Hi,

I just signed up on this group. I was very pleased to see that there
is a group for folks interested in home built spectrum analysers.

A bit of background.
I have dreamed of someday building my own analyser, for several years
actually, but about three years ago, I decided I would finally take
the plunge.

I built the analyser as described in Science Workshop.
It goes up to a little over 900 Mhz and I figured out how to get the
display working with both of the tuners I am using.  I also built the
tracking generator for my analyser.

I have had reasonable success with it. With an x-y monitor it works
decently and also with a scope.
I tried the program from Science Workshop and also some of the
software for soundcard oscilloscopes.  I am less than satisfied with
any of these.

I am not very satisfied with the fact that every time I have returned
to using the analyser I have found that I invariably have to re-
calibrate the display.  I don't mean it is just off by a few Khz, but
rather several Mhz.

Also I do not have much success with using the analyser and the
tracking generator at the same time.

I wonder if any of my present analyser can be used along with the
computer interface described on Scottie's site?

One of my main wishes is that I will end up with an analyser that I
can use to experiment with duplexer and other similar filters.

I would sure find it helpful to be able to exchange ideas and
experiences with anyone interested.

I have only just looked at the messages and I guess the group is just
starting up.  Perhaps this is not where I should be asking about TV
tuner based analysers and the problems associated with them! 

Cheers
Bruce, VE1II

When I mentioned 'when I return to using my analyser.....

By this I should have been more clear and said "when I start to use
my analyser after having not used it for several weeks..."

It maintains its frequency stability pretty well during extended
periods of use.  Of course it drifts around a bit.

bruce

Hi,

I just signed up on this group. I was very pleased to see that there
is a group for folks interested in home built spectrum analysers.

A bit of background.
I have dreamed of someday building my own analyser, for several years
actually, but about three years ago, I decided I would finally take
the plunge.

I built the analyser as described in Science Workshop.
It goes up to a little over 900 Mhz and I figured out how to get the
display working with both of the tuners I am using.  I also built the
tracking generator for my analyser.

I have had reasonable success with it. With an x-y monitor it works
decently and also with a scope.
I tried the program from Science Workshop and also some of the
software for soundcard oscilloscopes.  I am less than satisfied with
any of these.

I am not very satisfied with the fact that every time I have returned
to using the analyser I have found that I invariably have to re-
calibrate the display.  I don't mean it is just off by a few Khz, but
rather several Mhz.

Also I do not have much success with using the analyser and the
tracking generator at the same time.

I wonder if any of my present analyser can be used along with the
computer interface described on Scottie's site?

One of my main wishes is that I will end up with an analyser that I
can use to experiment with duplexer and other similar filters.

I would sure find it helpful to be able to exchange ideas and
experiences with anyone interested.

I have only just looked at the messages and I guess the group is just
starting up.  Perhaps this is not where I should be asking about TV
tuner based analysers and the problems associated with them!

Cheers
Bruce, VE1II

--- In spectrumanalyzer@yahoogroups.com, Will Wasson
<willwasson@c...> wrote:
> Hi Scotty,

> I usually use Netscape 6.2 to surf the web.  Yesterday I was using
> Internet Explorer and suddenly
> all your schematics and screen shots appeared.  These schematics
and
> screen shots were not
> appearing under Netscape.  I've turned off all web filtering
software on
> Netscape and still no
> schematics or screen shots.  It may be that there is some
inconsistency
> between Netscape and IE.
> You might want to add a banner at the top of your web page saying
that
> this site is best viewed
> with IE. 
>
> Thanks
> Will Wasson

  This a little beyond my expertese. I created the web pages in
Netscape Composer. I have both Netscape and I.E. on my machine and
both will view the pages without problem. Perhaps someone with
Netscape knowledge can lead you in the right direction.  Anyone?
Scotty

--- In spectrumanalyzer@yahoogroups.com, Will Wasson
<willwasson@c...> wrote:
> Hi Scotty,

