1993: First project: EL84 poweramp
1994: Second project: EL34 poweramp
2004: New tubeamp with KT88's
In may 2006, this KT88 amp was given a major upgrade. Instead of the 39Ω resistor between the two electrolytics, a choke is used in the power-supply. Together with the electrolytics, this forms an effective C-L-C filter, and this suppresses the ripple on the supply voltage a lot better compared with the C-R-C configuration. The compact choke has an inductance of 2H/300mA. This seems a bit low, because most of the times sometihing like a 10H choke is used here. But after a few simulations with PSU Designer II from Duncan's Amp Pages it showed that a big choke causes too much ringing. In the graphs below, the voltage across the second cap is shown, right after power-up.
200µF - 10H - 200µF 64µF - 2H - 200µF
On the left, a simulation with a 10H choke instead of the resistor can be seen, without further adaptions. On the right is a 2H choke with, in the end, a smaller smoothing cap. A 2H choke is a good compromise between ripple reduction and ringing in this case. To further improve the behaviour of the C-L-C circuit, the value of the first electrolytic has been lowered: instead of the 2x100µF JJ cap, a 2x32µF F&T is used.
The smaller smoothing cap and the extra impedance of the choke cause the supply-voltage to be lower. That is the price for such a configuration: a higher ripple voltage before the choke, and more voltage drop. This has been solved by using a 25V higher tap on the mains transformer.
Next to the addition of the choke, the decoupling caps across the kathode resistors have also been replaced. The 1000µ/50V Nichicon Muse (KZ) caps have been replaced by 220µ/50V Black Gates. The 1k8 resistors for the supply of the preamp stage and phase splitter were lowered to 1k5, to adjust the voltage to the new situation. Also, better quality types were used.
September 2008: time for the next jump forward: the Auricap coupling caps have been replaced by Mundorf MCap Supreme capacitors. For the preamp stage, the Supreme silver/gold was chosen, and the Supreme silver/oil was chosen between the phase-splitter and the output tubes. This choice is mainly based on the costs: the silver/gold types are more expensive, so these are only used in the first stage, where only one is needed per channel.
In a number of places, there were still 'common' carbon film resistors present in the circuit. The 470k grid resistors in the preamp stages have been replaced by 2W resistors from Kiwame, just like the 2k2 damping resistors in series with the grids of the output tubes. Also, the 100k grid resistors were replaced by PRP resistors. The 47k feedback resistor has been replaced by a Holco H4. A number of the kathode resistor turned out to be more than 10% out of tolerance. Enough reason to replace the cheap cement resistors with something better: 8W emamelled wirewound resistors (RWM series) from Vishay-Sfernice.
2007: EL34 amp 'revived'
The EL34 amplifier from 1994 had to donate its output transformers to the new KT88 amplifier in 2004. It was a very nice sounding amplifier however, that has been sitting unemployed in a corner for a few years, which is a shame of course. It has an ultra-linear push-pull output stage, with two EL34s in auto-bias. They each have a fixed cathode resistor that is bypassed by two electrolytics and a large MKC cap. One ECC81 doubletriode acts as a pre-amp and phase-splitter.
The amp is given a new set of output transformers and is made to play again. All transformers were also rotated by 90 degrees, because their orientation was actually wrong. The magnetic field of the power transformers was pointed towards the tubes and the output transformers, which caused a small hum-voltage on the output. The PCBs have received new capacitors and the signal wiring in the case is also replaced.
'Power amplifiers with valves'
A very interesting article, very pleasantly written by Claus Byrith, for anyone who wants to know more about designing a push-pull tube amplifier. The input-stage, phase-splitter and the output stage are all discussed separately. With some practical examples, this finally results in the complete amplifier: 'The 4-30'. The schematic looks a lot like that of the above EL84 amplifier, which uses a similar input-stage with the EF86 in triode, and the directly coupled phase-splitter with an ECC83.