[EE] Lossy bandwidth compression with analog circutry?

[Jason White-20]

Hi everyone,

Is it possible to "compress" the bandwidth/frequency spectrum of a signal
using an analog circuit? (with or without loss of information) For example
taking 100Mhz of radio spectrum and shrinking it (preferably without
involvement of digital electronics) into a 1Mhz span for monitoring
purposes. Specifically to look at received power vs frequency vs time in
order to identify approximately when and at what frequency transmissions
occur at over a long period.

I think the closest equivalent of this would be having a receiver scan
"quickly" and looking at each "slice."
A second might be to do some sort of "frequency binning" in the analog
domain.

Are there other ways?
--
Jason White
--
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[Alan Pearce]

Isn't this what Software Defined Radio does?
In effect you can look at the frequency binning and select the
bandwidth you want to monitor. It doesn't compress the frequency range,
but does allow automatic monitoring.

Have a prowl around the KiwiSDR web site http://kiwisdr.com/ and see
what add-ons people have added to process the frequency binning (some
of them take you to external sites). KiwiSDR only does it for 0-30MHz,
but I believe there are others that can do it for higher frequencies
(the main limitation is the ADC).


On Sat, 16 Jan 2021 18:14:54 -0500
Jason White <[hidden email]> wrote:

> Hi everyone,
>
> Is it possible to "compress" the bandwidth/frequency spectrum of a
> signal using an analog circuit? (with or without loss of information)
> For example taking 100Mhz of radio spectrum and shrinking it
> (preferably without involvement of digital electronics) into a 1Mhz
> span for monitoring purposes. Specifically to look at received power
> vs frequency vs time in order to identify approximately when and at
> what frequency transmissions occur at over a long period.
>
> I think the closest equivalent of this would be having a receiver scan
> "quickly" and looking at each "slice."
> A second might be to do some sort of "frequency binning" in the analog
> domain.
>
> Are there other ways?

--
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[Jason White-20]

That's right, as you mentioned a fast ADC and processor is needed to cover
large bandwidths. I was curious if there were other methods.

On Sat, Jan 16, 2021 at 7:10 PM Alan Pearce <[hidden email]>
wrote:

> Isn't this what Software Defined Radio does?
> In effect you can look at the frequency binning and select the
> bandwidth you want to monitor. It doesn't compress the frequency range,
> but does allow automatic monitoring.
>
> Have a prowl around the KiwiSDR web site http://kiwisdr.com/ and see
> what add-ons people have added to process the frequency binning (some
> of them take you to external sites). KiwiSDR only does it for 0-30MHz,
> but I believe there are others that can do it for higher frequencies
> (the main limitation is the ADC).
>
>
> On Sat, 16 Jan 2021 18:14:54 -0500
> Jason White <[hidden email]> wrote:
>
> > Hi everyone,
> >
> > Is it possible to "compress" the bandwidth/frequency spectrum of a
> > signal using an analog circuit? (with or without loss of information)
> > For example taking 100Mhz of radio spectrum and shrinking it
> > (preferably without involvement of digital electronics) into a 1Mhz
> > span for monitoring purposes. Specifically to look at received power
> > vs frequency vs time in order to identify approximately when and at
> > what frequency transmissions occur at over a long period.
> >
> > I think the closest equivalent of this would be having a receiver scan
> > "quickly" and looking at each "slice."
> > A second might be to do some sort of "frequency binning" in the analog
> > domain.
> >
> > Are there other ways?
>
> --
> http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
> View/change your membership options at
> http://mailman.mit.edu/mailman/listinfo/piclist
>

--
Jason White
--
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[Richard Prosser]

What sort of modulation?
If "standard" AM or FM there should be a constant carrier frequency you
could use for detection & signal strength. I'm not sure about some of the
digital modes though.
RP

On Sun, 17 Jan 2021 at 13:46, Jason White <[hidden email]>
wrote:

