Melbourne Audio Club

General Meeting Reports for 2025

February 2025

Tim Robbins and Ric Clarke from HRSA

Tim and Ric brought along an array of vintage Williamson amplifiers (valve, of course) to showcase the very great leap in practical amplifier design that Theo Williamson, a Scotsman, achieved in 1947. Tim and Ric's exceptionally entertaining talk was peppered with juicy morsels, including the cover of an electronic magazine of the time nominating the Williamson as "the amplifier to end amplifiers" (how often have we heard that claim made?) and a personal description of the dubious joys of adjusting the bias of an old valve amplifier when the adjustment potentiometer was un-insulated and carried 400 VDC. The two amplifiers on demonstration were fed by a very modern Technics SL-G700M2 streamer provided by the program co-ordinator, Dave Polanske. Dave also provided the Leak sandwich speakers, more of which anon.

In the days just before the meeting I had a couple of colleagues ask me what I thought about the impending demonstration. I explained in broad terms the historical significance of the Williamson amplifier and the novel design of the Leak sandwich speakers, and concluded by suggesting it could be a very interesting session but that I was afraid the system would struggle to fill the Willis Room. After all, the Williamson amplifier was rated at only something like 13-15 watts (on a good day, with a tailwind and a downhill slope) and the Leak speakers, despite being floor-standers, were likely quite inefficient and probably themselves rated at only 15- 20 watts. Having seem modern transistor and Class D amplifiers with power outputs that would rival those of a small nuclear reactor struggle in the cavernous Willis Room, I thought I had good reasons for some trepidation as to how the system would cope.

How wrong I was! The system made GLORIOUS music. (Note that I say 'music', not 'sound'.) Person after person I spoke to during the evening and afterwards, on the weekend, remarked on how magnificent the system portrayed music. One member, who I would describe as an ardent audiophile, said "They aren't hi-fi but they make fabulous music". Another, this time of more antiquated audio and musical interests, reported the set-up had among the best instrumental timbre he had ever heard (and he owns a pair of QUAD 57s, so that's quite some praise). He thought the rendition of the Dean Martin track was remarkable. I was won-over by the Elvis track played just after the coffee break: had "Can't help falling in love" (recorded in 1961 for Elvis' 1962 film Blue Hawaii - and in the film sung not to his lover but to his grandmother!) ever sounded so heart-felt, so emotive, so moving, so musically correct?

So, what was the equipment responsible for such unbridled pleasure? I'll start with the speakers, as I have a long-standing personal connection to the Leak sandwiches. The Leak sandwich speaker saw the light of day in 1961. It was called a 'sandwich' because the speaker cone was a thin (2 mm?) layer of expanded polystyrene foam sandwiched between two very thin layers of aluminium foil. The chassis was made of cast aluminium and the surround was made from cambric fibre. The bass driver was big at 13" and the cabinets themselves were also substantial (60 L) and - of course - finished in period-correct real walnut or teak veneer with a thick cloth grille. The tweeter was also a sandwich design, of 3" diameter, and it crossed over to the woofer at 900 Hz. Since the speakers were designed to be used with a valve amplifier, they had an impedance of 15 ohms. I believe this initial, large, floor-standing version was called the Model 2060.

I was seriously mistaken when I hazarded to guess that the system would be inadequate for the Willis Room. The speakers must have been far more efficient than I had guessed, as they had no problem at all in filling the room. I tried to find what their efficiency was via an online search, but came up with nothing. I now believe it must have been somewhere in the mid-to-high 90s (i.e. dB/watt at 1 m; and remember they were 15 ohm speakers). A very harsh critic might say the top-end was rolled off too - but what can you expect from a 3" cone tweeter with what would now be considered an impossibly heavy cone (i.e. it wasn't built from no beryllium or titanium or diamond wunder-material). In any case, it's likely that few of our members can hear beyond 10-12 kHz, so our hearing and not the drive units is probably the limiting factor. And, yes, the bass driver didn't plumb the depths down to stygian, sub-sonic frequencies: who cares? What the system did do magnificently was portray voices, and particularly male voices. Bravo. Was this because they were being reproduced by a huge (13") woofer that operated solo up to 900 Hz? Maybe.

The sight of these lovely floor-standing, realwood- veneered creatures brought back a flood of memories. About 25 years ago I owned a near-mint pair of their smaller and younger brothers, the Sandwich Model 200. They had a smaller (8") sandwich-cone bass driver and two purple (yes, purple!) 60 mm Mylar drive units, one for the midrange and the other acting as the tweeter. I drove them via a 1960s Japanese amplifier that went by the name of 'Star' which, if recollection serves me correctly, used four EL34 values in push-pull operation. It made a lovely, warm, comfortable sound, perfect for my office. Not a lot of deep bass, not a lot of extreme treble, but a magnificent mid-range and more than satisfactory dynamics. Once you get the mid-range right, who but lovers of heavy metal and other types of doof-doof needs ultra-extended bass or treble?

