This is what has been working for Romy for a while:
This is the essence of the DSET concept - amp stages dedicated to the specifics of the playback channel in regards to filtering, loading of the tube, match of the frequency range to the driver. No big transformers required where the music does not have to move a lot of current.
He uses a unique bias scheme that normally needs more than a quick glance.....
The MF channel does not use parallel tubes, it is either or. I think that he currently uses the YO186. There is a thread about the testing to choose between the various DHT tubes.
romys dset concept
#46 The proof is always in the pudding
Whenever I feel blue, I start breathing again.
-
- User
- Posts: 86
- Joined: Tue Jan 28, 2014 10:23 pm
- Location: Boston, MA
- Contact:
#47 Re: The proof is always in the pudding
Yes, the filtration in this amp is a bit confusing. The basic idea is that cathode shall be in ground and a good non-inductive resistor in grid is better than a capacitor in cathode. This is a virtual battery in grid where bias voltage is formed by dropping of voltage before and after (red and green) resistor R5, well visible in here:
http://www.goodsoundclub.com/PDF/Melquiades_SET.pdf
The supply of "soft type" of positive and negative voltage via gas tubes is different subject and it was made due to some sonic objective that are outside the scope of this thread. The idea of initial prototype is visible here:
http://www.goodsoundclub.com/Forums/Sho ... =4973#4973
Even the selection of filters is not final but you will see the progression of thinking. The objective were the following: the filtration element has be minimized, all resistors have to be brought to a common denominator in order to minimize the number of them. The filter has to be build that the resistor serve 3 duties in the same time: form a Bessel slope filter, accomplish necessary voltage division to equalize the difference in channels sensitivity, form a necessary different between positive and negative supply to milk 4V bias voltage. Also, the filter has to be subtractive and to eliminate any serial capacitance with exception of HF channel that is made immune to the capacitance problems.
The side effect of milking a resistor on grid is that it form a low pass filter with Miller capacitance of the tube. That itself is not a problem at as those tubes are very fast and are benefited from some grid stopper and the DSET amp has no objective to have good HF extension to begin with. All of the channels are explicitly low passed well below the natural Miller roll of.
http://www.goodsoundclub.com/PDF/Melquiades_SET.pdf
The supply of "soft type" of positive and negative voltage via gas tubes is different subject and it was made due to some sonic objective that are outside the scope of this thread. The idea of initial prototype is visible here:
http://www.goodsoundclub.com/Forums/Sho ... =4973#4973
Even the selection of filters is not final but you will see the progression of thinking. The objective were the following: the filtration element has be minimized, all resistors have to be brought to a common denominator in order to minimize the number of them. The filter has to be build that the resistor serve 3 duties in the same time: form a Bessel slope filter, accomplish necessary voltage division to equalize the difference in channels sensitivity, form a necessary different between positive and negative supply to milk 4V bias voltage. Also, the filter has to be subtractive and to eliminate any serial capacitance with exception of HF channel that is made immune to the capacitance problems.
The side effect of milking a resistor on grid is that it form a low pass filter with Miller capacitance of the tube. That itself is not a problem at as those tubes are very fast and are benefited from some grid stopper and the DSET amp has no objective to have good HF extension to begin with. All of the channels are explicitly low passed well below the natural Miller roll of.