Recently, I was tasked with the complete re-design of a disaster of a DiY tube amp that a customer had brought me. It was built from organ parts. As such, the audio transformers were unsuitable for high fidelity service, the amp was cathode biased and the phase inverter was not balanced properly. As arrived, this stereo amp with four KT-88 output tubes driven by 6SN7s was putting out about 10 watts with 7% THD and 19 watts with 20% THD.
I've kept it as simple as possible with the limited number of existing tubes in the chassis, but I've filed the many ideas presented here for future design efforts.
This amplifier, with conventional bias of a fixed value, would draw 255 watts off the mains at idle.
Due to limitations of the power transformer current delivery, I found that I could take advantage of short term dynamic power if I reduce the bias idle current to around 20mA.
With this modification involving signal responsive bias, the idle power consumption is 125 watts. At full output, the line consumption is 380 watts and the B+ potential sits about 98 volts higher than with conventional bias.
I'm taking the plate outputs through a high resistance network, blocking the DC and rectifying the result. The bias supply is -117 volts, regulated via a string of four 30V zeners. A resistive divider samples the bias supply and the signal rectifiers. At no signal, the bias on the KT88 grids sits about -85V. As signal increases, the bias is bucked by the output of the signal rectifiers sampling the plates and this voltage is summed with the bias supply. As signal increases toward full output, the bias voltage is bucked by the positive envelope voltage from the signal rectifiers, resulting in a maximum signal bias voltage around -35V. Using a 100V zener in series with a 100K resistor, tied to the -117V supply, I can limit the positive bucking voltage so that the bias never goes lower than -35V with maximum signal.
The only tricky aspect of this was getting the output tubes out of cutoff before significant signal appears. This limits me from having the outputs biased at -115V (which gets the line consumption down below 90 watts at idle). But I'm happy with 125W idle power consumption. The tubes run so cool that I can briefly touch the envelopes without pain. I expect tube life to be greatly extended via this signal responsive bias system.
The nice thing about this configuration is that the tubes run very cool. You can touch the bottles of the KT-88s briefly without getting a burn. The run no hotter than the driver tubes. It won't heat the room up, so it's a good choice for summer and warm climate use. And tube life should be greatly extended.
As for the sound, it feels comparable to a 240 watt solid state stereo amp, in terms of bass impact. Distortion is very low across the power range, thanks to a servo balanced phase inverter, which employs a transistor in the cathode circuit to sample the output of the two halves and amplify errors, using that current change to adjust the cathode current, keeping the inverter outputs within 1% of eachother at all times.
The customer's friend does custom cabinet work, and this is what the finished product looks like:
I'm thinking this is a good product for the Japanese market and I could build similar amps and sell them in Japan.
I've kept it as simple as possible with the limited number of existing tubes in the chassis, but I've filed the many ideas presented here for future design efforts.
This amplifier, with conventional bias of a fixed value, would draw 255 watts off the mains at idle.
Due to limitations of the power transformer current delivery, I found that I could take advantage of short term dynamic power if I reduce the bias idle current to around 20mA.
With this modification involving signal responsive bias, the idle power consumption is 125 watts. At full output, the line consumption is 380 watts and the B+ potential sits about 98 volts higher than with conventional bias.
I'm taking the plate outputs through a high resistance network, blocking the DC and rectifying the result. The bias supply is -117 volts, regulated via a string of four 30V zeners. A resistive divider samples the bias supply and the signal rectifiers. At no signal, the bias on the KT88 grids sits about -85V. As signal increases, the bias is bucked by the output of the signal rectifiers sampling the plates and this voltage is summed with the bias supply. As signal increases toward full output, the bias voltage is bucked by the positive envelope voltage from the signal rectifiers, resulting in a maximum signal bias voltage around -35V. Using a 100V zener in series with a 100K resistor, tied to the -117V supply, I can limit the positive bucking voltage so that the bias never goes lower than -35V with maximum signal.
The only tricky aspect of this was getting the output tubes out of cutoff before significant signal appears. This limits me from having the outputs biased at -115V (which gets the line consumption down below 90 watts at idle). But I'm happy with 125W idle power consumption. The tubes run so cool that I can briefly touch the envelopes without pain. I expect tube life to be greatly extended via this signal responsive bias system.
The nice thing about this configuration is that the tubes run very cool. You can touch the bottles of the KT-88s briefly without getting a burn. The run no hotter than the driver tubes. It won't heat the room up, so it's a good choice for summer and warm climate use. And tube life should be greatly extended.
As for the sound, it feels comparable to a 240 watt solid state stereo amp, in terms of bass impact. Distortion is very low across the power range, thanks to a servo balanced phase inverter, which employs a transistor in the cathode circuit to sample the output of the two halves and amplify errors, using that current change to adjust the cathode current, keeping the inverter outputs within 1% of eachother at all times.
The customer's friend does custom cabinet work, and this is what the finished product looks like:
I'm thinking this is a good product for the Japanese market and I could build similar amps and sell them in Japan.
