A wee while ago I bought a secondhand Kenwood RZ-1 scanner, I know not why. It covers 500KHz to 905MHz and has AM, FMn and FMw. As it is a fairly rare beastie, I wrote a magazine article about it and then began wondering two things: 1- why was the shortwave performance so ghastly, and 2- would it be possible to make it tune above 1000MHz? While I was looking into the SW performance issues, I slipped with the 'scope probe and popped the prescaler chip and its controller. Oh fiddlesticks, I exclaimed! A friend found a close equivalent for the prescaler and after fitting it the RZ-1 worked from 60MHz to 905MHz but SW seemed lost forever, thank God. So then I turned my attention to the +1GHz coverage. The easiest answer was to make a downconverter which would subtract 1000MHz from incoming signals, so that for example 1200MHz would appear at 200MHz.  The first thing to do was to decide on a way of generating a stable, clean 1000MHz signal. A crystal/multiplier? No, too involved and potentially sproggy. A simple free-running oscillator? Tempting, but drifty. So a phase-locked solution seemed the way forward, and I just happened to have a few 950-1050MHz VCO blocks in stock. In addition I would need a reference oscillator (25MHz blocks in stock!), a divider to bring the output from the VCO down to 25MHz and a phase detector that would work at 25MHz, all also kicking about in the shack.  Off we go then...

I assembled the parts on a piece of double-sided copper board, I managed to fit the PLL and mixer in about half the space available, leaving plenty of room for the RF section. I discovered the NE564 phase detector output needed inverting in order to control the VCO, so a simple transistor stage was squeezed in. After a little fiddling the loop locked and out came a clean, stable 1GHz signal which was amplified and then applied to the diode-ring mixer.
The signal section consisted of a HPF (roll-off below 1000MHz) and a single mmic RF amplifier, the signal from which went to the other input of the mixer. The mixer's output (0 to 1000MHz) was also amplified before finding its way to the output lead.
It worked! I almost installed it and called it done, but I was concerned at its sensitivity - or lack of. There's a broadcast link at about 1.5GHz around here - on my AR5000 it is a weakish but usable signal, but on the converted RZ-1 it was poor by comparison.
I tried adding another mmic RF amplifier, but the noise figure of these devices meant that the noise outweighed the signal.
Time for a Mk2.....

Another piece of board and another day at the bench! The design was almost identical to the Mk1, but the layout was changed to make way for a low-noise satellite line amp board. Here's a photo of the Mk2. Starting at the VCO (the largest tin box) and working anti-clockwise we have the divider (a wee board I built a while ago) which divides the 1000MHz by 40 to give 25MHz, then the 25MHz reference osc, then the NE564 phase detector. The inverter transistor is visible too. The mixer is at the bottom of the board. The green thing on the right near the BNC input socket is the two stage line amp. The sensitivity gives the AR5000 a run for its money.

This is the final installation. A miscalculation (alright, I forgot all about it...) meant that the internal loudspeaker had to come out, but I just plug an external one in.