|Elliott Sound Products||Valve Amplifiers - do they really sound different?|
Page Last Updated 28 Nov 1999
Despite a decade and a half of transistor development, many hi-fi enthusiasts have remained faithful to valve amplifiers. Now it seems amplifier manufacturers worldwide are having a second look at this once all-but-discarded technology.
|This article is a partial reprint from the November 1977 edition of Australian Hi-Fi Review, with only the discussion section and review of the hybrid valve amp I designed. Please see the copyright notice at the end of this document.|
When Harold Leak introduced his company's (H.J. Leak & Co) first transistorised amplifier, the famous Leak Stereo 30, he remarked that in his opinion, eliminating the output transformer was the greatest advantage of the new model.
The Stereo 30 was a very significant development. It was the first commercial transistorised amplifier from a leading British manufacturer, previously renowned for state-of-the-art valved designs. Its introduction reflected the very sudden shift world wide from vacuum state electronics to solid-state, at a time when the thermionic valve was still capable of further development and exploitation.
It's easy to understand why amplifiers and hi-fi electronics in general became transistorised so rapidly. Transistors saved costs, and in terms of standard test procedures, measured far better. Total Harmonic Distortion (THD), for example, that still-revered but relatively unrevealing yardstick of performance for hi-fi amplifiers, was invariably lower with transistorised units. Power outputs were far higher than from valved models of equivalent prices. Hi-fi amplifiers had now, it seemed, reached a peak of development, with minimal distortion and unlimited power availability (or so it seemed) to drive the new breed of inefficient speakers.
In the fifteen years or so since the great transistor revolution, the valve-versus transistor argument has constantly been debated. Many audiophiles regret having sold their old Quad 2's, their Fishers, their Scotts, their Leak TLl 2's, Stereo Twenty's and Sixty's and so on, having found the new transistorised designs less satisfying sonically despite vastly superior general specifications (right on! - Ed).
Until fairly recently, many of the problems associated with transistorised amplifiers remained unidentified, and there is little doubt that further problems have still to be identified and overcome. Yet there is no evidence that a transistorised amplifier (even one based on miniature integrated circuits) cannot sound at least as good as the very best valved models. Harold Leak's remark in connection with output transformers is one of the best arguments in favour of transistors, for it seems to be the output transformer more than any other component in a valved power amplifier which dictates performance quality. And good output transformers are very expensive indeed.
If we are to establish why a valved amplifier sounds better than a transistorised amplifier (and we're still far from convinced that this is so) we must look at the technique in both instances. It would appear that conventional methods of test and measurement give little real indication of how a particular component is going to sound, and this is probably because of the enormous difficulty of providing a controlled and quantified test signal simulating the sort of signal derived from musical sounds.
In a practical situation, an amplifier is faced with signals of astounding complexity, vastly different from the far simpler signals derived from test equipment. [See note for a comment on this topic.]
In addition, an amplifier is also faced with a practical loudspeaker for a load, and an investigation of the behaviour of a speaker as seen by the amplifier reveals a situation which often makes the dedicated audiophile want to give up the whole idea of high quality sound reproduction and retire to a desert island! Many amplifiers particularly transistorised ones-can be excused for complaining audibly about the sort of load they are called upon to feed.
And at the amplifier front end, there is another interface problem, that of the source (invariably the pickup cartridge and its need for equalisation and good S/N ratio poses the greatest difficulties) and its effect on the preamp input performance.
There is a lot of evidence to suggest that a good transistorised preamp is superior to a good valved one, not simply in terms of measured performance but in terms of the audible result.
There is no doubt that most valved amplifiers contribute to a categorically different audible result from that of most transistorised amplifiers, in fact an experienced ear can almost invariably judge whether a valved or transistorised model is in use. This points to an influence by design and manufacturing technique on the result - which might seem obvious but which should not happen in theory.
Most high quality valved amplifiers of recent manufacture and design (within, say, the past couple of decades) use the so-called ultra-linear push-pull output stage, which inherently cancels out high-order-harmonic distortion and has a number of other important desirable characteristics.
