G4DCV's 144MHz amplifier with a pair of GI-7B Russian triodes.

The design objectives for this amplifier were:

  • UK legal power output (400 watts) at true 100% duty cycle for WSJT modes without exceeding the tube ratings.
  • 25 watts drive for 400 watts output (Gain 12dB)
  • Triodes to simplify power supply
  • Reliable in contest operation
  • Easy to source components
  • Grounded plate line for safety
  • Power supply and RF deck in one box for safety
  • Cheap to build

If you do not understand high-voltage safety procedures do not attempt to build a tube amplifier.

The HT voltage can kill.

When I went QRT on 2m because of a house move in 1986 the only tube that was available at reasonable cost to amplifier builders in the UK was the 4CX250B. I have built several homemade amplifiers with these tubes and used a pair regularly on 2m in the 1980s to work 67 DXCC. Eimac triodes like the 8874, 8875 or 8877 were available but were very expensive.

Returning to 2m in 2004 things had changed. The Cold War was gone and Russian triodes and indeed submarines, tanks and fighter aircraft all were now available on ebay at bargain prices.

I decided to build my desktop amplifier using a pair of GI7B triodes. Reg G3UTS had bought four and let me have two of them. I was fascinated with the idea of building an amplifier capable of the maximum power output allowed under the UK licence, 400 watts, with a pair of tubes that cost under £15.00 (US$30) each new in the UK.

Initially I built the amplifier using a half-wave stripline and readily available, and free, microwave oven transformers. The HT voltage from the transformers was too high and so I bought a torodial HT transformer from Linear Amp UK. This was the most expensive component in the amplifier. Despite a lot of work I was unable to resonate the plate line. Plus the line did not meet one of my design objectives of being solidly grounded so that if the anode coupling capacitors failed the HT would be shorted out. So I abandoned the half wave line and used a quarter wave instead.


I wanted an amplifier which I could build as cheaply as possible, was reasonably compact and would sit on the table alongside my IC-746. Drive power was not a consideration as the IC-746 has variable power output up to 100 watts. Since coming back on the air in 2004 I had discovered the fun of WSJT and was using FSK441 for meteor scatter and even JT65b for occasional moonbounce. So it was important that the amplifier would comfortably run 400 watts output (the UK maximum) when using a high-duty cycle mode like WSJT.
G4DCV's shack. On the shelves from top to bottom, fully loaded FT736R, old FT221R, this amplifier and AEA LA-30E 3-500Z HF linear. On the desk Icom IC-746 and computer on ON4KST chat.
When I came to test the amplifier I was unable to get the half-wave anode line to work. The best I managed was 20 watts output for 600 watts DC input! So I rebuilt the amplifier using a conventional quarter wave line with a coupling loop. Plate tuning is with a flapper connected through a flexible strap. The cathode compartment. The input circuit comes from YU1AW's web site and works very well. Input VSWR is 1:1. The white pipes direct air from the anode compartment over the cathodes, an idea suggested by Ian GM3SEK. Also in the compartment is the heater transformer and TX/RX bias relay.
I wanted a single box amplifier and don't like the idea of a seperate power supply with an external HT cable. The rectifier board can be seen mounted vertically and behind the toroid is the step start relay. Not visible but bolted to the back wall are two series 22R, 200 watt 6kV anti-surge resistors in the HT line. Underneath the power supply compartment. The PCB in the centre is the bias board circuit copied from GM3SEK's Triode Board manual. To the left of the bias PCB are the metering resistors and to the right the relay supply transformer. The board to the left of the HT toroid is the bottom of the step start anti thump relay board. The case is made from 12mm angle aluminium with panels riveted or screwed to it.
A general view of the amp with the top covers off. Metering from left to right is HT Volts, grid current and plate current. I need to put some knobs on the plate tune and load and the grid tuning controls. Under the meters are LEDs to indicate mains on, HT on, TX and RX. The backside. The RF deck and PSU have seperate blowers. Overkill maybe but I want the amp to run cool on WSJT. The coax relay is an old Londex available very cheaply at Hamfests in the UK. It's 75 ohm but works fine on 2m. I have modified it to N sockets.
Dimensions of the anode bracket. The shape ensures that there is good airflow through the cooler fins. The shorting bracket is positioned 80mm from the tube end of the line and has slots for the screw holes so that the line can be brought to resonance with the flapper capacitor almost closed at 144MHz. A slow-mo drive is fitted on the front panel. If you copy this amplifier you must expect to have to adjust the line length on-test. The resonance peak is very sharp. If you cannot get power out the problem is almost certainly the anode line dimensions. There is a simple way to adjust line resonance and matching with power off. See my GS31 amplifier
The tuning flapper dimensions. The L shape is because I had to fit it around the hole I had already cut out for the cathode air-flow from the Mark 1 version when I unsucessfully tried a half-wave line. This was a kludge flapper that I threw together to try to get the 1/4 wave line working. It is still in use 3 years later. The flapper is grounded using the braid from RG-213. An even lower inductance connection would be better here. Thanks to Viktor HA5LV for suggesting a length of 170mm for the output loop which is positioned about 3mm from the line. The quarter wave anode line is 50mm above the chassis.