> I usually use Netscape 6.2 to surf the web.  Yesterday I was using
> Internet Explorer and suddenly
> all your schematics and screen shots appeared.  These schematics
and
> screen shots were not
> appearing under Netscape.  I've turned off all web filtering
software on
> Netscape and still no
> schematics or screen shots.  It may be that there is some
inconsistency
> between Netscape and IE.
> You might want to add a banner at the top of your web page saying
that
> this site is best viewed
> with IE.
>
> Thanks
> Will Wasson

   This a little beyond my expertese. I created the web pages in
Netscape Composer. I have both Netscape and I.E. on my machine and
both will view the pages without problem. Perhaps someone with
Netscape knowledge can lead you in the right direction.  Anyone?
Scotty

Hi Scotty,

I usually use Netscape 6.2 to surf the web.  Yesterday I was using
Internet Explorer and suddenly
all your schematics and screen shots appeared.  These schematics and
screen shots were not
appearing under Netscape.  I've turned off all web filtering software on
Netscape and still no
schematics or screen shots.  It may be that there is some inconsistency
between Netscape and IE.
You might want to add a banner at the top of your web page saying that
this site is best viewed
with IE.

Thanks
Will Wasson

--- In spectrumanalyzer@yahoogroups.com, "w4zcb77" <w4zcb77@y...>
wrote:
The question
> for the day is...
> Do pins 10,12,13 and 15 on the DB25 just float?
>
> Regards
>
> W4ZCB
  Yes, pins 10, 12, 13, and 15 are status lines back to the computer,
and, in this design, they are not used. Let them float. I would
suggest, for those people laying out boards, go ahead and make pads
for the pins. Later additions and experimentations may want to "talk
back" to the computer program with these lines.
Scotty

Gmorning Scotty

I thought I sent you an e-mail, but think it bounced. The question
for the day is...
Do pins 10,12,13 and 15 on the DB25 just float?

Regards

W4ZCB
(I tried to instant message you but I'm a yahoo neophyts.) Nice
picture.

Good news everyone.  The Libery Basic version 4.0, will run the SSA
program for FREE. Yes, I said FREE!  Go to www.libertybasic.com and
download it.  Scotty

Hi Scotty,

Thanks again!  You probably saved me days of lost work milling out an
aluminum block.
Silly me, I thought that a milled block should provide "battle ship"
type shielding for the
unit.  I was planning on providing screws every few inches to tie it
down to the PC board.
It would've taken me days to get the block milled and the screws tapped
and threaded.
And it might have worked, but as you noted "Murphy" would have come
along someday
and zapped a part , and if the block had to be soldered down that would
have been it! :-(

The coffee can lids is an idea that has merit.  I do have tin snips, so
my wife won't chase
me down the street with her ruined scissors inches from my rear :-)

A TIG is on my wish list, but until I can find a home repair that
"needs" a TIG to get it done
it will have to wait.  Of course my wife is still wondering why I needed
a mini-lathe to
redo the bathroom floor,  ("Oh yeah Hon, there were some special
connectors under the
floor that Home Depot doesn't carry anymore, and to reseat the toilet I
needed to cut
and turn new brackets". )

Will

>
>
>
>
> I have used the milled cavities in many professional designs. Mainly,
> for RF modules that were part of card bus systems (which are horribly
> noisy). The only problem I have had with the method is good RF
> grounding of the "guard rings" to the aluminum. In a couple of
> designs, having a mounting screw every 1 inch was not enough to keep
> 1 GHz from leaking into the next cavity.  I made the process
> engineers electroplate the aluminum, screwed it to the board, and
> then "sweat soldered" the alu. to the guard rings. That pretty well
> blew the mechanical engineers design for repairability. But, it
> passed all EMI testing. I would say, proceed with your plan. It will
> make for a beautiful single module.
>   As far as using the block for the cavity filter. I suppose you
> could mill the four cavities from a single block with the "top plate"
> integral to the block. The internal 1/4 inch stubs must be well
> soldered to the "bottom plate" and it, in turn, must be well soldered
> to the cavity block. If you are cool enough to have a milling
> machine, you are probably aware of the solderability of aluminum. Do
> you have a TIG welder?
> I use coffee can lids for shielding. Good solderability and can be
> cut with a good, stiff pair of sissors. Don't borrow your wife's.
> Divorce is likely!!
> Scotty
>

--- In spectrumanalyzer@yahoogroups.com, "willwasson_94550"
<willwasson@c...> wrote:
> Hi Scotty,

> Most of the single supply, rail2rail devices
> satisfy the output swing requirements, but
> the GBWP of the devices are pretty low.  I see
> many around 90KHz or so, and I'm not sure
> if they would be fast enough for your feedback
> loop.

