Home

 
: Nice spectrum analyzer
(<-- Back to message list)

Peter
Sun, 28 May 06 16:14:29 +0000

Excite DUT with white noise, convert output down to audio band using XO
and use spectrum analyzer running on PC audio card to visualise.
Schematic supplied. The nice part is that given a good mixer and LO this
whould work way up into microwave. Finally an affordable solution for
filter and amplifier design for homebrew UHF and microwave ?

http://www.qsl.net/7n3wvm/Fil_Meas.html

Peter
5 replies folowing
Vasile Surducan
Sun, 28 May 06 16:54:05 +0000
On 5/28/06, Peter wrote:
>
> Excite DUT with white noise, convert output down to audio band using XO
> and use spectrum analyzer running on PC audio card to visualise.
> Schematic supplied. The nice part is that given a good mixer and LO this
> whould work way up into microwave.


Would work yes, but will be almost unuseless. The simple mixer must
be replaced with an IQ one, else would be necessary two or three IF
stages with all the problems. The biggest one will be the frequency
fluctuation in the audio side and the limited bandwith or distortion.
Also will be necessary a real VCO and not a white noise generator.
There are many analyzers like this one on the web, the best one is
using a tuner
and claim to work ok till 2Ghz. I'm seeing every day how works a good
spectrum analyzer (for it's 30 years old age) TEK492 and if that one
can't be used for fast signal analyses, all analyzers having the same
principle (unfortunately) can't.
So depends what do you need.

greetings,
Vasile



Finally an affordable solution for

> filter and amplifier design for homebrew UHF and microwave ?
>
> http://www.qsl.net/7n3wvm/Fil_Meas.html
>
> Peter
> --
> > > >

dicsEE
Sun, 28 May 06 19:52:21 +0000
 
Peter
Sun, 28 May 06 19:52:21 +0000
Vasile Surducan gmail.com> writes:

> Would work yes, but will be almost unuseless. The simple mixer must
> be replaced with an IQ one, else would be necessary two or three IF
> stages with all the problems. The biggest one will be the frequency
> fluctuation in the audio side and the limited bandwith or distortion.
> Also will be necessary a real VCO and not a white noise generator.
> There are many analyzers like this one on the web, the best one is
> using a tuner
> and claim to work ok till 2Ghz. I'm seeing every day how works a good
> spectrum analyzer (for it's 30 years old age) TEK492 and if that one
> can't be used for fast signal analyses, all analyzers having the same
> principle (unfortunately) can't.
> So depends what do you need.


I wrote:

> > Finally an affordable solution for
> > filter and amplifier design for homebrew UHF and microwave ?
> >
> > http://www.qsl.net/7n3wvm/Fil_Meas.html
> >
> > Peter


Could you elaborate on this ? Why is an IQ mixer needed ? The way I see it the
most needed thing is low (very low) phase noise of the carrier so the LO does
not swamp the mixer and the audio after it (although I am unsure what happens
with the phase noise in a direct conversion receiver when an actual signal is
not present). The scheme is the same as used in DSB/CW direct conversion
receivers, whose performance is rather good usually. Also using a noise source
as input makes it possible to go to very high frequencies without the usual
problems. No ? Also it is a very simple and inexpensive scheme. With a $10 mixer
it could cover dc to 4GHz. If a stabilised Gunnplexer is used then microwave
filter (and Gunn and varactor bias, by voltage, to match an existing filter or
structure) tuning becomes possible. What am I missing ? Ok, dynamic range is
probably 55dB or less (sound card dynamic range). With a good sound card and a
preamplifier this could be 90-100dB. Many people already have such a good sound
card. If the LO is replaced with a stepwise DSS or PLL then a spectrum can be
taken with say 40kHz 'windows' from DC (almost) to 1GHz. The DSS could be
controlled by a USB to parallel control for example.

Peter

Dave Tweed
Sun, 28 May 06 20:31:20 +0000
Peter wrote:
> Could you elaborate on this? Why is an IQ mixer needed? The way I see
> it the most needed thing is low (very low) phase noise of the carrier
> so the LO does not swamp the mixer and the audio after it (although I
> am unsure what happens with the phase noise in a direct conversion
> receiver when an actual signal is not present). The scheme is the
> same as used in DSB/CW direct conversion receivers, whose performance
> is rather good usually. Also using a noise source as input makes it
> possible to go to very high frequencies without the usual problems.
> No? Also it is a very simple and inexpensive scheme. With a $10 mixer
> it could cover dc to 4GHz. If a stabilised Gunnplexer is used then
> microwave filter (and Gunn and varactor bias, by voltage, to match
> an existing filter or structure) tuning becomes possible. What am
> I missing? Ok, dynamic range is probably 55dB or less (sound card
> dynamic range). With a good sound card and a preamplifier this could
> be 90-100dB. Many people already have such a good sound card. If the
> LO is replaced with a stepwise DSS or PLL then a spectrum can be
> taken with say 40kHz 'windows' from DC (almost) to 1GHz. The DSS
> could be controlled by a USB to parallel control for example.


Your enthusiasm is causing you to extrapolate way beyond the concept
of the original project, which was to evaluate the details of a crystal
filter's amplitude response, which was already known to be less than
the bandwidth of the sound card. This isn't a "spectrum analyzer" so
much as a magnitude-only "network analyzer" (no phase information).

