data. The R820T potentially has some interesting image
issues if it is really delivering 3 MHz to the RTL chip and not 7 MHz.

These are TV devices. That means any analog filtering they may or may not
have is 6 to 8 MHz wide. So up front it has that is the bandwidth to worry
about for strong signals.

The RTL chip delivers as little as 250 ksps. The word I've heard from
people who test the chip is that it's not very good for image (alias)
rejection much below 1 Msps. So I simply live with the 2.4 Msps and
learn to like it. SDRSharp more or less properly (more compromises than
some) decimates this down to something usable. But there's no way to
tell SDRSharp, "Really, Essdee, I'm only interested in this 200 kHz
so please only show that in the FFT window." It could be done. The code
would be only a little more complex. I don't worry about it. I use the
zoom slider and Bob's my uncle.

The fellow who posted the question indicates he's interested in VOR
reception. That's a 10 MHz band between 108 MHz and 118 MHz. I have
a hunch there is a small band near the FM broadcast band that's more
or less useless. So 4 dongles with appropriate tuning could get all
but about 400 kHz of that range all at once on a suitable computer.
5 could do it with considerable overlap. But one, really, could do
the job if it knew what frequencies to tune to. One of the scanner
tools would maybe make that a little easier.

He'll likely want to use some 1/4 wave coax notch's since he indicated
two frequencies that had huge signals were 1 MHz apart up around 106 MHz.
That would make a big difference for his weak signal work.


On 2013/10/22 01:44, Ralph A. Schmid, dk5ras wrote:
>> The filters you'd need wouldn't be in the R820T anyway (it has filters but not
>> that narrow).
>> IIRC, the rtl2832 always sample at a fixed high frequency but it has an internal
>> downconverter + FIR filter inside to convert the sample rate down. But that
>> FIR probably can't handle such a big reduction ...
> Still we must not forget that we're talking about some piece of silicon with nm
> structures, and therefrom no extraordinarily sharp filtering can be expected.
> When analog meets digital, it is important that the analog part protects the
> ADC from being overwhelmed by strong signals that are not in the scope of
> interest.
> All those silicon-only broadband receiver concepts drastically fail when being
> directly connected to a real good antenna in urban vicinity. They are good for
> a rubber duck style antenna sitting on your table, but that's it.
> There's a reason for all the solid silver and brass stuff in real world receivers :)
>> Cheers,
>>     Sylvain
> Ralph.

More information about the osmocom-sdr mailing list