Transistors

No OP-amps ?
When designing larger professional units like mixer desks it is unpractical to use discrete amplifier stages, simply because hundreds of them are needed.  But a well designed discrete amplifier will always outperform an OP-amp simply because each stage of the amplifier is built with the absolutely best transistor for the purpose. An integrated circuit, like an OP-amp, always has a very limited flexibility in this respect (instead they rely on high feedback to give good measurement results).  I realize there are some high-end products that use OP-amps, but I never found their gear very musically involving, at least not to the level expected for the price.

Does the transistors make a difference ?
Even a very good transistor is by far the most non-linear device in the whole design, provided everything else is done right (e.g. grounding, decoupling, PCB layout).  The difference between transistors are VERY big, and must be given much attention if high-end performance is desired. The often used BC546 or BC550 are good examples of truly poor devices.  

If you want to do some tests, do what I did 10 years ago.  A 20W power amplifier was first built with BC- and BD-devices, and it sounded fairly OK.  I then replaced all transistors to the audio types below without changing anything else.  The difference in clarity and distortion was so big it was hard to believe, the distortion just disappeared !  This chocking test can be performed on any well designed amp.  As a more simple test, try to swap the input differential because this stage is often the dominant source of distortion.

I can only repeat my advice: if the transistors are not given attention you will struggle with reaching high-end performance.  You will probably at best end up with mid-fi.

Input differential stage
2SC1775A-E and 2SA872A-E (Hitachi)
A good example of a very linear bipolar device without the typical "transistor sound" you find in the BC550.  The Beta is totally flat as the collector current varies.  After about 1000 hours of brake-in the transistor is free from a slight coloration in the higher mid-range, and it works beautifully.

2SK30AGR (Toshiba)
This is the most neutral sounding jfet I ever crossed.  Its Gm is quite low which gives good overload margin even with moderate degeneration.  The sonic signature is very smooth, natural and totally grain-free, with a slight roll-off in the extremes as the only drawback.  If more bass slam is desired the bipolar transistor is recommended, but you will lose some of the smoothness (see how difficult it is to make full range amplifiers perfect throughout the audible range, that´s why frequency optimized bi- or triamp configurations beats everything in the world).

2SK43-2 (Sony)
This jfet is very close the 2SK30 and should work equally well.  Note the different pinning  - they are not pin-compatible.

Intermediate buffer stage
2SC1775A-E and 2SA872A-E (Hitachi)
2SA1038 (Rohm)
In higher voltage applications (Vce above 50Volts) I prefer the 2SA872A-E to the 2SA1038. In low voltage applications my preference is the opposite.  But both are very good.  If you need an NPN-buffer stage the 2SC1175A-E is the one to use.  

Driver stage and low level output stage
2SB1212 and 2SD1812 (Rohm)
2SB648A-C and 2SD668A-C (Hitachi)
2SB649A-C and 2SD669A-C (Hitachi)
2SA1837 and 2SC4793 (Toshiba)
For power amps where the driver needs additional cooling I use the TO126 devices from Hitachi.  The actual design of these transistors is very old but they are still very good.  If the power dissipation is too high for these devices use the Toshiba ones.  For low level output stages (e.g. line amps) I use the Rohm´s which work very good.

Output devices
2SB753-Y and 2SD843-Y (Toshiba)
2SA1386 and 2SC3519 (Sanken)
The Toshiba devices are 7A/40W TO220 types and are one of the best sounding power devices I ever crossed.  The design is old and they are hard to find even in Japan.  They have non of the grainy character you find in most bipolar power devices.  

The Sanken devices are my latest pick.  The Beta of this TO3P device is very high at high currents (>50 at 5A) so the stage driving them is not stressed even at high current peaks.  You can use a two-stage EF instead of a triple configuration, and thus stability is easier to maintain.  The data sheet shows an amazing linearity of this device so the feedback can be kept low while maintaining clarity, all resulting in increased dynamics of the amp.  

What about MOSFETs or IGBTs ? Well, the bass slam is always poor with MOSFETs.  I like tight bass so bipolars are my obvious choise.  On IGBTs I must say I have no experience, I´m just a hobbyist you know...

All rights reserved