This article is a companion article to the overview and tuning instructions found in the December 2004 QST article about this amplifier. It provides construction notes, guidelines and detailed photographs for the builder. The complete schematic for the amplifier is downloadable as a PDF document. Additional post-publication notes may also be available on the author’s web site at http://www.w6pql.com/.

I mounted the high-voltage parts above the high-voltage transformer on a piece of Plexiglas®, well away from other components, then drilled air-intake holes in the side panel next to the transformer to keep the power-supply components cool. The exhaust fan mounts on the perforated board that houses the low voltage components and Zener diodes that clamp the screen voltage. It exhausts through a screened vent in the top cover just above the fan (not shown). See Photo A. All wire (except for high voltage) is 600 V, #16 AWG wire from the local hardware store. The HV wire has a 5 kV DVM rating and is placed inside 0.187-inch ID vinyl tubing for extra insulation. 600 V wire would also have been safe enclosed inside this tubing.

I used three different military-style connectors for the main, control, and high-voltage interconnect cables; 11, 5 and 7 pins, respectively, and used only the very center pin on the 7-pin connector to carry the plate supply, providing extra isolation from other connections. You could use Molex connectors or Jones plugs here just as well, provided you have sufficient isolation for the HV connection. Figure 9 of the QST article shows these connectors on the back of the amplifier cabinet.

Above the bias/regulator circuit, note the multifunction meter switch and the methods used to meter the screen, grid and plate current, as well as plate voltage. I used an old round meter of WWII vintage that I've had in my junk box for years. It is a 1-ma full-scale meter, with the face marked 0 to 5 kV. You'll notice also that the meter switch has an off position. On this particular wafer switch, there wasn't a lot of physical spacing between positions, so the "off" position isolates positions on the switch that are roughly 400 V different in potential from one another. Had I allowed those positions to be adjacent, an arc could have damaged the circuits. If you have a better-insulated switch, you can eliminate this position.

See the Relay Sequencer schematic diagram with the following discussion. A simple analog switch/timer drives separate op-amp comparators, each switching at different levels on the charge/discharge cycle of the timing capacitor. The comparators drive the relay-driver transistors. If you want to include a preamp in the switching sequence, you need only to drive its relay with the appropriate driver circuit, controlled by the bottom comparator. This will cause the preamp to switch out first, and return last in the sequence. The 50-kW delay potentiometer provides some adjustment for the timing.

When I first looked over the cavity, I had no idea how it should be mounted. So, rather than ask the maker and risk looking silly, I just improvised and made a mounting "saddle" that would secure the unit in the proper position on the chassis (see Cavity Mount picture).

I had a piece of aluminum machined for the mount, and secured the cavity to it with a brass strap. You could always use another method. Take care not to obstruct vent holes, feed-through capacitors, or the controls.

To get the water in and out of the RF deck, I used a couple of ¼-inch ID brass fittings from the hardware store, and made a "keeper" out of a piece of 1-inch wide by ¼-inch thick aluminum bar stock (Coolant Connectors).

Drill a couple of holes large enough for the connector shafts to pass through, but small enough to stop the shoulder in the center of the connector. Drill and tap the two mounting holes for 6-32 screws. Then drill the chassis with the appropriate holes to pass the connector shafts and mounting screws. Sandwich the shoulder of the connectors between the keeper and the rear panel. I used short lengths of plastic tubing to connect the water jacket on the amplifier tube to the connectors. (see footnote 1)

One thing worthy of mention is that this brass-to-chassis connection serves another very useful purpose -- safety. Any current leakage from the anode to the water will be shunted to the chassis here, showing up on the screen meter as reverse screen current. It will be most visible as a small meter deflection during standby. As the water picks up contaminants, it will tend to conduct a bit, and the meter will thus indicate to you when it is time to change to fresh water. The water usually lasts me 90 days or more before it picks up enough contamination to need changing.

Since the meter switch is sometimes above chassis potential by as much as 380 V, isolating it with a Plexiglas® mount and an insulated panel shaft is a good idea.

Notice the way the RF connector adaptors mount on the rear panel. I made these adaptors from aluminum bar stock, ¼ × 1 × 1.5 inches, and drilled out just enough to pass a standard female type N cable connector, 11/16 inches, for the ones used here.

