SPECTRUM ANALYSER
The spectrum analyser is based on the W7ZOI design       PART 1       PART 2       SEE ALSO

I chose to design the analyser for 0 to 100 M Hz and settled on a 149.3 M Hz first IF.  This was because I already had a 160 M Hz Xtal oscillator and planned 10.7 M Hz (160 minus 149.3) second IF.
The VCO is a minicircuits POS300 and most of the other components were scavenged from the junk boxes., including a variety of double balanced mixers!
The POS 300 easily tunes from about 140 M Hz to over 300 M Hz with Plus 16V and minus 12 supplying the timebase.  
The first filter is critical and I found that a Toko 145 M Hz helical filter was quite selective enough and could be tuned to 150 M Hz.  However there were a few spurious responses and these were cleared by adding in series an old 144 transmitting bandpass filter which has very little loss but is too broad to be used on its own.  Its spurii must be different from the Toko filter as its addition cleared up all spurii.
The 10.7 filter is as in the articles and the narrow filter is a 25K FM Xtal filter, but as you will see in the pics below this is not ideal.
I used the modified Log amp using the ad603 and ad8307.  This chip is AMAZING!!  over 80db linear to log conversion and very linear on the output.
DISPLAY.
Only a scope with DC coupling and  xy input is needed. Low frequency response only
LAYOUT
Mindful of the need for screening I adopted a simple layout with all units in diecast boxes (mostly reused) and the many controls in the respective boxes fitted along the front of the analyser rather than design a custom front plate with leads to the controls.  This means that the only external interconnections are the power leads and the coaxial links.  The coax links were run through home made bulkhead type connectors to maintain integrity of the screening as suggested in the updates to the original articles. The faceplate was finished off with a half screen over the controls.  While not the most elegant solution this works well and looks reasonable..............
The finished analyser.  Note the method of construction used to avoid external leads and provide screening at a high level.
The final front.  Very functional and acceptable although not the most elegant
While the Pic a Star project can be constructed with minimal equipment this is a piece of test gear I wish I had had a lot sooner!!
PERFORMANCE
Well judge for yourselves also how useful it is!!
There are lots of screen shots below, but all reduced to 15K bits or so so should load quite quickly

The performance exceeded my expectations.  With the analyser set up for a reference of -20dbm there is 80 db of dynamic range and no spurs visible except marginally in the noise level at the -80 level.

The tracking generator also works well thanks to care in screening and has 70 db usable range.... feedthrough only limits the lowest 10db.


SCREENSHOTS
(Hand held digital camera shots!!) At the very bottom you will see comparison of the Pic and Mix DDS using the AD9851 (clocked at 160 M Hz) at 40 M Hz out and the Modified version using the 9951 clocked at 400 M Hz(Modification of G3XJP's code (Stddt08) thanks to DK5RM) also at 40 M Hz output. The spur difference speaks for itself!!

20 meter not M Hz!
NO FILTER ..............Spurs are 10.715 xtal and vfo at 14.4 M Hz (10.7 + 3.7)
absent with Tx filter in place!