I bought a "technician special" Heathkit SB-102 transceiver. Everything seems to work well until I attach a dummy load and try to go through the "initial tune up" procedure in the manual.
At the point where I switch the "Mode" to "Tune", instead of showing 50-60 mA plate current, I'm seeing an initial switching spike and then the needle settles around 150-250 mA and begins to creep upward. When it gets past 300 mA (as little as 10-15 seconds), the upward creep accelerates, and when it gets past 350 mA the breaker in the HP-23A power supply pops (presumably to protect various components). During this time, grid current appears to be negative -- the needle goes past the zero point on the meter (which I've zeroed with the zeroing screw when power was off).
Output was to an MFJ 300W/30 second dry dummy load (derating graph shows 100W up to 90 seconds, and hasn't even gotten warm in any test). If there's any resonance going on, it's inside the transceiver's housing. Further, it does the same thing on bottom of 80 m, 40m, and top of 10m that I've checked (as a recommended check against self-oscillation in the final stage).
I have very limited diagnostic tools -- a digital multimeter (cheap unit from Harbor Freight) and kit-built frequency counter (which I'm sure would need attenuation for use in this situation, even if it were likely to be useful). I could build an oscilloscope probe to plug into the headphone jack of an Android device, but I doubt that would be any more useful than my multimeter.
The manual doesn't ask for more than a VTVM -- but makes it clear this isn't the same as a VOM, specifically for resistance testing circuit segments with diodes, and as I understand it, for this type of testing, my DMM is more like a VOM (low test voltage) than a VTVM. Comments have suggested that this modern digital meter is more like a VTVM than a VOM from the era when the SB-102 manual was printed, so should be okay for these tests.
I've replaced the mismatched 6146 and 6146A final tubes that were in the rig when I got it with actual 6146 tubes -- I now have three with varying date codes, and I'm using the closest-dated (on Internet advice), about a year apart. No change.
Not sure it's relevant, but turning down the variac (feeding the power supply ~100 V instead of ~120 V), thereby reducing plate voltage from 800+ to just under 700, has reduced this effect, lowering the initial settling point and slowing the upward creep in plate current.
Question is, at this point, what kind of failure do I need to look for? Bad grid leak resistor on one of the 6146 tubes? Failed capacitor in the final stage?