The gain bandwidth product of the op amp does not need to be very
high.  In the presently designed loop, the "0" gain point is at 41
KHz.  The PLL 1 phase frequency is about 350 KHz and the pre filter
pole in the loop is 340 KHz. An op amp that is "lousy" at these
frequencies is actually beneficial. It will help attenuate the side
bands in the VCO that are created by the PLL's phase discriminator.
The minimum GBWP of the op amp should be at least 2 times the maximum
of the loop BW, in this case, 82 KHz. However, the experimentor using
faster loops may want to have a faster op amp, probably a 500 KHz
would be the maximum he would need.
Scotty

> With the lowering of the +12v requirement to +10v,
> I can see this whole unit being powered by a
> single 12v wall wart supply. That is if I can
> get rid of the -5v supply :-)

We still need about +20 v for the op amp. The VCO needs about +18v
control voltage to get to 2013 MHz. I have had best results when
double regulating all of the voltages in the SSA. For example, the
+12 regulators should be fed by a +15v or +20v regulator.  The +5v
regulators should be also be fed by a pre-regulator. The +20v and -5v
to the op amp need not be preregulated. Scotty

> An alternative would be to add a dc to dc converter
> to convert the +12v input to -5v.  But this
> has the problem of pushing switching noise into
> your design, and I'm not sure if this would
> be good idea.

If it is well filtered and shielded, it might not create a problem.
It is certainly worth trying. It would manifest itself when seeing
the switching frequency as sidebands on high level signals you are
analyzing.
Scotty

Idealy, I'm planning on using a
> block of aluminum the size of the circuit board,
> and using my milling machine to make cutouts for
> each module.  The walls of the cutouts will contact
> guard rings around each module on the PC board.
> This will provide the interstage shielding along
> with the full ground on the other side of the PC
> board. So maybe a dc to dc converter wouldn't be too
> bad.
>
> In fact, I was wondering if the cavity filter could
> be milled into the same aluminum block.
>
> In a previous post you said the you were a real
> stickler for layout design.  So I'd really value
> your opinion.  Am I getting too far out in left
> field with this aluminum block shield idea :-)

> Thanks!
>
> Will


I have used the milled cavities in many professional designs. Mainly,
for RF modules that were part of card bus systems (which are horribly
noisy). The only problem I have had with the method is good RF
grounding of the "guard rings" to the aluminum. In a couple of
designs, having a mounting screw every 1 inch was not enough to keep
1 GHz from leaking into the next cavity.  I made the process
engineers electroplate the aluminum, screwed it to the board, and
then "sweat soldered" the alu. to the guard rings. That pretty well
blew the mechanical engineers design for repairability. But, it
passed all EMI testing. I would say, proceed with your plan. It will
make for a beautiful single module.
   As far as using the block for the cavity filter. I suppose you
could mill the four cavities from a single block with the "top plate"
integral to the block. The internal 1/4 inch stubs must be well
soldered to the "bottom plate" and it, in turn, must be well soldered
to the cavity block. If you are cool enough to have a milling
machine, you are probably aware of the solderability of aluminum. Do
you have a TIG welder?
I use coffee can lids for shielding. Good solderability and can be
cut with a good, stiff pair of sissors. Don't borrow your wife's.
Divorce is likely!!
Scotty

Hi Scotty,

Thanks for the reply.

Most of the single supply, rail2rail devices
satisfy the output swing requirements, but
the GBWP of the devices are pretty low.  I see
many around 90KHz or so, and I'm not sure
if they would be fast enough for your feedback
loop.

