K2VCO 813 Class-C RF Amplifier

This amplifier employs a pair of 813s in class C. That's right, it is suitable for CW ONLY! With a total plate dissipation of 250 watts, it's capable of 800 watts output on 160-15 meters, and somewhat less on 10. It was built for fun, and I used mostly parts on hand or could scrounge. It isn't supposed to be an example of perfection in design or construction (for a more serious effort, see my 4CX1000A project)! Credit goes to Kurt, W6PH for providing the cool chart frame and some other parts.

Circuit Description
The amplifier uses a basic grid-driven circuit, with a link-coupled input tank preceded by a 3 dB attenuator. The output circuit is the usual pi-network. Initially, I used an unusual circuit design in which the high voltage is fed through a coaxial tank coil (see pictures). This places the RF choke at a low-impedance point in the circuit. This was ultimately replaced with a conventional parallel feed circuit because of problems setting up the tank circuit, although it turned out that the coaxial coil was not the cause of the difficulty.

Plate voltage is about 2250 under load, with screen voltage of 400 volts provided by a separate supply. Fixed bias at slightly less than cutoff is provided, along with a grid resistor to bring it up to class-C level when drive is applied. This arrangement preserves the waveshape of the drive so that it does not generate key clicks (although it does sharpen the keying a small anount). T/R switching is accomplished by a fast reed relay in the input along with a vacuum relay in the output. A similar reed relay used to switch the bias with keying. Although the relays switch very rapidly (around 4 ms), there is no sequencing employed. This means that QSK operation requires a transceiver like the Elecraft K3 which provides a reasonable delay between its key-out signal and the start of RF, and after the end of RF before key-out ends. But with a K3, it works great!

The pictures below were mostly taken during the early stages of construction, so there have been numerous changes.

This is the clever coaxial tank circuit. The coil is used for 20-10 meters, and the DC connection for the HV passes through it, exiting at the low impedance point. The large RF choke shown has an inductance of several mH, so it provides a high reactance on 160 meters without worries about resonances on the bands above 40 meters. Ultimately I replaced this with a conventional parallel feed, but I thought it was cool, so here it is.
Here is the under-chassis view. The HV power supply and input circuit are at the bottom. The compression trimmer just above the partition is half of the bridge neutralization circuit; there is a fixed plate above the chassis near the tube plates. Neutralization is done by adjusting this trimmer. Note the copper sheet on the tube sockets. The idea is to provide a good common ground for the various bypasss capacitors. A little 'tongue' of copper passes through the socket hole and is bolted to the top of the chassis, to provide the shortest possible return path from the plate circuit to the filament and screem bypass capacitors. Does it help stabilize the amplifier? Who knows? The transformer to the left of the sockets is for the screen and bias supplies; just above it are the power and soft-start relays. T/R switching circuitry is to the right of the relays.
The amplifier seen from the back. The little boxes at the back contain the T/R relays (see next picture). My theory was that it is good to isolate them in a high-gain amplifier, such as this. Note the feedthrough capacitors on the boxes. Latter I added a box containing a line filter where the power cord is. You can see the completed tank circuit. The big coil is for 160-40 meters, and the tubing coil is for 20-10. This was entirely redone when I changed to a conventional paallel feed. The vertical spiral between the tubes and the tank circuit is to provide a small amount of inductance between the tubes and the pi-network, making it an "L-pi." This made it possible to obtain a reasonable Q on 10 and 15 meters, for good efficiency. This picture was taken before the enclosure was built. The enclosure has a muffin fan located above the tubes.
Here are the contents of the little boxes.
The link-coupled input coils and the grid bandswitch. I later replaced the 10-meter coil with an air-wound version, for slightly more efficiency.
The HV power supply board. Only a total of about 19 uF, but I don't hear any hum. And even if I did, this is a 1955 design! These capacitors are somewhat old, but I used my ancient Eico resistance/capacitance tester to make sure they were formed up and held full voltage
A bad 813 I bought on EBay. Note the broken filament coil. The guy said that the tube 'looked unused' and the damage must have happened during shipping. The carbon plate was very badly pitted, too. I got my money back, but was out the shipping cost. Caveat emptor on EBay!
I had intended to document the construction as I went along, but I didn't do a good job of it. Here are some of the parts I started with -- you can see that I made some changes, like the vacuum capacitor instead of the bread slicer shown.
Here I am before building the enclosure or mounting Kurt's chart frame. Boy, I am happy it's finished.

Back to K2VCO main page