I have chosen to work with the 813 tube on my build due to it's reputation of stability against self oscillation. I know that there are other tube choices that offer higher performance but the old 813's are still available at reasonable prices and I have several gathered up.
I don't know for a fact that the 813 is any more stable than other tubes. The main reason to use one is that you have some - and that's a very good reason.
This is the design that I am going to try and replicate and I have all of the parts now to construct this amp. I have read that there is some concern as to the safety of this particular design. Particularly around area in which the B- is metered. Would anyone please describe exactly what they would do to make this design safer?
The meter reads the plate current. It is placed in the B- lead so that the whole meter movement is near ground potential. It's a very old trick and there's nothing wrong with it. It is much safer than putting the meter in the B+ lead, which is the alternative.
The only hazard is "what happens if the meter opens up?"
The solution is to put a "safety resistor" from B- to ground right inside the power supply. The safety resistor is chosen to have a value that is 20 or more times the meter resistance, so that practically all the current flows through the meter.
If you have a 0-500 mA meter, as called for in the schematic, its internal resistance is probably less than 1 ohm. (Measure to be sure; rig up a battery and resistor to drive the meter just full-scale, then measure the voltage across the meter. A little Ohm's Law and you know the meter resistance).
The meter will read slightly low, but not enough to matter if you choose the safety resistor properly. Say you put a 100 ohm safety resistor from B- to ground in the power supply - less than 1% of the plate current will flow through it during normal operation. If the meter opens up, and the plate current is 500 mA, there will be no more than 50 volts across the safety resistor.
Note that the safety resistor has to have fairly high power rating to insure that it won't burn out if a meter failure occurs. The minimum wattage I would use is 50 watts (500 mA times 50 volts is 25 watts, so with 50 there's a 2x safety factor). Normally the resistor will only dissipate a fraction of a watt; the high rating is for the what-if scenario.
One way to get the high wattage is to parallel several resistors. For example, five 10 watt 500 ohm resistors would do the job.
I have a few comments on that amp design:
1) The 813 is being run very hard; 3000 volts at 350 mA is well above the normal max ratings of the tube in both voltage and current (download the specs and see). For unprocessed SSB, this may be OK, but for CW you may not want to smack it so hard and for high-duty-cycle modes like PSK31 you REALLY want to run it at lower power.
How much risk you want to take is up to you; they're your tubes. Just be aware that the design really pushes the tube way beyond ratings. (A single 3-500Z would be within ratings at 3000 volts 350 mA).
2) If the 813 is mounted horizontally, it is very important that the socket be aligned so that the filament is in the correct position. The tube data sheets will tell you the correct alignment (usually there is verbiage about Pin X and Pin Y being in the vertical plane). Be SURE you know it's right! If the alignment is wrong, the filament can sag and touch the grid, and then all sorts of very bad things start to happen. Expensive fireworks!
3) The amp (as designed) does not go below 7 MHz. Whether or not that's a concern is up to you.
4) There is no metering of the grid/screen current.
5) The input circuit is not unique; there were grounded-grid amps built using the coaxial-cathode-coil method more than 40 years ago. That method was never very popular because it adds a tuning control, but it is just as valid as the more-common bifilar-filament-choke method. Do not expect 1:1 input SWR on all bands with it as designed.
6) The plate RFC should be a 500 mA unit.
7) ZL1AXB made his very compact; you don't have to. But be careful about layout; you don't want long RF leads.
8) Cooling is key. The filament of the 813 dissipates 50 watts all the time, and when you're transmitting the tube is dissipating hundreds of watts! You want good air flow in and out. You also want RF screening; the fan over the open hole may not be enough.
You may wish to use more than one fan. Old computer power supplies usually have 12 volt muffin fans; often old AT supplies can be had for the asking and torn down for the fan and some small parts.
9) The TR circuitry is show with a relay contact selecting either the Zener diode or a high-value resistor. Note that during the transfer time of the relay the cathode circuit is opened. I would put the Zener diode and the high-value resistor in series, and use the relay contact to short out the resistor on transmit. That way the circuit is never open. Less wiring, too.
73 es GL de Jim, N2EY