• Dirty little EL84 amp

A friend was interested in having an amp for slide playing that was low wattage so he could crank it and get power tube distortion. He liked EL84 tubes, and was looking for around 15 watts of power. He won this amp on ebay. To those design goals I added that it should cost as little as possible to convert it. This meant using as many as possible of the components that were already in the amp, and whatever I had in my junk box.

I don't recognize the type of tone control circuit it used, but I left it alone because it provided a little tone shaping, and would decrease the number of new parts that would otherwise be needed.

The preamp tube is a 12AX7. I configured this like a Fender preamp with the tone circuit and volume control sitting between the 2 tube stages.Primarily this was to avoid having to find another coupling capacitor that would be needed if I configured it like a Marshall with just the volume control sitting between the tube stages. But I also didn't want to have too much preamp distortion. I was aiming for a fairly clean preamp that slammed the output tubes hard.

The power amp looked ready to go.When I first fired it up, the power tube plates started glowing red so I increased the cathode resistance. The power transformer was mounted slightly crooked, and didn't fit the holes in the chassis, so it had probably been replaced. B+ was a little high for EL84 tubes, but I think Vox amps put about 400 volts on the plates, so I figured the bias being off was due to this old repair.

The phase inverter was the type used by some Gibson amps where the signal from the plate of one tube stage is attenuated and fed to the grid of another tube stage. I thought that changing it to a long-tailed phase inverter would give it more gain and make it distort more easily. EL84 tubes need much less drive than 6V6 tubes, so I took the phase inverter design used in Fender amps. (Nuke wrote up a great explanation of this in alt.guitar.amps.) I figured this should make the EL84s distort very easily. The phase inverter tube is a 12AU6, and I figured if this didn't work out well, I had the option of using a 12AX7 for even more drive.

The power looked fine and luckily the 3 section filter cap seemed OK.

The amp with this set up was dirty and distorted (in a good way) right at the start. Just opening up the volume control a little brought you into distorted tones, so that goal had been addressed. There had been a kind of presence control that had been lost when I changed the phase inverter, so there was an unused pot that I used to add negative feedback to tame some of this distortion. On first test with this wired in, the amp squealed whenever you tried to apply NFB, so I reversed the output transformer connections to the power tube plates and that was fixed. The NFB allowed you to get some clean tones, but it really dampened the character of the amp. I left it because you could dial it out, and maybe there would be a use for it later.

So Kevin comes over, we're playing with the amp to see where to bias it, and after a while we notice there's a fading-out volume problem. I thought at first it was him playing with the volume control on the guitar. In playing it loud, the volume would decrease like turning down the volume pot 1 or 2 numbers, and then come back by itself. Also the distortion would tail off in a shelf-like curve--a quick transition with a large change in the amount of distortion while before and after the change the distortion was tailing off more smoothly.

It turned out the coupling caps on the outputs of the phase inverter were leaky. Pulling the power tubes and measuring the voltage from the grid to ground showed a positive voltage. Unsoldering the ends of the caps and measuring the voltage at this point across a 1M resistor to ground showed 3 volts of DC--there should be 0 volts. So new coupling caps, and the problem went away.

This was also the cause of the bias problem mentioned earlier. Now when the original cathode resistor was used the plates didn't glow. The positive voltage on the plates was messing up the bias.