Post
by Albert H » Sun Apr 21, 2019 6:15 pm
Back in the 80s and 90s, there were quite a few PCB exciter designs around. Roger H (now at Broadcast Warehouse) had a neat little PCB that gave about ½ Watt and was quite stable and clean. He later tacked a PLL PCB beneath them to ensure their frequency stability.
My PCB design in the 80s was a double-sided board that could be used on any band from Band I to Band V just by changing the coils and a few capacitors. These were designed to have a CMOS PLL, and the VCO would work between 40 and 70 MHz. The Band I version was "at frequency", the Band II version used a doubler stage, the Band III version used two cascaded doublers, and the Band IV and V versions used Varactor multipliers and grounded base driver and output stages. All of them delivered between 750mW and 1.2W,and all used the same PLL with diode programming to set the frequency. They were designed to fit into a standard tinplate box.
Dawson had a PCB that Kenny Myers designed. This was a doubler design, and used a four-chip CMOS PLL. If built correctly, with the right components, it was actually pretty good for its day.
There were many others around. Bonex sold a kit for a "1 Watt" VFO exciter that used a neat row of Toko S18 coils, but wasn't particularly frequency stable, and could be tuned up to deliver lots of different frequencies at the same time!
There was a big, square PCB that the Invicta rigs used. It had the most complicated coil assembly that I ever saw in a rig - several coils on the same former, with some of them overwound with secondaries. This board went beneath the chassis of their standard rigs, and the top of the chassis had the mains transformer, smoothing capacitors and the amplifier stage valves. Their final was usually a QQE06/40 which could be critically tuned to give as much as 100W when the anode voltage was cranked up a bit. Their rigs always drifted because of thermal problems, but Bob Tomalski (RIP) and the rest of them couldn't accept that their design wasn't ideal!
These days, surface-mounted components seem to be the way to go - particularly at higher frequencies. "Leadless" components get rid of the problems of lead length, but introduce problems of their own.
A recent PCB design I saw (from a well-known rig supplier) had a problem with going out of lock after it had run perfectly for an hour or two. I put one of these on the bench and tested it, and discovered that the VCO Control (Tuning) Voltage would slowly drift downwards as the PCB warmed up. The final stage was a pair of BFG135s running in parallel to give about 1.5 Watts from the board - ideal drive for their following amplifier stages. Unfortunately, the output stage heat would warm the oscillator enough through the groundplane so that the Control Voltage would head towards ground, and when the loop filter op-amp couldn't get any lower in output voltage, the loop would go out of lock, shutting down the rig. It would cool off, and re-start, and then go off again!
Silly design errors like that have always helped me in my design work - I learn from them, and don't make the same errors myself!
"Why is my rig humming?"
"Because it doesn't know the words!" 
