A personal route of amplifier design.
My findings along the road to where I am now!
Available in print in Common Ground Issue 1
Over the past 100 years the realm of cutting edge electronics has inevitably passed from the individual to the corporate world. Most of us now live with many pieces of equipment without which life would be very different. For almost all of these devises we only have a passing understanding, at best, of how they function.
For many of us this is not really a problem, why worry about the hows and wherefores, all that matters is the job it does. (I shall own up now and say that I think like this about most things, but this always leads to major problems when I want to buy something for a specific task.) We sit and ponder, while reading the sales literature, attempting to disseminate some resemblance of sense out of all the psuedo-techno babble that passes for information in a vain attempt to decide just what it is we do want. Some technical knowledge doesnt help either, after all the guy that wrote the sales blurb doesnt know anything either, and the sure wont let the designer write it because he may tell you something useful, or honest. So what do we do? Forget it and buy on price, the easiest and most common route out, and for items that you will have no emotional attachment to, why not? After all I dont find one brand of washing machine inherently more fulfilling than another, do you?
If a piece of equipment is supposed to induce an emotional reaction, what then? The second option is to buy some magazines and try and read up, in what is almost certainly a vain attempt to discover something useful. Most magazines want to tell a good story, rather than provide any information that may be of real use. The writers primary task is to sell the magazine to the advertisers and public. The feeling you get is one of insecurity. I dont have a Dominator X1", or mine isnt that fast etc. because new is always better. By retaining this feeling of insecurity we continue to buy the magazine, and never really feel relaxed about what we have bought.
So what other options are there? Well to be honest I dont know much about washing machines, I know very little about cars, and only a bit more about telephones. Since I have no emotional attachment to these machines, I don't much care about them either, but I do have a love of music, and a background in electronics and acoustics, so I tend to have a raging obsession with building my own audio equipment.
I have no real desire to write something extolling the virtues of one of my designs, telling you all other designers are deaf or stupid, somebody else will do that. Nor do I wish to bamboozle you with mathematics, electronics or tales of black magic in the name of audio Nirvana. All I wish to do is take a general walk through some of the ideas and processes I explored over the past few years with my triode amplifiers. I shall not be attempting to prove the superiority of this format, those who fail to see this will never accept it, the rest of us already know it.
Around four years ago, as Single Ended Triode amplifiers were beginning to appear, I set about developing my first SET power amplifier. Up until that point I had been using some PP 2A3 amps or PP EL84 amps. Phono stages and Pre-amplifiers had always consumed my time without me picking up any real interest in power amps. For several reasons I chose a parallel 2A3 amp as the starting point for a power amp. The main reasons were ones of power, 7 Watts was the minimum power output I could live with for my loudspeakers, and power valve considerations, 2A3s from China being cheap and reasonably reliable while the 300B had only recently started to appear with mixed reports. The question then arose, where to start? Like many SE virgins I begun by building an existing published circuit design, learning about it along the way, then using it as a springboard for more research and experimentation.
The first amp built was a straight copy of the Audio Note Neiro. SRPP 6072A input stage, parallel cathode follower 5687 driver stage and fixed bias parallel 2A3 output stage. Hopefully most of you have got a copy of the circuit, if not it is available from Audio Note (UK), or any of their representatives. The first change I had to make was to dc heat the output valves, rather than ac heat, due to hum problems. This presented the first important change. DC heating sounds much worse! With a directly heated triode the filament is the cathode, so anything in the filament supply is a primary part of the audio circuit. The best solution I have found so far is to use the best components possible in this area, in an effort to reduce the negative effects of electrolytic capacitors and silicon rectifiers. Thus I now use Schottky rectifiers and Elna or Black Gate capacitors, anything else is clearly audible as a grey metallic edge to the music.
How did it sound? To be honest it was a major revelation after all those PP amps. It sounded sweet, large, dynamic, effortless and utterly musical. It remained untouched for a couple of months., which in my system is a major compliment. During this time I learned a lot more about driving the output tubes, about cathode followers, about SRPP etc. My major problem was that I really didnt like cathode followers. I was trying to eliminate all forms of feedback, and sat there in the middle of my power amplifier was a stage that was 100% feedback. So I took it out. In theory the resulting amp should have sounded awful, there is no way the SRPP 6072A, with its significantly higher output impedance and much lower power, should have been able to drive the parallel 2A3s. The amp sounded much much better. Life and dynamics increased enormously, while a lot of fog was removed and clarity increased hugely. This was not a subtle improvement! It must be said, however, that the lower drive capabilities of the 6072A did cause the weight and scale of the presentation to diminish. It was after this experience that the magic spell of this circuit being by Kondo San was broken. It was time to really go to town! The goal now was to work towards a position with the perceived weight and scale of the original circuit, without closing the window I had just opened.
The 6072A was replaced by an SRPP 5687, plenty of drive but a mixed sonic result. Stronger sound, but without the finesse or joy that I had previously. Next up was a 6072A SRPP input stage capacitor coupled to a 5687 driver stage. And finally I re-configured the SRPP to a plate loaded anode follower. I have never used SRPP since! This version of the amplifier is shown in Figure 1. I had achieved what I had wanted, the only penalty was a higher input sensitivity.
This was a very good sounding amp. All the easy relaxed feel and large scale presentation of the original but with greatly enhanced clarity, focus and depth of image. It also illustrates two of the other major improvements that were set up during the above period. The first was a separate rectifier and power supply for the input and driver stage. This brought about great improvement in clarity and composure under difficult signal conditions. The second was much more important; a tube rectifier for the negative bias supply. Wow! Getting rid of that sand was just the ticket. Everything improved; and I mean everything.
