High Power RF "PIN switch" for VHF



For switching the high power levels of our homebrew amplifiers (see the GS35-PA coming soon, the Semelab PA and Dirk's GI-7B amp project) we decided to build a solid-state rf switch instead of using coaxial relays which would require the usage of sequencers and make this annoying "klick-klack" sound.

One great advantage of this switch is that it's installed in an extra box outside the power amp, so you can mount it directly at the antenna.
You will need one big coax to bring your power rf up to the switch and you can use a rather lossy coax and a preamp at the RX output of the Switch which goes back to the TRX.
If the switch was installed in the amplifier, you would only need one big cable to the antenna but if you would like to use a preamp for RX, you would have to get two high power coax relais to switch between TX and RX right at the antenna. With this solution you just have to unplug the RX coax and plug in a preamp! Much easier, isn't it?

The switching is done by UM9415 PIN diodes with a quarter wave transmission line.

Here a picture of the pcb containing the directional coupler for power and SWR measurements and the PIN diode switch:

It's made of PTFE material to ensure low loss; after etching the pcb I've silver coated it (so at least it looks nice ;-)

Here the same board with the switching components:

As you can see there are many ground connections made by silver wire through the PCB and by copper foil soldered across the borders of the board.
The quarter wave transmission line is installed on the bottom of the pcb and made of UT141 semi-rigid cable:

The transmission line was calculated to be 36,4 centimetres long :
Lambda/4 = c / (f * 4)= 300 / 144,3MHz = 51,98cm
Due to the propagation factor in this cable (v = 0,7) the physical length is 51,98cm * 0,7 = 36,38cm.
But the first measurement showed that this calculated length transformed at the wrong frequency, as you can see in the following diagram (center is 144,3MHz = design frequency):

Insertion loss in transmit mode = 0,35 dB @ 144,3MHz

The peak (lowest insertion loss) was at 111 MHz instead of 144,3 MHz, so the quarter wave transmission line was too long about factor 144,3 / 111 = 1,3
so the short of the pin diode wasn't transformed to an open at 144 MHz.
So I've cut the transmission line; now it's only 27,9cm long (36,38cm / 1,3).
The new measurement proved this calculation:

TX insertion loss after optimizing the quarter wave transmission line
for minimum loss at 144,3 MHz

Now an insertion loss of 0,1 dB could be achieved at the design frequency! A really great rf switch !
And here the same measurement for RX operation:

Insertion loss in RX mode = 0,2 dB @ 144,3 MHz

The only "minus" is that the isolation is only about 22 .. 25 dB but that's why we've added the Omron Relais in the input switch (~35 dB isolation) and the additional "pull-to-ground-relais" at the input side of the RX path.

(c) DJ9KW, 25. 05. 2008