EMRFD Message Archive 1445

Message Date From Subject
1445 2008-02-20 18:13:23 mdgolfbum SA Narrow Crystal Filter?
Has anyone done a design for a ~3Khz crystal filter for the RBW section
of the W7ZOI spectrum analyzer? I can take a run at doing one myself
with the info on the CD.

I'm working on my SA project now and for my purposes I'd like all
three: 300Khz, 30Khz and 3Khz. Of course I'll have to make some
adjustments to the sweep circuitry as well.

I'm planning on using a separate 1590A for the 30Khz 8pole filter and
will do the same for the 3Khz. I'm boxing each of the major
subassemblies in Hammond boxes. I'm also planning on doing the triple
helical filter at 110Mhz rather than the three section one in a single

All this will take a little longer but I think it will be worth it.

jim ab3cv
1446 2008-02-20 20:19:38 Wes Hayward Re: SA Narrow Crystal Filter?
Hi Jim, and gang,

I have tried a 3 kHz filter in my variation of the spectrum analyzer,
but the phase noise in the VCO is just to high. Some sort of phase
lock system is needed. A VCO up at the same frequency that had a
much smaller span might be quiet enough. There are some spectrum
analyzers out there that go down to a RBW of 1 kHz with a free
running YIG microwave oscillator. (The example I'm thinking of is
the HP-8558B.) But the POS 200 does not seem to be that quiet.

This is speculation at this point. It might be that the VCO is
quiet enough, but noise from the sweep system generates oscillator
noise when it is applied to the VCO. If this is the case, a large
capacitor on the VCO line right at the VCO input might do nice things
to filter sweep generated noise. It would, of course, distort the
sweep, but one can live with a little bit of this.

I was able to experiment with a 3 kHz filter merely by dropping one
into the system in place of the 300 kHz filter, which was a stand
alone box in my system. This made it handy to discover the real

Most of us have built the analyzer as a collecti
1447 2008-02-20 22:19:55 kerrypwr Re: SA Narrow Crystal Filter?
Since all crystals (at least those available to experimenters at an
affordable price) are different, there is no single set of values for
a particular design.

I obtained 50 5-MHz crystals from Futurlec
( http://www.futurlec.com/cgi-bin/search/search.cgi ) last year and
characterised them using both the G3UUR technique and the resonance
technique; both are well-covered in EMRFD.

I spent much of today "building" filters with the AADE program;


The outcome of this massive learning experience will be my version of
Wes' Measurement Receiver; the mixer module is finished (generally in
accord with the EMRFD circuit but I have included a diplexer) and the
next step is a pair of filters (250 Hz & 2500 Hz, reed-relay-switched)
, possibly a post-IF-amp and output to my AD8307 meter.

So; if you are going to build crystal filters, that is the journey you
should undertake.

It's a pleasant journey; few projects have given me as much pleasure
and sheer fun as this one!
1448 2008-02-21 04:05:56 Jim Miller Re: SA Narrow Crystal Filter?
Hi Wes

Thanks for the info. My unit will be all Hammond boxes with SMAs so it will
be easy to move things around and try out new modules. I got a bunch of
females with hardline pigtails from ebay to make the boxes easier.

Even the 30Khz piece will be in it's separate 1590a sidecar to the main
300Khz filter so it will be easy to substitute a 3Khz one.


jim ab3cv
1452 2008-02-21 08:45:45 rotfunkblau Re: SA Narrow Crystal Filter?
Hi Wes ,

in old spectrum analyzers (e.g. HP141T) the stabilisati
1453 2008-02-21 08:45:51 rotfunkblau Re: SA Narrow Crystal Filter?
Hi Wes ,

in old spectrum analyzers (e.g. HP141T) the stabilisati
1456 2008-02-22 04:19:57 pamunoz2001 Re: SA Narrow Crystal Filter?

The 8555A/141T has a stabilizer, but it is a frequency locked loop,
and its purpose is only to avoid the frequency drift of the LO, that
is noticeable in narrow spans. Since it is not a PLL, it does not
reduce the phase noise.