I say my Leaks were 'near-mint' because the cambric roll surrounds had hardened with age over the then 40+ years of their existence. I took them a speaker repair fellow, based in Northcote I think, and he recommended replacing the original sandwich bass drivers with some hideous, cheap, Chinese rubbish that had bright yellow cones (but obviously were not real B&W kevlar units, "copy-cat yellow" being then the choice of colour for the cones of nearly all drive units). I recoiled in horror at the thought, mentally equating it to the travesty of replacing the 4.2 L straight-six DOHC XK engine in an E-type Jaguar with a pushrod Chevy V8 on the grounds that the latter was more reliable. Barbarian. Instead, in what can be subsequently described only as a fit of madness, I sold the amplifier and speakers. And sold them for a pittance, such stuff being almost unwanted at the time by audiophiles who prided themselves on being totally up to date with all the most modern equipment. Three words describe that action too: Regret. Regret. Regret.

Now onto the Williamson amplifier. It has been described in innumerable reports, of which I have the published paper by Lankshear (1990), the magnificent unpublished paper by Stinson (2020), and the three-part series of books on valve amplifiers by Popovich (2016). Scott Frankland wrote a detailed three-piece analysis in 1996 and 1997 for Stereophile on the history of push-pull amplifiers and their relation to earlier, single-ended, typologies. The first of Frankland's articles (December 1996) described the historical precedents of the Williamson design, its dependence upon the earlier (1934) amplifier design by W.T. Cocking, and how in turn the Williamson became the basis for much further development by other audio designers. The invention of ultralinear operation by David Hafler and Herbert Kereos in 1951-52 was one such development.

The Williamson design first appeared in the April (Part 1) and May (Part 2) editions of Wireless World in 1947, and was updated in the August, October and November 1949 editions. Revisions suggested in the updates included a change from the original four L63 single triodes to the use of two 6SN7 dual triodes in the preamp stage. Interestingly, 6SN7 valves are still used today (my Cary uses them as phase splitters). Popovich (2016, p. 190) argued that "The Williamson's design was not novel even in its day. There is nothing in it that hadn't been seen before, except, perhaps, the triode connection of the [tetrode] output tubes. However, it combined a few clever design choices, resulting in a relatively simple yet (for the time) well performing package." Frankland (1996, p. 115) was slightly kinder, concluding that "Williamson's amplifier enjoyed unprecedented momentum in the marketplace" and "has become the prototype for feedback amplifiers the world over." Lankshear (1990, p. 153) was kinder still, concluding that "The real importance of Williamson's work was that he demonstrated that extremely low distortion was achievable by using plenty of negative feedback, combined with carefully designed output transformers. His design set a standard of performance that is still acceptable today."

Figure 1 shows the circuit diagram for the early version. You will see the use of L63 triodes in the pre-amplifier stage and two KT66 valves in the output stage. (The KT in the valve's name stands for "kinkless tetrode", the kinkless bit being a reference to the shape of the value's performance curve, representing an effective way to circumvent the similar improvement in response recently patented for pentode valves.) 'The rectifier valve was a U52. In summary, it is a four-stage, Class A triode design using deep global negative feedback (20 dB) and a push-pull typology for the output tubes. The fact it was a four-stage design is important because an additional amplification stage was required to recover the input sensitivity lost due to the use of the global feedback. A push-pull typology using triode valves was considered in the 1940s to be the optimal basis for the design of a high-quality audio amplifier (notwithstanding the triode's chief drawback, high input capacitance).

The Williamson amplifier is significant in audio design, and this is because it eliminated the multiple inter-stage transformers that had been widely used in earlier designs and it DCcoupled the first two stages, both innovations being critical in minimising phase shifts. Williamson recognised it was vital to keep phase shifts to a minimum with a push-pull, negative-feedback design, given that the output transformers were integral to the global negative-feedback loop. The transformers therefore had to be of exceptional quality, otherwise they would be responsible for introducing large shifts in phase at frequency extremes. Were these to develop, what was intended to be a global negative feedback loop would quickly morph into a global positive feedback loop. The amplifier would then become a massive oscillator - with disastrous results for your speakers.