With the lowering of the +12v requirement to +10v,
I can see this whole unit being powered by a
single 12v wall wart supply. That is if I can
get rid of the -5v supply :-)

An alternative would be to add a dc to dc converter
to convert the +12v input to -5v.  But this
has the problem of pushing switching noise into
your design, and I'm not sure if this would
be good idea.  Idealy, I'm planning on using a
block of aluminum the size of the circuit board,
and using my milling machine to make cutouts for
each module.  The walls of the cutouts will contact
guard rings around each module on the PC board.
This will provide the interstage shielding along
with the full ground on the other side of the PC
board. So maybe a dc to dc converter wouldn't be too
bad.

In fact, I was wondering if the cavity filter could
be milled into the same aluminum block.

In a previous post you said the you were a real
stickler for layout design.  So I'd really value
your opinion.  Am I getting too far out in left
field with this aluminum block shield idea :-)

Thanks!

Will




--- In spectrumanalyzer@yahoogroups.com, "Scotty" <wsprowls@y...> wrote:
> --- In spectrumanalyzer@yahoogroups.com, Will Wasson <willwasson@c...>
> wrote:
> > Hi Scotty,
> >
> > As far as I can see the Op27 amplifier is the only device that
> requires
> > a negative voltage, (-5v). Do you know if there is a single supply,
> > rail to rail op-amp that could replace it to save on having a
> negative
> > supply.
> >
> > Will
>
> I didn't do a great deal of searching for single supply op amps. Maybe
> someone knows of a good one. The requirements for the op amp are:
> The output MUST go down to 1.0 volt or lower to swing the VCO to its
> lowest frequency.
> The input offset current and input noise current should be as low as
> possible, in the pico amp range would be best. Offset voltage is of no
> concern.
> Must allow +20v operation on Vcc, and the output should swing to
> within 2 volts of Vcc to allow VCO to swing to its highest frequency.
> My OP 27 is drawing less than 2 ma from the negative supply. -5v is
> not critical.
>  As a matter of fact, my original breadboard's negative voltage was
> created by ac coupling the positive bridge rectifier in my power
> supply and rectifying it to -3v.
>   The OP 27 will swing to within 1.5 volts of either rail. Some of the
> newer ones will swing closer than that.  I used the OP27 because I had
> a few, and they have been the workhorse of loop filters for many
> years. But, it is really out of date.
> Scotty Sprowls

--- In spectrumanalyzer@yahoogroups.com, Will Wasson <willwasson@c...>
wrote:
> Hi Scotty,
>
> As far as I can see the Op27 amplifier is the only device that
requires
> a negative voltage, (-5v). Do you know if there is a single supply,
> rail to rail op-amp that could replace it to save on having a
negative
> supply.
>
> Will

I didn't do a great deal of searching for single supply op amps. Maybe
someone knows of a good one. The requirements for the op amp are:
The output MUST go down to 1.0 volt or lower to swing the VCO to its
lowest frequency.
The input offset current and input noise current should be as low as
possible, in the pico amp range would be best. Offset voltage is of no
concern.
Must allow +20v operation on Vcc, and the output should swing to
within 2 volts of Vcc to allow VCO to swing to its highest frequency.
My OP 27 is drawing less than 2 ma from the negative supply. -5v is
not critical.
  As a matter of fact, my original breadboard's negative voltage was
created by ac coupling the positive bridge rectifier in my power
supply and rectifying it to -3v.
   The OP 27 will swing to within 1.5 volts of either rail. Some of the
newer ones will swing closer than that.  I used the OP27 because I had
a few, and they have been the workhorse of loop filters for many
years. But, it is really out of date.
Scotty Sprowls

To all,
Much to my dismay, I found errors on both the PLL 1 and PLL 2
schematics.  The LMX 2325 pins 1 and 10 are reversed.
The schematics on the web page have been changed to reflect:
The RF sense from the VCO should go to the input of the LMX2325 on
pin 10.
The reference clock to the LMX2325 is pin 1, not pin 10.
For those of you who have finished layout and printed your board, I am
on my way to Mexico to escape drive-by shots into my house.
I'm sorry for the goof.
Scotty Sprowls

Hi Scotty,

As far as I can see the Op27 amplifier is the only device that requires
a negative voltage, (-5v). Do you know if there is a single supply,
rail to rail op-amp that could replace it to save on having a negative
supply.

