~ Updated 10/29/2024:  Under Construction…some thoughts about Linear Amplifiers ~

Many articles about RF amplifiers have been published and it’s impossible to list them all here, so I’m going to pick and choose some I like.   Mostly, I gravitate to what can be built from the junkbox or scrapyard and I tend to concentrate on that.  In other words, I’m cheap.

I also gravitate to simple designs, so this article concentrates on Grounded Grid (GG) amplifiers, due to their simplicity, ease of construction and because most of us have an 80 – 100 watt transceiver with which to drive a GG unit.  Properly driven, a GG linear amp amplifies the input power by a factor of  about 10, which means your signal is theoretically twice as loud (3 db).

If you are a QRP operator and have only a QRP unit to use as a driver, you probably don’t want a big linear amp anyway, but if you do, the old 813 can be driven with only about 5 watts – but not in GG.  Search the web for inspiration – there’s lots of it.

It does not matter how the watt is radiated; from a $25 813 tube or a $2,000 solid-state class D amp – the watts sound the same on the receiving end.  Amplifiers aren’t so much the great equalizers – antennas and operating skill are.

My favorite tube:  The venerable 813 tube was first developed in 1936 and was so robust and useful that it is still in production, 87 years later, in China and Russia.   They are inexpensive; I have two NOS American ones with graphite plates bought for $25 each and one NOS with a steel plate given to me.  Compare those prices to 572’s or other tubes!  Use the graphite-plate ones rather than the cheaper steel plate tube.  The graphite ones are more robust.

My first home-built KW linear used a 4-1000A tube, with parts salvaged from a discarded 10 KW AM aircraft beacon transmitter.  I used that for about 28 years until the power transformer died.

Next, I constructed a 2 x 813, 1 KW GG linear amp,  but it contained power supply and filament xfmrs and the RF deck in one box and it was too heavy for me to easily move it around (I’m 83).  So I’ve taken it apart and am rebuilding it into two separate ventilated HeathKit cabinets (P.S. in one and RF deck in the other).  I’ll put some photos of that here “soon.”

Since I recently had a cardiac pacemaker installed, and since it wants a limit to RF exposure in the HF region, I’m opting for a single 813, 500-watt amp this time.

One 813 can easily provide 500 watts input and grounded-grid circuits provide a simple, stable design.  Here’s a tube data sheet, courtesy of Dr. Greg Latta, AA8V.

813’s are rated for a 2500 volt plate supply but units have been run for years with 3000 volts or more.  There is no discernable difference on the receiving end as the result of using more voltage so just go with 2500 – 2800 volts and the tubes will last 20 or 30 years.

A Grounded Grid linear amplifier means ground the grids!  Use the shortest, widest leads possible to ground them.  Don’t strap them to ground through an 0.01 capacitor.  W8JI presents excellent information about GG amplifiers and it’s well worth studying.

Here’s an example of what is probably the simplest dual-tube, 1000 watt design, using the 813 tube as an example.  If you want only 500 watts input, just use a single tube.  The “Z” in the signal input line is just the usual resistor/coil arrangement commonly found in the plate lead.  If the unit is built properly, a single Z in the input will suppress transients, but feel free to put one in each plate lead if you prefer.  813’s are quite stable in GG circuits.

Be mindful that the glass envelope of the tube will become HOT so provide adequate ventilation.  I’ve measured temperatures of 290 degrees F on certain parts of the glass envelope.  I’m using a small fan and a fancy home-made, heat-dissipating plate cap on my tube which is mounted horizontally (with pins 1 & 6 in the vertical plane).

Here’s an example of a power supply to use with the above 2 x 813, 1 x 813 or the Lew McCoy design (below).  It uses two discarded microwave oven transformers (both of approximately equal size & ohms in the secondaries) to provide HV.

As designed, there is no AC voltage on the transformer frames.  A revised circuit, showing how to use half the B+ voltage for tuneup purposes, will be added soon.