The well-known and respected English transformer maker Partridge Transformers Ltd was responsible for building the output transformers. The primary required 4,400 turns, the windings divided into ten primary and eight secondary sections and as by Lankshear (1990) noted "all interleaved into two balanced halves." They must have cost a fortune to make, and the highlight the deep skill-base of the English audio industry immediately after WW2. Negative feedback was optimised at 20 dB, and levels greater than that, Williamson concluded, served little or no useful purpose.

By demonstrating an Australian-made amplifier of the time that used slightly smaller transformers than those developed by Partridge, Tim and Ric showed just how essential high-quality transformers were to the Williamson design. The single most critical component in any valve amplifier is the output transformer, and as Tim and Ric noted, these are the single biggest cost in a valve amplifier, commonly accounting for at least a third of the total. In this situation there will always be a financial incentive to use cheaper transformers, i.e. ones that are smaller, lighter, less complex, or use lower quality wiring or non-grain-orientated steel in the transformer core. Tim and Ric showed plots of output power and phase shift of the Australian model to show how the amplifier with the smaller transformers had a markedly poorer performance than one with the big, expensive Partridges. It was still a good amp, just not as good as the original design would allow had better (i.e. dearer) transformers been used.

The Williamson amplifier is astonishingly significant in audio history because through it Williamson proposed - and then implemented - a set of design specifications that still hold today. Slightly later than its 1947 debut,

Williamson collaborated with Peter Walker (of QUAD fame) on an article published in a 1952 issue of Wireless World that expanded upon these requirements: Total non-linear distortion (i.e. harmonic and intermodulation) should be less than 0.1% at all power outputs (1-2 % was typical for the time) Linear frequency response within the audible spectrum of 10 Hz to 20 kHz Frequency response should be better than -3 dB at 3 Hz and 60 kHz, in order to minimise phase shifts through the audio bandwidth (40 Hz to 10 kHz + 1 dB was typical at the time) Phase shifts within the entire audio bandwidth should be less than 20o, in order to prevent the amplifier becoming an electronic oscillator Good transient response, with a power supply sufficient to accommodate large dynamic peaks in the music Output impedances as low as possible, and always "much less" than the speaker impedance, in order to provide adequate electric damping and limit undesirable peaks in the bass response of the speaker Hum and noise at least 80 dB below the maximum output. They concluded (p. 357) that this was "a formidable specification, and by no means every amplifier styled as "high quality" will meet it."

Nevertheless, specifications as tight as these were required because of the very great advances in recording quality that had been made in the late 1940s. An example is the introduction in June 1948 by Columbia Records of the 33 1/3 rpm microgroove LP record; until then, far looser specifications for frequency range etc were acceptable as they were sufficient for the shellac 78s and (AM) radio broadcasts that then made up all the program material. Greg Milner, in his 2009 book Perfecting sound forever: the story of recorded music, showed the degree to which recording processes had been improved after WW2 (e.g. as seen in Decca's FFRR records), microphones and speakers had become much better as a result of wartime technical developments in sonar etc, FM radio broadcasts (invented in 1933 but first provided in 1948 in New Jersey), and tape recorders using high-frequency bias (a wartime invention in Nazi Germany by AEG) were just coming onto the market (e.g. the Americanmade Ampex Model 200 in 1948). What an exciting period it must have been for those interested in music reproduction in the late 1940s and early 1950s!

To conclude - we heard last month an amplifier that was designed in 1947, teamed with speakers that first saw the light of day in 1961. In other words, an amplifier from eight decades ago and speakers from six decades ago. What glorious music they made, and made on track after track after track, regardless of genre or period of recording. What perfect, unbridled pleasure they provided. And the track that stood out for me - Elvis' "Can't help falling in love" - was recorded back in 1961 too. In other words, at the same time the Leak speakers were introduced and thus also over six decades ago, recorded using valve microphones and mono valve tape recorders and valve mixing desks etc, etc, etc. Yet we are told relentlessly by audiophile manufacturers that "new is best", that the most recent amplifier and speaker designs are light years ahead of what was claimed as first-class only a few months ago, that only modern 196 kHz 24-bit recordings will do as sources, that we need at least 24, preferably 36, speakers in our living/music/ theatre room to obtain the best sound, and this must include at least six sub-woofers.

Rubbish to all that self-serving baloney. The Williamson amplifier/Leak speaker combination is a superb corrective to the debilitating audiophile disease of upgradism. It makes beautiful music and what our hobby is about is music, not which amplifier has 0.000000001% total harmonic distortion at 2 Hz at a rated output of 3 kW per channel. It provides a remarkable antidote to the Cult of the New. Thus my new credo: "Long live audio 'anachronisms'!"