I drilled and tapped two holes into the face of each adapter and also drilled and tapped a hole for a no. 6 setscrew into the side of each. Then I attached them to the rear panel with 6-32 screws and used the setscrew in the side to hold each RF connector in place.

A perforated breadboard holds most of the components and control circuits. It’s mounted to the floor of the chassis with a couple of aluminum brackets made from scrap 0.060-inch-thick aluminum strips.

I used RCA connectors for the T/R key lines to and from the amplifier and mounted them just below the fan on the rear panel.

The cooling fan draws air from holes drilled into the floor of the chassis, and expels it out the back. It already had a speed control built into the cable, but still, I wired this fan in series with the one that blows air across the cavity, slowing them both down for quieter operation. You don’t need a great deal of airflow here, just a whisper; the water is doing most of the cooling.

Note the mounting location of the bias adjustment control, just above the coolant connectors. This control is above the chassis by about 400 V; so use an insulated mount like the meter switch. (see footnote 2)

The cavity is mounted with the plate tuner connected to the tuning shaft. The shaft passes through a ¼-inch bushing in the front panel. I used an insulated coupler, but it isn't necessary to insulate here. (I just ran out of regular couplers.) The cathode-tuning shaft does not require a panel control, as it only requires adjustment once during initial tuning, or when the tube is changed.

A second fan is mounted above the cavity, positioned to blow down through the vent holes in the body of the cavity. The mounting bracket was made from 0.060-inch-thick aluminum, bent into a Z, and mounted to the cabinet bottom with the 2 screws shown. Holes for the fan, cathode-tuning shaft and grid feed-through capacitor were cut with chassis punches.

Now that the big parts are all mounted, and while there is still some room to work, assemble the RF cables. Route them and connect them to the various amplifier connection points, following the complete schematic diagram. Avoid using RG-213 here, as it will not hold up very well at these power levels. LMR-400 is a better choice; it can handle the power, is flexible enough, and has half the loss.

Input and Output cables are fitted with a male N connector on one end, and a female on the other. These are routed top and bottom, respectively, to keep cross-coupling to a minimum. The output cable is bent to allow an inch of clearance from the water jacket on the tube (there will be 2 kV there). The cavity-to-relay cables are male-to-male. The bypass jumper is ¼-inch semi-rigid coax with a male connector on each end (LMR400 can't quite make the bend in this space). You can use UT-141 or RG-213 on this piece if you like, as the power level is low and the jumper is very short, so loss is not really an issue. I only used the ¼-inch coax because I had it, not because it was necessary.

I used no. 18-gauge, 600 V wire throughout, with the exception of the keying lines and fans. Number 22 wire is okay there. Like the power supply, the plate supply lead is 5 kV wire run inside vinyl tubing. Once the wires are laid out and connected, they can be loosely bundled with cable ties. This photo shows how the wiring was done from the connection side of the perfboard.

On the component side, resistors that will get warm (like the cathode cutoff and screen safety resistors) should be kept away from other components that may be sensitive to heat.

Place the low-voltage components away from those carrying higher voltages. The sequencer components are on the left, and the rest on the right. Locate the bias regulator transistor in the exhaust-fan air stream, and use computer-style jumpers to adjust the filament-supply resistance for proper emission. The RF deck wiring photo shows the routing of the coolant lines to the tube, and the anode supply wire.

I used a 9-pin Molex connector (only the center pin) as a service disconnect. I put the wire inside vinyl tubing, and routed it to the high-voltage metering resistors towards the top of the amplifier, away from other cabling. There are also disconnects for the other cavity electrical connections. I used three different connector styles to prevent accidental misconnections. These disconnects are convenient when the cavity must be serviced, such as when changing the tube.

There are three other disconnects that are useful in initial testing of the amplifier, located on the rear of the perfboard. For example, the filament can be disconnected in order to test other circuits without risking damage to the tube. The screen and cathode can be disconnected, and the HV line unplugged from the rear chassis connector in order to test the bias circuit. The connection to the plate voltage metering resistors is as far away from other components as possible. The fan speed control, a part of the cable assembly on the exhaust fan, is also visible in the photo.