As its description indicates, this type of output stage employs a pair of valves in which one 'Pushes' and the other 'pulls' - a mechanical analogy being a two handled saw. Both valves operate throughout a complete cycle, swinging from positive to negative as demanded by the input signal. The use of two valves rather than one increases output amplitude to give more power than would be achieved via a single-ended output (quite feasible with valves). In fact many ultra-linear push-pull output stages will produce a useable signal with one of the valves removed, at the expense of power loss and increased distortion! (We don't recommend you try this, incidentally, if you own a valve amplifier).
Most transistor output stages are rather different. Again, a pair of transistors is used, but in so-called complementary or (quasi-complementary) mode. In effect one transistor of the pair handles the nominally positive side of the signal while the other remains dormant. These roles ire reversed with the nominally negative side of the signal - the previously operating transistor handing over work to the formerly dormant transistor.
Some transistor amplifiers operate in a similar fashion to the type of valve units just described, although there are currently severe practical limitations mainly the voltage handling ability of power transistors-resulting in poor efficiency and excessive heat output.
Class A push-pull output stages are inherently distortion-cancelling and each amplifying device, be it valve or transistor, operates in a complementary electrical sense at all times. Deviations of performance from the ideal of either device are compensated by the other. But with so-called Class B stages, one transistor of a pair is constantly switching in and out of operation, handing over to the other transistor during its period of no operation. An obvious difficulty here is to achieve a smooth changeover or 'crossover' (not to be confused with loudspeaker dividing networks), and the main reason for the 'transistor sound' of most early (and unfortunately some present-day) transistorised amplifiers is/was poor crossover performance (where the switch from one transistor to another failed to give perfect signal continuity) leading to a spiky distortion most evident at low Output levels and high frequencies. Many transistor amplifiers actually had better performance at high output levels, quite the reverse of most good quality valved amps.
And then, of course, there is the output transformer. This is used in valved amplifiers to match the valves (which demand a high impedance load) to the loudspeaker, which is a low impedance device.
Transformers are extremely costly and in theory are responsible for a number of serious performance deficiencies including reduction of damping factor, phase shift toward the high frequency end of the spectrum, and, in a practical situation using a complex loudspeaker load and extended upper frequency response instability with consequent oscillation. No wonder Harold Leak was keen to eliminate this component!
On the other hand, the transformer serves as a 'buffer' between loudspeaker and amp proper. This can be advantageous with practical loudspeakers in avoiding the 'ricochet' effect of transistorised amplifiers faced with back-EMF from loudspeakers resulting from unwanted diaphragm motion. This output from the loudspeaker can, in some circumstances, penetrate back to interfere with a signal passing out toward the speaker; a consequence of the inherently low impedance of transistorised amplifying circuits. Valves are inherently high impedance devices, and the effects of back-EMF from the speaker are therefore far less likely to achieve significant penetration at sufficient level to degrade performance. This must not be confused with damping factor per se which is generally rather better in transistorised amplifiers than in valved models.
A further factor in favour of transistorised amplifiers is long-term consistency of performance. Although total failure of transistorised amps seems to be more frequent than with valved units, repairs normally cost less and are quicker. Even so, overall reliability of transistorised amps seems to be better, especially in terms of performance consistency over prolonged periods.
Valves deteriorate at a fairly steady rate, and regular replacement is essential for top performance to be maintained. And, with very high quality amplifiers, this isn't always a matter just of replacing a valve - almost invariably, matched sets are required and it is also wise to optimise grid bias voltage level when valves have been replaced. Regular transistor replacement, on the other hand, is not necessary.
Generally speaking a transistor either works or it doesn't - and only requires replacing if it has an inherent fault or has been abused.
Of course, there are other fundamental differences between valved and transistorised amplifiers, and a study of these shows that a transistorised amplifier should be able to out perform any valved amp. This is certainly true of low-level amplifier stages, transistors generally having better signal-to-noise ratio, better frequency response in terms of overall bandwidth, better linearity across that bandwidth, and greater efficiency.
At the present time, hybrid valved/ transistorised amplifiers would seem to offer the greatest potential for best audible performance-transistors being used for all amplifying stages up to and including the driver stage, and a valved output stage.
