Filters and Spectrum Analyzer.

[Geoffrey Mackenzie-Kennedy-2]

Hi Brett,

My personal preference when looking at the response of any filter is to use
a RF Spectrum Analyzer and a tracking signal generator. When it comes to
looking at the response of IF filters already built into a receiver the
tracking generator can work at the antenna frequency rather than at IF,
which makes the generator - pad - receiver connection easy to make, provided
that the front end selective circuits have no effect on the measurement of
the IF filter i.e.their working bandwidth is >>> IF filter bandwidth. The
penalty is that the input of the Spectrum Analyzer needs to be connected to
the IF strip at some point before the product detector, and everything after
this point has to be disconnected unless the working bandwidth of this part
which follows is >>> IF filter bandwidth, *or* has no effect on the
measurement. One also has to ensure that the Spectrum Analyzer and its
connecting cable are transparent to the cascade under measurement, a high
impedance probe helps. AGC has to be OFF of course if  the AGC system could
be active. Spectrogram does not require this fiddling, AGC must be OFF, but
being in effect a Spectrum Analyzer working at audio it gives a limited
view - about 20kHz IIRC. What I like about using a RF Spectrum Analyzer and
tracking generator is that a cleaner picture is produced without fuzzy noise
at the top, and it is possible to sweep over a wide bandwidth to view the
ears and spurious responses of the filter. It is also usually possible to
see those other unwanted responses caused by radiated feedback or coupling
via pcb traces etc.

It could be argued that Spectrogram is easier to use for routine alignment,
but has limitations for use as a diagnostic tool.

73,

Geoff
GM4ESD

 


_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Jack Smith-6]

Geoff:

Your approach assumes that the tracking generator has a programmable
offset, doesn't it? RF Generator output, for example, sweeps from 7 -
7.4 MHz with the K2 tuned to 7.2 MHz, but the detector needs to sweep
5115 - 4715 KHz, if I've got the maths right. (Sweep also must be
inverted to see the display correctly due to K2's inversion on 7 MHz band.)

My VNAs don't have that feature, but I believe frequency offset is
available in some of Agilent's current production VNAs. With an
auxiliary detector, such as an HP3400 RF voltmeter, however, I could use
my HP8752B VNA's "vertical axis input is auxiliary DC voltage" feature,
taking the 0...1V recorder output from the 3400 as the VNA's auxiliary
input. That gives the advantage of a high Z detector, but the
significant disadvantage of broadband input. The other option would be
to lash up a program to command a signal generator and spectrum analyzer
via their GPIB control buses with the flexibility in frequency stepping
that provides.


Jack

Geoffrey Mackenzie-Kennedy wrote:

> Hi Brett,
>
> My personal preference when looking at the response of any filter is
> to use a RF Spectrum Analyzer and a tracking signal generator. When it
> comes to looking at the response of IF filters already built into a
> receiver the tracking generator can work at the antenna frequency
> rather than at IF, which makes the generator - pad - receiver
> connection easy to make, provided that the front end selective
> circuits have no effect on the measurement of the IF filter i.e.their
> working bandwidth is >>> IF filter bandwidth. The penalty is that the
> input of the Spectrum Analyzer needs to be connected to the IF strip
> at some point before the product detector, and everything after this
> point has to be disconnected unless the working bandwidth of this part
> which follows is >>> IF filter bandwidth, *or* has no effect on the
> measurement. One also has to ensure that the Spectrum Analyzer and its
> connecting cable are transparent to the cascade under measurement, a
> high impedance probe helps. AGC has to be OFF of course if  the AGC
> system could be active. Spectrogram does not require this fiddling,
> AGC must be OFF, but being in effect a Spectrum Analyzer working at
> audio it gives a limited view - about 20kHz IIRC. What I like about
> using a RF Spectrum Analyzer and tracking generator is that a cleaner
> picture is produced without fuzzy noise at the top, and it is possible
> to sweep over a wide bandwidth to view the ears and spurious responses
> of the filter. It is also usually possible to see those other unwanted
> responses caused by radiated feedback or coupling via pcb traces etc.
>
> It could be argued that Spectrogram is easier to use for routine
> alignment, but has limitations for use as a diagnostic tool.
>
> 73,
>
> Geoff
> GM4ESD
>
>
>
>
> _______________________________________________
> Elecraft mailing list
> Post to: [hidden email]
> You must be a subscriber to post to the list.
> Subscriber Info (Addr. Change, sub, unsub etc.):
> http://mailman.qth.net/mailman/listinfo/elecraft   
> Help: http://mailman.qth.net/subscribers.htm
> Elecraft web page: http://www.elecraft.com
>

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Jack Smith-6]

I should add another option would be to use an outboard mixer and signal
generator and create a mirror image of the K2's internal process.