Dummy load testing. The amplifier is running at 1200 watts input with 600 watts output so efficiency is 50%. The cooling is good enough that the amplifier can be run at 100% duty cycle at 600 watts output without exceeding the tube ratings. At lower power levels the efficiency is better. At 400 watts output it is 56%. For 400 watts output the amplifier needs 25 watts of drive which means that my FT-736 can also drive it to UK legal output without an intermediate amplifier. In retrospect my only regret, apart from wishing I had finished the project earlier, is that the HT volts are rather low, about 1550 on-load. I'm now building a new amplifier with a GS31 and when it is finished I will probably convert this pair of GI7Bs to 6m.

Please note that I originally built this amplifier "in my head" and didn't have any drawings of it. I drew the following from memory several months after I finished the amplifier so there will be errors! Please do not blindly copy these circuit diagrams- if you need to do this and don't understand howe the circuit works you really should not be building a high power amplifier with lethal voltages.

I put this information on the web for general knowledge of radio amateurs. This is not a construction article and if you try to blindly copy this it surely won't work.

If you do not understand high-voltage safety procedures do not attempt to build a tube amplifier.

The HT voltage can kill.

Schematic1 - Schematic 2 - Schematic 3 - Schematic 4

GI7B Amplifier Notes

Last updated 8 August 2011

The 2m amplifier with the pair of GI7Bs is now 6 years old and has been on loan to Kevin G1KAW for the past couple of years since I built my GS31 amplifier. Kevin has regularly used it to acivate JO00 square mainly on MS but also with some tropo and EME. In his own words the amp has worked a lot of very nice DX!

He reported to me that one of the tubes was running a lot hotter than the other and so I have recently had the amp back in and given it a full service including fitting new tubes.

Since writing the web article on this amp I have made a number of modifications, hopefully improvements, to the amplifier.

  • Replaced the input coax relay. The original small BNC relay fell apart in a contest in 2005 and a "temporary" replacement was hung on the back panel with a cable tie. Finally fixed to back panel properly August 2011!
  • Cleaned out 6 years of dust!
  • Fitted M4 Nutserts on all top and bottom covers (about 60 of them!).. Threads in aluminium angle don't work very well and so when a cheap Silverline Nutsert tool (GBP 10.00) came up on ebay I bought one.
  • Oiled the main blower motor bearings. It is now much quieter!
  • Replaced the thick film HT anti-surge resistors with a tubular ceramic wirewound (43R, 25W from ebay for a few GBP). Thick films are no good in this application and fail open circuit (or high resistance) under flash over surges.
  • Fitted a grid over-current protection circuit. See below. I have learned that the surest way to kill a triode is to over drive it! When I used a FT736R as my main 2m rig this was not a problem, the 736 only outputs 25W. But my IC-746 runs up to 100W output on 2m and one day I forgot to turn the power control down. The result was a flash over, a dead pair of tubes and an open circuit thick-film anti-surge resistor. The amp was off the air for a while while I repaired it. An overdrive protection circuit is very simple and would probably have saved all the work and the cost of replacing the GI7Bs.

The grid trip circuit.

grid trip circuit















grid trip photo



I used a latching relay which was in stock at the local Maplin electronic shop. The input to the circuit sits across the grid meter shunt resistor at the junction of the collector of Q1 the TIP147 bias transistor. For a pair of GI7Bs the 10k preset pot is adjusted so that the latching relay pulls in at 200mA of grid current. The relay open circuits the PTT line and lights a warning LED on the front panel. Pressing the Reset button enables the PTT again provided the fault has been removed. The LED across the SET coil is optional but warns when the trip is about to operate and will flash with increasing grid current. It is very useful when adjusting the trip point.

Operating from the back of a van as M8C in the March 2005 144MHz contest.

I have also been using the amplifier from home for WSJT meteor scatter and EME and the amplifier has been used for over 40 EME QSOs with a very small antenna. These are of course high-duty cycle modes. The only problem I have had was when I forgot to turn the drive down and put 100 watts from my IC-746 into the amplifier. One of the GI-7Bs flashed over and after I replaced the HT transformer fuse I found it had died. I didn't have a spare at the time and so continued to run the amplifier on the one good tube, leaving the dead one in place. Under these conditions I could get up to 350 watts output. Over a period of about 10 days the dead tube healed itself. Slowly it came back to life and now the amplifier is back to normal. So don't throw those dead GI-7Bs away too fast. You may be able to heal them.

The total cost for the amplifier was about £200, (US$400), more than half of this was the cost of the transformer.

73 Paul G4DCV

Email: paulATg4dcv.co.uk Replace AT with @



I used this amplifier for about 5 years before building one with a GS31. Kevin G1KAW now has it on loan and uses it regularly from JO00 square. It has done several 24 hour portable contests and made 1000s of QSOs

We used it as M8C in the September 2005 2m Trophy and won the UK Open Section. We used it again in the November 2005 CW contest and won the Multi-op section. And it was again used for M8C in the March and September 2006 contests and then for M0MAD during March and September 2007. On-air signal reports have been very good and many stations have been very complimentary about the quality of our signal.