In our experience, a top valved amp gives a sweeter, smoother result than a top transistorised amp, but this is a generalisation and we have heard at least one transistorised amplifier, currently being developed by AMW, a local manufacturer, which seems to have none of the audible deficiencies of either type. Nevertheless this (prototype) unit is still unhappy with some complex resistive/ reactive loads presented by multi-way loudspeakers, but the designer and development engineer are confident this problem can be overcome.
So much for technicalities. We've only scratched the technical surface here; there are far more behaviour characteristics of each type of amplifier which account for audible differences despite indications of identical expected performance from standard measurement techniques.
Unfortunately delays in production prevented our reviewing AMW's fascinating new amplifier in this issue. This amplifier, as reported last month, is based on a valve output stage but uses transistor front-end and driver stages.
An interesting feature of the preamp is a high-gain pickup input designed for use with low-output moving coil cartridges, plus of course the usual RIAA-compensated facility for cartridges of higher output. Following latest trends for top performance equipment, the preamp has no tone controls, and is fitted only with volume, balance, input select and tape monitor facilities.
The power amp has an ingenious system for optimising grid bias of each output valve, which contributes to minimum distortions and increased valve life. A pair of light-emitting diodes (LEDs) and a screwdriver-operated potentiometer associated with each valve is fitted to the front panel and adjusting the potentiometer for equal light intensity from both LEDs (of the pair) provides correct grid bias voltage. Level controls are fitted for each channel.
The power amp looks quite a monster and the sight of those enormous KT88 output valves, visible through the gaps in the T-section aluminium cover members is quite nostalgia-invoking.
The preamp, provided with a remote power supply which helps improve signal-to-noise ratio, has, by comparison, an unobtrusive appearance although we understand production models will be slightly different.
Power amplifier performance was excellent, low frequency sounds being particularly well defined and with good extension to extremes. There was no characteristic valve sound, the overall result being essentially neutral and with good definition of fine detail, impressive transient performance and excellent stereo imaging.
The preamp, whilst again having no specific performance deficiencies in terms of tonal balance, tended to sound a shade 'grainy' at times, especially when called upon to deliver high outputs. However this was not unpleasant but rather gave a powerful, 'gutsy' sound which one observer described as giving the amplifier 'balls!' (blush, blush!).
Production is currently in full swing and samples of the amplifier should be available from AMW retailers during December. Rated output of the amp should be in the vicinity of 100 watts per channel, thanks to the use of very carefully wound and efficient output transformers - the prototype was claimed to yield some 60 watts per channel but sounded far more powerful than most amplifiers of this rating.
We're very much looking forward to seeing and hearing production samples of this very fine new Australian product, which upholds the firmly established AMW reputation for fine performance. Price is expected to be around $2,300 for the preamp/power amp combination.
The main electronics of the power amplifier and preamplifier referred to in this article were designed by me - I cannot take credit for the design of the power amp casing nor the specification for the output transformers, as they were the domain of John Burnett. The amp was designed on contract to AMW Acoustic Labs. The phono stage was an early version of the Project 06 preamp presented in the projects section. We designed the amp to be as good as a valve amp could be, which meant that it was a hybrid - part valve and part transistors (as well as early Class-A opamps). At the time, nothing else on the market in Australia came close to the performance achieved, and today it could be even better if reliable valves could be sourced.
Also note that if some of the comments seem a little dated, this article was written in 1977, so they are indeed dated. The basic ideas are still the topic of debate, so although everything changes, it still stays the same (except the price, of course - if you could build one of these for $2,300 today you must have stolen the parts!).
Please note that I have reproduced the text verbatim - I do not necessarily agree with the comments regarding the difference between valve and transistor amplifiers, nor with all the technical matters raised.
The company that originally contracted the design (AMW) went out of business not long after the review was published. For what it's worth, this left John and me well and truly out of pocket - a very expensive exercise indeed.
|Copyright Notice. The sections of the article shown are an excerpt from Australian Hi-Fi Review, November 1977. Not all of the article has been re-published here, as the remainder reviewed other amplifiers. Of the section re-published, this has been done without any alteration to the original text. Reviewers' names have been removed but otherwise all text is verbatim.|