Tracking Generator --> MIXER --> to K2

Mixer LO driven from a signal generator set to provide the correct
offset and frequency inversion.

Adds a bit of complexity to the equation, but should work.

Jack

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Geoffrey Mackenzie-Kennedy-2]

Jack,

Thank you, I should have clarified this point. As I do not own a tracking
generator with built-in programmable offset, when looking at a filter when
there is a receiver signal mixer between the generator and the spectrum
analyzer I use the outboard mixer plus generator approach, with the spectrum
analyzer's swept oscillator providing one of the two injection sources for
tracking. This swept oscillator works in the same frequency region
regardless of the input working frequency of the spectrum analyzer. A stable
generator provides the second fixed frequency injection source for the
outboard mixer, with the mixing scheme - high side or low side - allowing
choice of whether sweep inversion is introduced or not. Using a tracking
generator with programmable offset would be a better choice to avoid any
spurious products produced by the outboard mixer, but by using a H-mode
mixer in this application potential spurs have not intruded.

I use the same hardware for looking at stand alone filters and other
circuitry when an offset is not required, so in effect the outboard mixer,
generator and buffers is my tracking generator.

73,

Geoff


Jack Smith wrote:

> Geoff:
>
> Your approach assumes that the tracking generator has a programmable
> offset, doesn't it? RF Generator output, for example, sweeps from 7 - 7.4
> MHz with the K2 tuned to 7.2 MHz, but the detector needs to sweep 5115 -
> 4715 KHz, if I've got the maths right. (Sweep also must be inverted to see
> the display correctly due to K2's inversion on 7 MHz band.)
>
> My VNAs don't have that feature, but I believe frequency offset is
> available in some of Agilent's current production VNAs. With an auxiliary
> detector, such as an HP3400 RF voltmeter, however, I could use my HP8752B
> VNA's "vertical axis input is auxiliary DC voltage" feature, taking the
> 0...1V recorder output from the 3400 as the VNA's auxiliary input. That
> gives the advantage of a high Z detector, but the significant disadvantage
> of broadband input. The other option would be to lash up a program to
> command a signal generator and spectrum analyzer via their GPIB control
> buses with the flexibility in frequency stepping that provides.
>
>
> Jack

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Ron D'Eau Claire-2]

Spectrogram is also limited by the dynamic range of many sound cards. I've
used it on some computers that had an internal noise floor barely 30 db
down! Also the accuracy of the crystal clock in the sound card affects the
accuracy of frequency measurements.

Even so Spectrogram is an extremely worthwhile tool for the purposes it has
been recommended: align the BFO frequency to center the bandpass on the
receive frequency and to see, clearly and graphically, how changes affect
the bandwidth and position relative to the center frequency. Considering the
price (free) when downloaded from Tom, N0SS's site
(http://www.n0ss.net/index_k2.html) it's hard to beat the "investment" for
anyone who has a PC running Windows handy.

Ron AC7AC

-----Original Message-----
Hi Brett,

My personal preference when looking at the response of any filter is to use
a RF Spectrum Analyzer and a tracking signal generator. When it comes to
looking at the response of IF filters already built into a receiver the
tracking generator can work at the antenna frequency rather than at IF,
which makes the generator - pad - receiver connection easy to make, provided

that the front end selective circuits have no effect on the measurement of
the IF filter i.e.their working bandwidth is >>> IF filter bandwidth. The
penalty is that the input of the Spectrum Analyzer needs to be connected to
the IF strip at some point before the product detector, and everything after

this point has to be disconnected unless the working bandwidth of this part
which follows is >>> IF filter bandwidth, *or* has no effect on the
measurement. One also has to ensure that the Spectrum Analyzer and its
connecting cable are transparent to the cascade under measurement, a high
impedance probe helps. AGC has to be OFF of course if  the AGC system could
be active. Spectrogram does not require this fiddling, AGC must be OFF, but
being in effect a Spectrum Analyzer working at audio it gives a limited
view - about 20kHz IIRC. What I like about using a RF Spectrum Analyzer and
tracking generator is that a cleaner picture is produced without fuzzy noise

at the top, and it is possible to sweep over a wide bandwidth to view the
ears and spurious responses of the filter. It is also usually possible to
see those other unwanted responses caused by radiated feedback or coupling
via pcb traces etc.

It could be argued that Spectrogram is easier to use for routine alignment,
but has limitations for use as a diagnostic tool.

73,

Geoff
GM4ESD


_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com