After enjoying the amplifier for a few months it started to blow the secondary winding’s fuse when the amplifier was running for a few hours. Now this happened in the summer with temperatures between 25 – 35°C ambient temperatures. The temperature inside of the enclosure was obviously a bit hotter. That confused me a bit because when I powered up the amp, currents were normal and everything played nice. No microphonics, nothing…
So is this a temperature issue? Is the fuse overheating? The type of fuse I used was a 250V normal temperature range one. Is the value too low?
I replaced the fuse with slightly higher value. No change. I replaced the fuse with a 600V high temperature model. No change.
Then it struck me, runaway bias? Measured the bias voltage over time. As stable as it can be. The shunt regulator and the high precision cathode resistors did what they were supposed to do. While pulling hairs and growing a few more grey ones, I was staring at the tubes and the volt meter measuring the voltage over one of the cathode resistors and it happened. Slowly the anodes of channel started glowing red and the current started running away. What!!!
The answer is actually quite simple, the characteristic of the tubes change over time and change when they warm up. The quiescent current balancing did what it must and balanced the current of the tubes perfectly. The run-away tube took the other tube for a ride. The solution was simple, increase the negative bias voltage and all stabilized. However, I do not want to run the tubes cols (all the time). If I run them at a safe 50 to 52mA we are OK. I wanted a bit over 60mA through my KT77.
The issue is that because of the hybrid fixed/cathode bias the regulation that the cathode bias brings is reduced. Increasing the cathode resistor would help. Changing to auto-bias seems the ultimate solution. But I do not want to give up the quiescent current balancing nor do I want to give up the mixed fixed/cathode bias.
What to do? The answer is to adjust the fixed bias voltage per channel in slow speed to counter the changing tube characteristics. I keep the current balancing as is and let the transistor pair keep the tubes aligned real-time. The fixed bias voltage is controlled by a digital potentiometer controlled by a μController. But that is for another time…