The rear panel is pretty Spartan. Nobody ever looks back there anyway, but there should still be labels. I took greater care with the front panel. I used inkjet printer labels; it was a bit of a challenge to match the background color on the label to the paint on the chassis, but I was able to come pretty close. Cut them out with a pair of scissors, stick them on, and you’re ready to go.

The GS15B single tube grounded screen 250 W+ PA for 1296 MHz is totally stable up to 300 W output if operated within the parameters listed below (all voltages are referenced to the cathode):

Anode voltage = 1600 to 1800 V dc

Anode current = <350 mA

Screen voltage = 320 to 350 V dc

Screen current = + 1 to –1 mA when tuned correctly.

Grid voltage = –22 to –32 V dc

Grid current = 5 to 30 mA

Idle current = 50 to 75 mA

Heater voltage = 5.5 to 5.8 V ac

Drive power = 10 W

Typical efficiency = 50%

This PA can produce up to 400 W with minor thermal drift in class B or C, if operated within the parameters listed below:

Anode voltage = 2200 to 2300 V dc

Anode current = <325 mA

Screen voltage = 350 V dc

Screen current = + 1 to –1 mA when tuned correctly.

Grid voltage = –30 to –40 V dc

Grid current = 8 to 20 mA

Idle current = 0 mA

Heater voltage = 5.5 to 5.8 V ac

Drive power = 15 W

Typical efficiency = >55%

Since this tube is generating negative screen current, it will self-supply screen voltage. There must be a resistor of less than 100 kW between ground and cathode at all times to bleed off self-generated screen voltage. Never switch the screen voltage on and off without this resistor in place to bleed off the screen voltage.

Disconnecting the screen with anode voltage present and no load resistance will cause a failure. The screen voltage will increase to the anode voltage potential and cause an arc-over in the grid decoupling insulator or an internal arc in the tube.

The screen power supply must be of shunt type to prevent a screen voltage increase when the amplifier is tuned in correctly, or you could get an anode current run-away.

I also recommend that you use a sensing circuit that will trip and shut off the amplifier if the screen current goes more than 5 to 10 mA negative.

Watch the screen current carefully while tuning the PA. The screen current will go drastically negative (–15 mA) if the cavity is tuned too low in frequency.

Apply 5 to 10 W of drive and tune the output for maximum power by alternately adjusting the anode tuner and rotating the output coupling loop. Tune the input cavity for best SWR by alternately rotating the tuner and adjusting the input coupler in and out.

Re-adjust the output for maximum power. Watch the screen current and adjust the anode tuner slowly CW until you see a decrease of about 1 mA (negative current). You should now be about 10 W below maximum power. This will ensure that the amp will reach full power immediately after an idle period.

Use a distilled-water cooling system with good flow rate. It should be able to handle greater than 300 W load without over-heating.

The recommendation is to use a small, quiet muffin fan to blow on the input cavity. The amp is generating very little heat so a low-capacity fan is sufficient.

A heater voltage that is too high will drastically shorten the tube life.

This minimizes any cathode back-heating. Tests have shown that many tubes have full emission down to 5.4 V. Higher heater voltage than necessary will only shorten the tube life with no power gain.

Leave the heater on for at least 2 minutes before transmitting to allow the cathode surface to reach uniform temperature.

The cavity top has to be removed to replace the tube.

All decoupling is done inside the amplifier, and no external RF chokes are needed.

The cavity’s RF leaks are almost zero so no extra shielding is needed for this reason.

You are working with lethal voltages, so install in a cabinet or other place where the high voltage cannot be reached.

Footnotes:

1. During the first year of operation, I noticed that electrolysis (due to leakage current) was leaching some of the metal from the brass connectors. This would show up as a greenish-blue tint in the water. I was able to minimize this by lengthening the plastic tubing from the original length of around 12 inches to about 24 inches. This reduced the leakage current and the resulting electrolysis. Over the next 6 months of operation, I observed no evidence of electrolysis acting on the connectors.
2. My control held up as shown (without an insulated mount), but I later changed to an insulated one just to be safe.