Will

As W4ZCB kindly pointed out to me, I have made an error in my designs
of the PLL 2 and the Tracking Generator.  I show a +12v, LM7812
feeding the ROS-1500 VCO's.  The VCOs are specified for 10v with
a
max input of 11v.
I changed the parts list, and both schematics to reflect the 10v.
Change your voltage regulators to 7810 or 78M10.  If you are using an
adjustable regulator at these two points, change them to +10v.  The
schems are now marked REV A for this change.
Scotty Sprowls

Absolutely, you can substitute the filter in the DDS module. Use any
narrow band filter you have on hand. The 10.7 or 10.965 I show is
because they are junk box specialties. I have experimented with a 5
MHz / 5 KHz bw xtal filter, and an 8.9 MHz/4KHz bw filter, both with
success. The ONLY stipulation is: the filter must not allow a
subharmonic of the DDS clock through. This will show up as a spur on
the output.  For example, if the DDS clock is 64 MHz, the 1/6 spur is
10.666667 MHz. Don't use a 10.70 MHz filter with a bandwidth greater
than 34 KHz. Actually the chances of grabbing a narrow crystal filter
out of the junk box that WON'T work is slim.  Play with the resistors
feeding and loading the filter to match its impedance. In reality the
impedance can be pretty sloppy around this filter and it will still
work fine.
It may be obvious, but I will mention that the filter MUST be ac
coupled to and from the DDS chip.  I didn't show any blocking
capacitors in the circuit, because I have never run into a crystal
filter that was dc coupled to ground.
Scotty

I found an error on my PLL 1 schematic. When adding the optional
Tracking Generator divider at VCO 1, I left out the coupling
capacitor. It can be any value between .001 and .01 uf. Without the
cap, the PLL RF input would be pulled to a low voltage and become
inop. Sorry about that! I would also mention, if you use a VCO that
does NOT have an ac coupled output, a coupling capacitor must be added
at its' output for the same reason.
The schems on the website are now updated for this error.
Scotty

I found an error on my PLL 1 schematic. When adding the optional
Tracking Generator divider at VCO 1, I left out the coupling
capacitor. It can be any value between .001 and .01 uf. Without the
cap, the PLL RF input would be pulled to a low voltage and become
inop. Sorry about that! I would also mention, if you use a VCO that
does NOT have an ac coupled output, a coupling capacitor must be added
at its' output for the same reason.
The schems on the website are now updated for this error.
Scotty

For those of you who want to experiment with an interdigital filter
rather than construct my cavity filter, this is an on-line computation
for the mechanical design of an interdigital filter.  It should give
you some pretty good results.  http://www.wa4dsy.net/cgi-bin/idbpf
I added it to the Link page.

I received this from Peter Vekinis this morning and thought is was
relevant to the group page:

Peter.Vekinis@...
Thu, 5 Feb 2004 12:02:49 +0100
Scotty, thanks for putting together the spectum analyzer. I have been
saving to buy one but now maybe I can save that money
Questions: Why do we need a cavity?
Can this be simplified further?
Can we not use a PC to drive the DDS directly, use one Mixer and A/D
through the Joystick port (which is an A/D detector (8 bits)? Will
you offer kits? Great idea! Peter, KC1QF/LX1QF, in Luxembourg

And my answer to him was:
Hello Peter,
The reason for the cavity is to keep the analyzer to a 2 conversion
process. I could have designed for a 3 conversion and used a
conventional filter in the first IF stage. However, with more
conversions, more internal intermodulation spurs are created.  The
cleaner the Spectrum Analyzer, the better the measurements you will
be able to take.  Your suggestion of single mixing with a DC output
is a viable plan at higher frequencies.  There are no VCO's that will
tune from DC to 1000 MHz, but it would work well in the range of 1000
to 2000 MHz.
I don't have a joystick port on my laptop so I can't comment on its
operation.
No kits yet, but stick around. I have a very good feeling someone is
going to come up with one. Keep tuned to my website for updates. (be
sure to use your "refresh" button)
Best wishes,
Scotty Sprowls

> Hi W4ZCB,
>
> What lay out program are you using?
> I've just started creating my schematic elements
> and footprints in WinQcad.  I should be able to
> start schematic capture, layout, and routing in
> a couple of days.