Actually, in the 8555A that I have there is a slight increase in
phase noise when I turn the stabilizer on. I'll have to investigate
that later... The 8558B doesn't have a stabilizer, and doesn't need
one for the 1KHz rbw.


1457 2008-02-22 04:28:10 pamunoz2001 Re: SA Narrow Crystal Filter?

I have not tried the POS-200, but I think it should be quiet enough
to use with a 3Khz filter.

According to the specifications of the 8558B, the phase noise is
about 81dBc/Hz at 10Khz from the carrier. This is quite similar to
the phase noise of the POS-200 at the same frequency offset. The
problems with a resolution bandwith of 1 or 3Khz come from the shape
factor of the filter. The 8558B has filters with a shape factor of
1:15 (3/60dB), while the typical crystal filter for communications
has a shape factor of 1:2 or less. The very steep skirts of this type
of filters is what makes the phase noise visible on the screen.

I have been playing for a long time with an old 8555A spectrum
analyzer front-end. I bought it "as is" on ebay, so I donĀ“t have the
140T/141T display or the 8552B IF, and I don't intend to buy them
because they are are too bulky and heavy. Therefore, I am in the
process of building a suitable IF, log amplifier, etc. to make it
work. For this I have used some sections of the W7ZOI spectrum

I am building the rbw filters inspired in (well, copying) the filters
of the 8558B. These are very complex, but a beautiful design that is
capable of variable bandwith from 50KHz down to 1KHz or less with the
same gain. The complete filter has 4 crystals, and a shape factor of
1:15. I have built only 2 sections, and only with very narrow
resolution (a few hundred Hz) the phase noise of the YIG is
detectable. On the other hand, with a 15KHz-wide crystal filter from
a FM receiver, the noise sidebands are easily visible because the
skirt is steeper.

Another problem is that as the selectivity is increased the sweep
time needs to be very long (maybe ten seconds or more). As I am using
a regular scope, the persistence of the display is not sufficient. So
eventually I will need to put an A/D converter and use a PC as a
display. That opens a lot of new posibilities, buy is a lot of work
and I need to finish the RF part first.

For HF work, I would like to build a new front end. The 8555A only
goes down to 10MHz and its IIP3 is quite poor (about +5dBm). Besides,
the first mixer is very fragile, and if I destroy it the replacement
could cost more than the complete 8555A. I would use a more strong
diode ring mixer. For the VCO, I would like to experiment with an
helical resonator. That should give better phase noise than a POS-xxx
type VCO. But that is another projetct...

1462 2008-02-22 15:01:31 rotfunkblau Re: SA Narrow Crystal Filter?
Hello pamunoz2001 (sorry didn't get your name)

... hmmm, maybe I've not been quite correct with the older spectrum
analyzers. But nevertheless the concept I've outlined is valid and
found in the magazine UKW-Berichte 3/1987, used by DB1NV to build
a SA with a few kHz of bandwidth.

My understanding of the concept is that the PLL (maybe unlike a
frequeny loop!) is reducing (at least theoreticaly)every 'noise'
within the bandwidth of the loop. This 'noise' could also be ac hum
or whatever.

By building my analyzer I've found that one have to keep an eye on
hum and also magnetic shielding (i.e. ferrite cores can
be 'modulated'). Also of importance is the noise of the power
supplies, special the tuning supply.

I fully agrees with your outline of the filters and the POSxxx.

The filters should have Gaussian shape. An realized design can be
found at

To me the discussion seems to be obsolete, because going to narrower
bandwidth is not neccesary, because these measurements can be done
today with the help of a PC, a mixer, simple xtal oscillator and
spectrum analyser software.(I think this is also outlined in EMRFD).

The bandwiths of the W7ZOI analyzer are choosen very well, the
desing is careful balanced. 'Improving' on one end might cause
trouble with other things that end in an effort not worth spending
(at least for me). No it is not the perfect SA (W7ZOI never stated
this), but regarding the effort/outcome rati