One of the 68 ohm screen resistors on the EL-37 output tubes has burned open, and previous history component replacement shows that this is a repeat problem on this and another model 100. Running screen voltage is 470, and this value is significantly exceeded during warmup. An output tube arcs over, and there goes the screen series resistor. Should I move the screen tap down the power supply filter line to a point more in line with the tube manual maximum screen voltage for 6L6/EL37 types?
The Photofact shows some production units had a 500pf cap between cathode (8) and plate (3) of one of the output tubes. The schematic shows V4, the pictorial shows V3. There may be a correlation between this added component and units produced with 5881 tubes. Is this a move to quell a parasitic oscillation, and should I premptively add the change to the model 100's on hand?

I would recomend a 1K ohm half watt grid stopper resistors on the G1 grid and the screen grids , also change the , the plastic caps to the 6L6 grids and both the large can caps under the chassis ,
They are 40UF/40UF at 450 WVDC if i remember corectly , Use this for the second cap http://angela.com/100uf100uf500vdcclamp ... citor.aspx and this one closest to the rectifer
http://angela.com/32uf32uf500vdcpolariz ... citor.aspx you will need a capacitor clamp for both of them .You will also need a 50UF at 50 wvdc cap across the cathode resistor .

Wouldn't use such a large cap as 500 PF ,100 PF would be more like it , the grid stoppers should take care of oscilations beter than the cap

Thanks for the input on taming oscillations, supposedly in the rf range. I have looked at like-application schematics in a 1958 Heathkit catalogue that I didn't throw away (schematics of all their product offerings) and the RCA Receiving Tube Manual. What I see is control grid de-q resistors ranging from 1K to 22K, depending I suppose, on how tough of an oscillation beast the designers were trying to tame. There is a need for some degree of stiffness of the screen grid voltage. Value of screen grid resistors go from 56 to 270 ohms. Screen draw of a 6L6 can go to 22ma, so a 1K resistor dissipation could be right at 1/2 watt.

The folks at Fisher were apparently addressing a problem. The output transformer plate leads are in that spiral shielding that Fisher used. Also the existence of the screen grid resistors. Did the resistors open because of tube internal arcing from the power supply voltage per-se, or a combination of supply voltage and runaway voltage from an unloaded rf oscillation? I don't know, and don't care to repeat the event.

This I do see, Fisher is(was) definitely pushing it as far as tube rated voltage, and voltage rating of power supply capacitors. For the transformer, 600 to 700v caps is more like it (what other schematics show). My thought is to move the first filter cap from the rectifier side to the load side of the series 100 ohm / 10W resistor, and make this more of a "choke" input filtering setup. Ideas? This also gets away from pushing the dickens out of the 6L6's.

A 5U4GB will drop about 25 more volts than the 5V4G , if you use 6L6GC's they will handel the extra voltage with no problems .

Summary of repairs made and outcome:

These are known as “locomotive” amplifiers. You see one, and you see the resemblance. There were three types of problems to be dealt with. (1) The installation had the amps in a confined space that led to over temping and parts failure, design layout that contributed to heat related part failures, and running tubes well in excess of ratings that led to oscillations, internal tube arc-overs, and blown parts.

A general run-hot issue was addressed by (1) don’t cram the amplifiers into a confined space with scarce ventilation, (2) mount the chassis on ½ inch high feet for better cooling, and (3) drill a number of 3/8 inch holes in the chassis bottom cover to facilitate air circulation.

The original filter cap installation was in the form of two “cans” mounted in a compartment under the power transformer and blocked-off from any cooling air circulation. Read many parts failures and replacements over the years. Replacements were done with individual axial units which are smaller than the originals. Layout was changed to avoid jamming parts under the high-heat transformer area. An all-in-one power chassis-clip-mount resistor unit was used to for B+ filter isolation, cathode bias resistor, and variable damping function components. This toasty arrangement had burned wires and run potting tar out of the output transformers. A prayer session was held in hopes that the power and output transformers were still good. Prayers were answered. The power supply resistor segment was abandoned in place, and a separate 10 watt rating part was used and placed well away from other parts.

Coupling capacitors had failed many times. Count where wires were cutoff, and new parts tacked in. Again all coupling capacitors were replaced with metalized polyester construction parts. The time was taken to carefully remove all the component lead debris, and wire in/solder the parts as it were an original installation.

The 6L6 output tube screen connection anti-oscillation resistors, 68 ohms, had failed many times. On initial power up, one failed again. Pop. Investigation showed that with Fisher running a capacitor input power supply filter that the B+ voltage to the 6L6 plates and screens was 550 volts. Max rating for the basic 6L6 is 350 volts, and the final version of the tube, 6L6GC, is 500 volts plate and 450 volts screen grid unless Williamson Ultra-linear operation is being run. In that case 500 volts screen voltage is permissible.

Several fixes were done. First, the capacitor input filter arrangement was changed to a series input resistor. This was easy enough to do by moving the high side of one filter capacitor lead. The result is 450 volts on the plates and screen grids.

One K-ohm control grid anti-oscillation resistors were also installed, and the screen grid resistors were changed to 270 ohms to follow observed practice with other high sensitivity power amplifier designs.

A new set of SOVTEK 6L6WXT+ output tubes were installed in both amplifiers. One amplifier had a bad 5V4 rectifier. This was replaced with a 5AR4/GZ34 tube that has better voltage and current ratings while needing slightly less filament current. Call this a plus all around.

One more suggestion: get a quiet fan to keep air moving over the amplifier.

-Chuck