In order to turn this into a direct-conversion spectrum analyzer,
you'd need to build a filter with a fixed 20 kHz bandwidth whose
center frequency could be varied from "DC to 1 GHz". Good luck!

If you want a rather good design for a low-cost spectrum analyzer,
as well as a discussion of some of the design issues, look for the
article by Neal Martini in the next issue (#192) of Circuit Cellar.
It's based on a MAX3550 and a PIC18F4520.

-- Dave Tweed
Peter
Sun, 28 May 06 21:30:23 +0000
Dave Tweed dtweed.com> writes:

> Your enthusiasm is causing you to extrapolate way beyond the concept
> of the original project, which was to evaluate the details of a crystal
> filter's amplitude response, which was already known to be less than
> the bandwidth of the sound card. This isn't a "spectrum analyzer" so
> much as a magnitude-only "network analyzer" (no phase information).
>
> In order to turn this into a direct-conversion spectrum analyzer,
> you'd need to build a filter with a fixed 20 kHz bandwidth whose
> center frequency could be varied from "DC to 1 GHz". Good luck!
>
> If you want a rather good design for a low-cost spectrum analyzer,
> as well as a discussion of some of the design issues, look for the
> article by Neal Martini in the next issue (#192) of Circuit Cellar.
> It's based on a MAX3550 and a PIC18F4520.


Thanks for answering.

My enthusiasm is that of an optimist with experience. I have specan
experience with homebrew, HP, and Elbonian devices. I think that I
know what can be built, bought and rented, and what it's worth.

My question was, assuming this is a $20 project including 1 hour of
soldering, and that it will be used for low signal work, in closed
circuit (not exposed to QRM for example), and that the user can always
turn off the power to the DUT and to the signal source if suspecting
something is not right about the output, are there any serious flaws
in this scheme, which would prevent it from being used to tune VHF,
UHF and microwave circuits using the noise input method (noise which
is inherently incoherent and thus makes an IQ demodulator unnecessary) ?



As to the CC article, thanks for posting that.

Peter

Vasile Surducan
Mon, 29 May 06 11:48:48 +0000
On 5/28/06, Peter wrote:
> Vasile Surducan gmail.com> writes:
>
> > Would work yes, but will be almost unuseless. The simple mixer must
> > be replaced with an IQ one, else would be necessary two or three IF
> > stages with all the problems. The biggest one will be the frequency
> > fluctuation in the audio side and the limited bandwith or distortion.
> > Also will be necessary a real VCO and not a white noise generator.
> > There are many analyzers like this one on the web, the best one is
> > using a tuner
> > and claim to work ok till 2Ghz. I'm seeing every day how works a good
> > spectrum analyzer (for it's 30 years old age) TEK492 and if that one
> > can't be used for fast signal analyses, all analyzers having the same
> > principle (unfortunately) can't.
> > So depends what do you need.
>
> I wrote:
>
> > > Finally an affordable solution for
> > > filter and amplifier design for homebrew UHF and microwave ?
> > >
> > > http://www.qsl.net/7n3wvm/Fil_Meas.html
> > >
> > > Peter
>
> Could you elaborate on this ? Why is an IQ mixer needed ?


Well, a DC to microwave spectrum analyzer can't be made with one IF stage
only if you're using an zero IF IQ mixer, a damn good reference oscillator
-not a DDS which has a really poor SFDR on the whole oscillator band
(50-60dB if you're a very good PCB and filter designer) with the
eliptical filter adjusted on the maximum frequency range and a sinx/x
amplitude dependence (and you need about 90...120 dB SNR if you want
to see something on your screen with small signals on the input)
-nor a PLL oscillator which has a poor signal to noise ratio (dBc)
because of it's working principle. So you need something else or a
combinations above...
A mixer from DC to 4GHz is good, but not if you're trying to get
audio, mixing 4.000.000,000 MHz local oscilator with an input signal
of 4.000.000,010 MHz, and you believe you'll really get 10KHz without
crazy fluctuations...
If you're interested I could send you a few links about home made
spectrum analyzers. But I don't recommend building any if you may buy
a new one, with about $1500, covering 1Mhz to 7GHz range and showing
you at least the frequency and amplitude.

greetings,
Vasile




The way I see it the

> most needed thing is low (very low) phase noise of the carrier so the LO does
> not swamp the mixer and the audio after it (although I am unsure what happens
> with the phase noise in a direct conversion receiver when an actual signal is
> not present). The scheme is the same as used in DSB/CW direct conversion
> receivers, whose performance is rather good usually. Also using a noise source
> as input makes it possible to go to very high frequencies without the usual
> problems. No ? Also it is a very simple and inexpensive scheme. With a $10 mixer
> it could cover dc to 4GHz. If a stabilised Gunnplexer is used then microwave
> filter (and Gunn and varactor bias, by voltage, to match an existing filter or
> structure) tuning becomes possible. What am I missing ? Ok, dynamic range is
> probably 55dB or less (sound card dynamic range). With a good sound card and a
> preamplifier this could be 90-100dB. Many people already have such a good sound
> card. If the LO is replaced with a stepwise DSS or PLL then a spectrum can be
> taken with say 40kHz 'windows' from DC (almost) to 1GHz. The DSS could be
> controlled by a USB to parallel control for example.
>
> Peter
>
>
> --
> > > >