> That's right, as you mentioned a fast ADC and processor is needed to cover
> large bandwidths. I was curious if there were other methods.
>
> On Sat, Jan 16, 2021 at 7:10 PM Alan Pearce <[hidden email]
> >
> wrote:
>
> > Isn't this what Software Defined Radio does?
> > In effect you can look at the frequency binning and select the
> > bandwidth you want to monitor. It doesn't compress the frequency range,
> > but does allow automatic monitoring.
> >
> > Have a prowl around the KiwiSDR web site http://kiwisdr.com/ and see
> > what add-ons people have added to process the frequency binning (some
> > of them take you to external sites). KiwiSDR only does it for 0-30MHz,
> > but I believe there are others that can do it for higher frequencies
> > (the main limitation is the ADC).
> >
> >
> > On Sat, 16 Jan 2021 18:14:54 -0500
> > Jason White <[hidden email]> wrote:
> >
> > > Hi everyone,
> > >
> > > Is it possible to "compress" the bandwidth/frequency spectrum of a
> > > signal using an analog circuit? (with or without loss of information)
> > > For example taking 100Mhz of radio spectrum and shrinking it
> > > (preferably without involvement of digital electronics) into a 1Mhz
> > > span for monitoring purposes. Specifically to look at received power
> > > vs frequency vs time in order to identify approximately when and at
> > > what frequency transmissions occur at over a long period.
> > >
> > > I think the closest equivalent of this would be having a receiver scan
> > > "quickly" and looking at each "slice."
> > > A second might be to do some sort of "frequency binning" in the analog
> > > domain.
> > >
> > > Are there other ways?
> >
> > --
> > http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
> > View/change your membership options at
> > http://mailman.mit.edu/mailman/listinfo/piclist
> >
>
>
> --
> Jason White
> --
> http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
> View/change your membership options at
> http://mailman.mit.edu/mailman/listinfo/piclist
>

--
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[Sean Breheny]

Hi Jason,

There are many ways to do what you are asking. One way (maybe not the
most practical) is to use a length of highly-dispersive transmission
line (or dispersive delay line or SAW structure). This would convert
frequency-domain into time-domain (in other words, if you put a 1
millisecond burst of the entire radio spectrum in one end, you'd get a
longer (let's say 10 milliseconds) burst out the other end where the
lower frequencies would come out first and the higher frequencies
later, or vice-versa. You could then use a demodulator to produce a
power vs frequency output which was represented as a voltage vs time.
Some old radars did things like this and so did early spectrum
analyzers. The section of transmission line needed would be very long
- probably at least 10s of meters even when made of very special
ultra-low velocity factor materials.

Of course another, simpler way would be a wideband tunable filter
which you sweep across the spectrum. This would produce a similar
output (frequency vs time). This may be easier to do if you first
upconvert your 0-100MHz to, say, 1 GHz to 1.1GHz and then use a
YIG-tuned filter (YTF) or dielectric resonator filter.

Could also be done with clever use of a comb filter and undersampling.
In other words, you intentionally produce a highly-aliased version of
the signal but preceed the sampler with a comb filter which selects
out several narrow ranges of frequency so that the aliased copies
don't overlap in the frequency domain. Tough to make sure everything
lines up with no gaps or unintentional overlaps.

All of the above, though, would be harder to implement today than a
digital solution. They would require careful tweaking and quite
possibly temperature compensation to make them workable. I've seen
1970s microwave test gear where a large part of the system consisted
of arrays of op-amps and multi-turn pots to create complex calibration
curves (e.g. for linearizing the tuning of a VCO) in an analog
circuit, which all had to be hand-tuned for each unit at the factory.

Sean

On Sat, Jan 16, 2021 at 6:17 PM Jason White
<[hidden email]> wrote:

>
> Hi everyone,
>
> Is it possible to "compress" the bandwidth/frequency spectrum of a signal
> using an analog circuit? (with or without loss of information) For example
> taking 100Mhz of radio spectrum and shrinking it (preferably without
> involvement of digital electronics) into a 1Mhz span for monitoring
> purposes. Specifically to look at received power vs frequency vs time in
> order to identify approximately when and at what frequency transmissions
> occur at over a long period.
>
> I think the closest equivalent of this would be having a receiver scan
> "quickly" and looking at each "slice."
> A second might be to do some sort of "frequency binning" in the analog
> domain.
>
> Are there other ways?
> --
> Jason White
> --
> http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
> View/change your membership options at
> http://mailman.mit.edu/mailman/listinfo/piclist

--
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[Sean Breheny]

Don't forget that if you can accept lossy signal processing, you can
use aliasing to your advantage. For example, if you undersample but
can control the precise sampling frequency then you can make small
changes in the sampling frequency and shift aliased signals around to
get a good picture of what the spectrum looks like without having to
actually sample at 2x the highest frequency. If, however, you need an
actual accurate instantaneous snapshot of the entire 100MHz of
spectrum then there is no way to do that without sampling at least at
200 Msps.

Sean


On Sat, Jan 16, 2021 at 7:46 PM Jason White
<[hidden email]> wrote:
>
> That's right, as you mentioned a fast ADC and processor is needed to cover
> large bandwidths. I was curious if there were other methods.
>
--
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist