Disclaimer: No warrantee provided, your mileage may vary
Unless you are using a 28 volt military vehicle electrical system, you'll have some kind of external power supply for the R-392. The nominal supply is 28 volts at 3 amps. Of this, 2.5 amps is for tube filaments, 0.5 amp for plates. While the manual reports full sensitivity at supply voltages of 25 and above, many people report improved RF performance by running the plate supply at 28 to 35 volts while keeping the filaments at 28 volts. It has been my experience that the receiver, when properly aligned, meets its sensitivity specifications at 26 to 28 volts on a combined filament/plate supply.
Al Marshall reports that it is possible with some minor modifications to run the filament string on 28 VAC, reducing the regulated 28 VDC power requirement to 500 ma. The mod consists of lifting one end of the antenna relay coil. DO NOT USE THIS MOD IF YOU HAVE THE SOLID STATE AUDIO MODULE -- It runs on the filament, not plate circuit.
There are some military and commercial power supplies, with beefy outputs suitable for running a room full of R-392s (and prices to match), that are available from Fair Radio and other sources. Examples include Lambda (28V, 19A, 62 lbs, $195), PP-2309C (0-36V, 15A, 75lbs, $145), and PP-4763/GRC (28V, 50A, 145lbs, $195). A variety of less expensive alternatives include the PP-3700/PRC-41 (24-28V, 4.5A, 15 lbs, $35-$50), copier switching and linear power supplies in the 25-30V range, as well as homebrew designs.
Some problems with sensitivity (and unresponsive S-meter) can be traced to weak tubes, leaky caps, or out-of-spec resistors, particulary those in the S-meter bridge circuit (V502, R509, R622, R621, R512, R624, C714, C624). A common problem is a leaky C509 in the cathode circuit of V502, the 2nd IF. Tubes in the RF/IF chain should be checked by substitution; tubes that test OK on a tube tester can still have dramatically different in-circuit performance. Place "hotter" 26A6 tubes in the 1st & 2nd RF Amp, and first few IF stages. The 26FZ6 can be substituted for the 26A6 in the RF/IF amps for an increase in sensitivity. (Paul Bigelow and George Rybicki)
While I don't experience problems with intermod, some folks do. Step one in reducing these problems is keeping VHF/UHF energy out of the front end with a regular amateur or CB low pass filter. These filters pass RF energy below 32-35 MHz to the receiver and keep all the FM and TV stuff out. Proper alignment and putting "select" 26A6s as the RF amps is another step. And double check your AGC chain. The first RF amplifier has its own AGC circuit; the other AGC circuit controls the 2nd RF, following IF stages and metering circuit.
Always check suspected intermod with another receiver. It is possible for the intermod to occur OUTSIDE the receiver -- bad electrical connections in antennas, feedlines, power and telephone wiring, even fences and aluminum gutters.
Jerry Johnson, K0CQ, recommends 200V orange drops as replacements for the oiled paper capacitors, and dipped silver micas for the micas. Orange drops should not be used to replace micas, as they have higher inductance and less stability than micas. Resistors should be replaced with those of equal or larger power rating as the originals (1/4 to 2 watts). Also look at Mike Dinolfo's recommended cap/resistor replacement parts
Gord Hayward, VE3EOS, reports the following fix for IF output saturation/overload.
The IF out from my R-392 (3075-PH-51 #498) saturated when I tuned a strong or even a moderate station. I did the calculation for the cathode follower and found that it saturated where the theory said it would. Has anyone else found this?
The fix was to replace the follower stage with a unity gain op-amp buffer. I took power from the plate pin of the 12AU7 and the input from the grid connection. There was a 2.2 meg grid resistor to ground so I used a 2.2 meg to B+ to bias the op-amp into the middle of its range. Since these are high value resistors, I used a J-Fet amp. The output is capacity coupled to a BNC connector that gets the stock cable to the front panel. The op-amp board is mounted on the stock chassis BNC connector that also gives the ground connection.
The connections are made through a 9 pin plug/socket with the follower plate and grid connections diverted out. The 12AU7 plugs into the socket so the filament string and the other half's circuit aren't disturbed. I found a tall tube cover with the heat sink insert to complete the installation.
Vacuum tube purists will probably want to skip this section...
The final AF audio output amplifier tube is a 26A7GT. This tube consumes 16 watts of filament power (26.5V, 600ma) to produce only 200 milliwatts of audio output. Helps keep the radio nice and warm, too. A number of plug-in solid state audio replacements have been made over the years. They all provide greater audio output and consume less power than the 26A7GT. The greater output is obtained without overloading the B+ circuit by using the beefier (in amperage) filament circuit. Here is the schematic of the "official" solid state audio module. Mike Dinolfo's friend "John Doe" says that garden variety darlington NPN power transistors, and the 6AJ5 audio amps by 2N2222 or 2N3904 NPN types. (Both will require bias resistors). Rudy Salomon has contributed a LM383-based amp schematic. It should be noted that some audio chassis changes may be required to use either the "official" or after-market solid state audio modules. R629, a 8.2 megohm resistor that provides feedback to V606, the first AF amp, is used in receivers that use the original 26A7GT, but not used in receivers with the "official" SS audio module. This feedback is provided inside the solid state audio module. Many users have reported feedback and squeals at low audio gain settings when changing from 26A7GTs to the SS audio module. As previously noted, don't try the AC filament mod if you have any sort of solid state audio module.
If push comes to shove, and you really can't round up the requisite number of 26A6 or other tubes, a 40763 MOSFET can be more or less directly substituted in the signal and B+ path. Or so it has been said. Hey, I told you that you might want to skip this section.
Mike Dinolfo's friend "John Doe", reports that the 2N3819 or 2N3823 can be substituted directly for the 26C6 mixer tubes. Solid state diodes can be substituted for the diode sections of V601, V602, and V603.
Barry Hauser has a R-392 which has been completely converted to solid-state active devices and to 12 VDC vice 24-28 VDC. Don't blame him either, it was that way when he bought it. The front panel has 3 of the fuse holders (plate and 2 spares) removed (in addition to all panel markings being painted over). The IF/AF/PTO/Calibrator modules clearly show tubes replaced by either single transistors or IC modules. Barry is currently collecting some basic performance data to see how this radio performs vis-a-vis the original R-392 specs.
Additional general information on building solid state replacements for tubes is contained in "Synthesizing Vacuum Tubes", Parker R. Cope, QEX, August 1997.