For the 500 watt version, one microwave oven transformer would certainly be enough if it is a robust one but remember that, in this case, the frame of the transformer will have two thousand or more volts on it so be sure to protect it from human contact and insulate it from the chassis!  In this case, it might be prudent to incorporate a relay to kill the power to the transformer if the lid of the case is opened.  Remember, these high voltage supplies can kill you.  Safety first; belt and suspenders.

Be sure to install a HV microwave oven fuse immediately after the filter caps.  These are available on Amazon.com in various milliampere ratings; a single 813 will draw no more than 250 ma so a 500-750 ma fuse would do.  They are only about $7 for a 4-pack and well worth it.  Never use an ordinary 120/250 volt AC fuse in a high voltage circuit!

Following the fuse, install a 50 ohm “glitch” resistor of as high a wattage as you can find (50 watts being ideal).  If there is a major flash-over or instantaneous HV fault, the glitch resistor will absorb most of the surge and the fuse will likely survive.  It the fuse does open, it will open in a millisecond or less, protecting your expensive power supply parts.  Both the fuse and the glitch resistor are cheap insurance.

Put a thermistor or a soft-start device in the HV transformer’s primary winding circuit or the power supply might blow a house breaker at turn-on, if it turns on at exactly the wrong point in the 120 VAC phase (if you are using a microwave oven transformer).  I’m using a 20 amp thermistor, which drops the line voltage by about 1 volt once it settles down after turn-on.

The 813 requires 10 volts at 5 amps (50 watts) for its filament and a way to provide this must be found.  A 10 volt transformer can be purchased or a microwave oven transformer can be rewound.  Put a thermistor in its primary, to ease the start-up surge to the filament and to prevent popping the house breaker.

My filament transformer is an old one, with multiple primary input taps.  The line voltage here is 120-121 volts so I’m using the “123 volt” primary tap on the transformer.

The transformer from a discarded UPS unit could be used (run backwards) with a few secondary turns removed or with a suitable dropping resistor or a ‘bucking’ transformer in the primary as these would otherwise provide about 12 volts or a little more, under load.

Each 813 requires 50 watts for the filament (10 volts X 5 amps), so size the transformer properly.

The Lew McCoy single-813 design can be found here.   Note that you can ground ALL the grids directly which is a better design.
You can also skip the fancy metering scheme in this design and replace it with a simpler one:  For plate volts, measure across the final resistor in the equalizing string as he has done.   For plate MA, put the meter and its shunt in the center tap of the filament transformer.  This is the most simple approach but be aware that the milliampers shown on the meter will include the leakage through the capacitor equalizing resistors and the bleeder resistors.  You can mentally adjust for that.

Here’s another, nice clean design.

Here is a link to an alternative design that eliminates the need for a filament choke and provides some input tuning capability (see below for more about filament chokes).

Here’s an excellent and extensive article about pi networks for RF amplifiers.  Even though written in 1972, the simple math and the tables provided in this Sept 1972 Ham Radio issue are as valid today as they were then.

If pi network space is critical, look at this very detailed article about using a T200-2 iron powder core toroid for the pi network coil.

Filament Chokes:  The old B&W FC30 or FC30a filament chokes are convenient, if an affordable one can be found.  The Heath SB-200 used a 10 uh choke wound on a ferrite rod but this ran out of steam on 80 meters.  More uh is better; say 12 -14 uh, or even 20.

A classic homebrew solution is to wind a bifilar choke on a ferrite antenna rod.  A good one would be 2 x 26 turns on a 9.5mm diameter rod, with the winding using about 100 mm (4 inches) of the rod.  There are many Google references to this.

Another alternative is to use an FT-240-61 ferrite toroid.  Use the ferrite core calculator link below to calculate the number of bifilar turns needed (hint:  10 or 12 would do; 14 is even better).

Calculator to find the inductance or calculate the number of required turns on a ferrite core.

Calculator to find the inductance or calculate the number of required turns on an Iron Powder toroid.  For example, 35 turns on a T-200-2 would provide very adequate RF choking.  Use two of course, one for each side of the filament lead.