I use Autocad and have for years. It doesn't have an autorouter, but
I can have more widths of lands and pads than too big and too small.
>
> Since you are ahead of me, do you have any warnings
> or cautions that I should look out for?

Not really, I'd make the lands for the "IF strip" 50 ohms and vias
all along the interface however. That's a lot of gain to keep stable.
I used a different MMIC, and have 60 dB of nice low noise stable
gain. But I have also gone to extremes in terms of decoupling and
impedance matching to do it.
>
> Where are you planning on having your boards fabbed?
> I have been very happy with Olimex out of Bulgaria.
> Yeah, I know, Bulgaria?  They're fast and cheap though.
>
I make my own prototypes with PnP blue. ABout 20 minutes from
finished layout to finished board. I've just received 22 boards for
another project from the Bulgarian folks. Sent them the Gerbers and
they got one drill hole wrong. Not the end of the world, an easy work
around, but I don't do that when I make my own.

Faster and more expensive, is Express PCB in CA. You use their
software and have the boards 3 days ARO. The quality is outstanding.

W4ZCB

I've spent most of the day looking at RSSI chips. You are right, the
AD series are outstanding. These are usable for the 10MHz range..
AD8306 @ $18.50, AD8307 @ $7.10, AD8309 @ $9.75, AD8310 @ $4.95,
AD606 @ $27.45
Without having any of these in my grubby hands to test, the specs of
the AD8310 would be my favorite.  It's range is -90 dBm to +5 dBm.
Same slope as the others, but would require less amplification in
front of it to drive it to full scale.
Scotty

Hi W4ZCB,

What lay out program are you using?
I've just started creating my schematic elements
and footprints in WinQcad.  I should be able to
start schematic capture, layout, and routing in
a couple of days.

Since you are ahead of me, do you have any warnings
or cautions that I should look out for?

Where are you planning on having your boards fabbed?
I have been very happy with Olimex out of Bulgaria.
Yeah, I know, Bulgaria?  They're fast and cheap though.

Will

Well, the boards are etched. The PLL 1 board fought me to the bitter
end. No matter where I put a land, I didn't want it there later. At
any rate, it finally came together. (Those 0.025 pads for the 2353's
are a bear, glad I've done them before!) It all became worthwhile,
when I realized that with a little judicious editing, I could use the
same artwork for PLL 2 that I used for #1.

Re the log amps, There is RSSI and then there's RSSI. For log
linearity, no prisoners approach, you can't beat the Analog devices
AD8307. Second place and VERY close, is the 8310 from the same folks.
I've used Mot's and Philips best and they aren't even in the same
galaxy.

Regards

W4ZCB

I have done a little digging and have found some potentially good
substitutions for the Log Det, MC3356. Although there are several
suppliers who have stock of the 3356's, we could play with other
alternatives.  Seems that Philips has some pretty good ones.
SA604AD,  SA605D,  SA614AD, SA615D,  SA636DK,  SA639DH. From looking
at these, the SA615 has the largest dynamic range log output. All of
these will work up to 25 MHz. Also, the packages are the itty bitty SO
surface mount series.  Prices are darn good, though: $3.39 to $6.59
each.
The SA615 is SO20 package at $4.77 from Arrow Distributors, qty of 1.
All of the above can be found at:
http://www.semiconductors.philips.com/similar/SA604AD.html
I will put this on the link.

Hi Scotty,

Just checking in.

Thanks for creating this site, this journey is going to be FUN!

Will

I'm here Scotty. but it's also bedtime. CU tomorrow.

W4ZCB

I'm Scotty Sprowls, the guy behind the Scotty Spectrum Analyzer.
Most everything you need to know about this Spectrum Analyzer project
can be found at  http://users4.ev1.net/~wsprowls/
I'll try to monitor this group page as much as possible and respond
to questions and comments directly on this page.