Various different input and driver valves have also been tried during the life of this amp: ECC88, ECC82, E182CC...; but nothing seems to have the same balance of the 6072A & 5687 combination. The 6072A appears to have a highly analytical clear presentation, which when married to a similar driver tube, such as E182CC, produces an overly forward presentation. Likewise the rather foggy presentation of the ECC82 when coupled to the slightly dark 5687 produces a rather sat upon sound. Thus the tube complement has stayed the same overall.
The audio circuit design was nearing completion, the only major change remaining was switching to direct coupling between the input and driver stage. The elimination of the coupling capacitor, and associated low frequency corner, improved speed, clarity and dynamics to give a more direct sound. Attention was then redirected towards the power supplies. For some time I had been using a choke smoothed power supply. This involves having an input capacitor after the rectifier, followed by an inductor and capacitor filter to remove ripple from the power supply. The inductor is useful because for a given resistance it will help you achieve a lower level of noise on the power supply, a very beneficial effect. However having an inductor filtered power supply, as most are, does not bring sonic heaven, it is not the be all and end all of a good amplifier. I had been looking into using a choke input filter, where there is no first capacitor after the rectifier, and believed that there were good reasons for trying it. When this was finally implemented, on the HT supply for the output valves, it came as a major revelation. The solidity and dynamics of the music was very impressive. The sense of grip and poise was something I had never heard before from SE amps, or any valve amplifier come to mention it. A real, deep, strong, controlled bass was emanating from my speakers. SE amplifiers are always compromised by soft, woolly, bass and poor control, every review I have read says this, even from people who like SE amps! Choke input power supplies are the means to improve this area. The choke input filter was then quickly applied to the bias supply with similar results.
This is basically the stage at which the Arcadia amplifiers were developed. Each utilises separate HTs for the two output stages, with a shared bias supply and input stage supply. Component quality is good, with Holco resistors, 1% Wire Wound for power resistors, Hovland Musicaps, Black Gates, BHC ALS20 series HT caps, silver wire for signal, ht and ground wires, copper chassis and PTFE sheet for hardwiring all the components on. The result is a very good balance of dynamics, control, power, scale, clarity and long term listening pleasure. Figure 2
More recently a single SE 300B project has developed along a completely different route. It started far more recently and was based upon different design ideals.
The first 300B amp built was a simple stereo unit, and aimed to be a version of the classic three stage 6SN7-6SN7-300B combination. Versions of this amplifier have been around in Japan for many years. This circuit offers one of the least expensive means of getting a 300B amp on the road. The 6SN7 allows for good levels of drive to the 300B, about the correct level of overall gain and elegant direct coupling between the first two stages is possible. This amp is currently in the form shown in Figure 3. This amplifier offers a very competent level of performance. For the relatively low cost and ease of construction it is an excellent project. A similar approach in a commercial product would most probably cost a few thousand pounds, and still offer good sound per pound.
Another similar version was tested which utilised a parallel 6SL7 anode follower, in a two stage circuit. This was less direct, clear and dynamic, as well as having a reduced high frequency bandwidth. The reduced bandwidth was almost certainly due to an inability to drive the 300B grid as effectively as the 6SN7, whilst on several occasions I have witnessed a marked reduction in clarity and focus when paralleled stages are used.
The second 300B SE amplifier takes a radical departure from standard designs. For many years Japanese transformer companies such as Tango and Tamura have been building Interstage Driver Transformers. Up until the past 12 months they have had very little impact upon amplifier design in Europe and the USA. Recently the work of Nobu Shishido, and Susumu Sakuma, has received far more attention from SET amplifier builders in the west. These Japanese designers have been using, transformer coupled amplifiers almost exclusively for the past 15 to 20 years. Both designers talk of the organic quality that music takes on when transformer coupling is utilised, with a much greater level of emotional involvement, and a less processed presentation.
Spurred on by these ideas I set about building a single 300B amplifier utilising an interstage transformer to drive the grid of the 300B. The results of these works can be seen in Figure 4. Here a 5842 (or 417A) low impedance, high transconductance, medium mu single triode is used with a 1:1 ratio interstage driver transformer to power the grid of a fixed bias 300B. The gain of the 5842 being around 45 times means that the power amplifier has a sensitivity of around 1.2Vrms to full output. This is much higher than many amplifiers such as Leaks or Audio Innovations, but not really a problem if an active line stage pre-amplifier is used. The simplicity of this two stage transformer coupled approach leads to a direct and emotional result quite unmatched by any other method. Resolution of low level information is exemplary, with the whole presentation having an organic, complete feeling that quite belies the electronics of the reproduction chain.
Later versions have carried this design to its logical conclusion. Separate power supplies for the driver and output stages, and the adoption of fixed bias have resulted in greater control, clarity, dynamics and scale. Figure 5 The improvements in musicality brought about by transformer coupling has lead to work on transformer couple line stage preamplifiers, and transformer inputs, but more of that at a later date.
These HTML documents, and associated graphics, are Copyright 1996, SJS
Electroacoustics. These documents may not be reproduced in any manner without the
permission of the copyright owner. The schematics presented here are Copyright 1996, SJS
Electroacoustics and are offered for construction on a non-profit basis only. Any persons
wishing to profit from these designs, or requiring similar should contact SJS
Electroacoustics for licensing. We invite the audio and music community to link to this
home page, which will be periodically revised.
This document has been revised and updated April 4, 1999.