I had questions about my amp(s) -a pair of monos- but I can tell only what I know. It is my first ever tube project and it has been "evolving" for more than 15 years. Most of what I know about tubes -together with what I don't- came out of this. If anything, I post it here as evidence of what a laboratory mouse can do provided with time and resources. 🙂
It all started from the schematic bellow which seemed very attractive for an amateur equipped only with a DMM. Still, not informative about that 6,3V going where. Go figure!
First implementation shown bellow. Very heavy, kitchen oven wiring, integrated stereo with all kind of inputs and twelve region graphic equalizer per channel. Custom made transformers by a local manufacturer.
Soon it was split in two halves with solid state omitted and these served as "guinea pig" platforms for whatever I found on the internet before I came across to diyaudio.com. In fact, I became a follower of this forum searching for solutions for this amp.
From the above to its current state, all steps were taken one by one. The picture bellow shows what stayed on. Many other things have been tried.
The final schematic is following. Admittedly, not what an engineer would do out of parts ready to deliver 35W in class AB or even 50W in class B. This is just 16W in class A.
I know exactly the difference each new part made to the sound but I'm going to keep it brief. There are some obvious modifications for reliability, like the power section negative bias circuit and the heaters DC elevation. Then the NFB was removed together with the equalization network at the original EF86 input tube. Also, for this the coupling caps to the output stage had to be increased. Sparkling gain was addressed by wiring input and output tubes in triode mode -and moving from EL34 to KT88. One of the most important changes was to replace the ECC82 with the 6N6P with tail CCS. The soundstage got its third dimension after that. The particular CCS has to deal with 140V/34mA and you can see the flimsy heatsink on the top of the pic showing the inner parts. It is floating at 140V and I've found that shunting it with a cap to ground reduced third harmonic by about 5dB.
The power supply is something one will never find in an "orthodox" schematic. Definitely, it's not going to work inside a feedback loop let alone class AB. The idea was to decouple all three stages taking advantage of the steady current consumption. For this, the input stage moved to differential double triode with tail CCS, with ECC40 chosen because it's electrically similar to the EF86 and was fitting to the same hole. Otherwise it should be 6SN7. Accordingly, the output stage was converted to class A by means of two power resistors to take B+ from 500V down to 375V. Originally meant to be a temporary test but it stayed.
PSUDII was extensively used for simulations. I was looking for low ripple and fast clean transition on load transients -if any.
Finally, tube operating points and basic measurements.
ECC40
6N6P
KT88
Distortion @1W
I ended up with a relatively colorless amp... Whenever I want the tuby sound I have to connect some solid state! 😛
It all started from the schematic bellow which seemed very attractive for an amateur equipped only with a DMM. Still, not informative about that 6,3V going where. Go figure!
Soon it was split in two halves with solid state omitted and these served as "guinea pig" platforms for whatever I found on the internet before I came across to diyaudio.com. In fact, I became a follower of this forum searching for solutions for this amp.
From the above to its current state, all steps were taken one by one. The picture bellow shows what stayed on. Many other things have been tried.
The final schematic is following. Admittedly, not what an engineer would do out of parts ready to deliver 35W in class AB or even 50W in class B. This is just 16W in class A.
The power supply is something one will never find in an "orthodox" schematic. Definitely, it's not going to work inside a feedback loop let alone class AB. The idea was to decouple all three stages taking advantage of the steady current consumption. For this, the input stage moved to differential double triode with tail CCS, with ECC40 chosen because it's electrically similar to the EF86 and was fitting to the same hole. Otherwise it should be 6SN7. Accordingly, the output stage was converted to class A by means of two power resistors to take B+ from 500V down to 375V. Originally meant to be a temporary test but it stayed.
PSUDII was extensively used for simulations. I was looking for low ripple and fast clean transition on load transients -if any.
Finally, tube operating points and basic measurements.
ECC40
6N6P
KT88
Distortion @1W
I ended up with a relatively colorless amp... Whenever I want the tuby sound I have to connect some solid state! 😛
Hi, non expert question:
in the circuit of post #1, what difference would do if the 1.5k and cap where after the cathode, 680R to ground, and NFB applied in between?
in the circuit of post #1, what difference would do if the 1.5k and cap where after the cathode, 680R to ground, and NFB applied in between?
Hi,
Non expert here too. Just speculations. Perhaps Tadaatsu Atarashi - the designer- had decided for this ratio between feedback and cathode degeneration. He also had published a KT88 ultra linear version with less feedback and half way rectifier.
Non expert here too. Just speculations. Perhaps Tadaatsu Atarashi - the designer- had decided for this ratio between feedback and cathode degeneration. He also had published a KT88 ultra linear version with less feedback and half way rectifier.
Speaking about cathode resistors, I forgot to mention the 82 ohm at the cathodes of the KT88. This is my interpretation of Kiebert's paper discussed elsewhere -attached pdf. It is a method to reduce H3 by inserting a resistor of value equal to 1/2gm to smooth crossover distortion in class AB. I found experimentally using a wirewound pot that also works in class A but the value is now 1/gm.
Very interesting evolution. Well written, thank you for sharing your journey.
Was the reduction in power worth the gain in listenability.
Thanks!
Surely, all the mods resulted in big improvement of sound quality. Power reduction simply made it easier to achieve. Needless to say this amp goes together with sensitive speakers.
Surely, all the mods resulted in big improvement of sound quality. Power reduction simply made it easier to achieve. Needless to say this amp goes together with sensitive speakers.