Roger brought his new Welborne Labs / Stereo 70 by for a smoke check this month. The amp is his first scratch-built project, and he asked me to check his work and bring the beast upon a Variac for him.
There were a couple of minor wiring mistakes which were due to slightly misleading drawings in the Welborne manual. We found these through the checkout procedure I'll list here. If you're doing a first construction project, having someone with experience check your wiring can save you quite a few hours of tail chasing if something's wrong. A person who hasn't been staring at the project for three days and nights can pick out the obvious flub easier than the builder whose eyes are crossing from hours of concentrated wiring. Roger's wiring looked pretty good, but there were a lot of differences in the layout from a stock Stereo 70, and some wiring was hidden under PC boards, so a cautious approach to smoke testing was taken.
It was time to check things with an ohm meter. I checked across the filter cats from positive to negative and checked for shorts or very low resistance. Everything was nice and high, so I went on to the ground buss, making sure all connections were reading near zero, indicating good solder joints.
With the visuals and general disaster avoidance checks done, it was time to fire up. We placed 20W 8 Ohm resistors on the output posts and shorting plugs in the inputs. A lot of people think this is overly cautious, but an amp with a problem con go into oscillation, and this can take out an unloaded output transformer if it's bad enough. Better safe than sorry, particularly if it's your amp I'm working on.
I plugged the amp into a 5 Ampere Varitrans and connected a 1200VDC meter to the B+, and a pair of digital VOMs to the bias circuit test points, one for each channel. Then we brought the juice up slow, to about 60VAC, and let things sit for a couple minutes. Everything was cool, so I went for 90VAC.
At this point some B+ started to show on the meter, and filaments showed a faint glow. Everything continued to look OK, so we hooked up speakers and a source.
I brought the B+ up slowly to full tilt, but nothing happened.
Turned out the filament voltage on the output tubes was about 2 Volts instead of 6.3. The two 6.3V filament windings from the power trannie had been joined out of phase because it looked like they should be on the wiring diagram. Once this was corrected, we got the bias adjusted and things sounded good. All was well.
For a week anyway.
Roger called and said the amp just stopped playing after about three days. I took the amp back. Everything checked out fine with an ohmmeter, no damage done. Again I fired up slowly. This time the B+ came up fast, not like a 5AR4 does, and slowly dropped to about 40V lower than last time. It seemed to be slowly dropping as it got warmer.
Eric had told me that Russian 5AR4's, which this was, are really 5Y3's, and seldom hold up when put under a big load. I put two and two together and replaced the Russian rectifier with a good ol' GE NOS 5AR4.
B+ was right where it belonged after a nice slow ramp up. I monitored things for an hour or so and all seemed well. Watch those Russian tubes!