One of the things that really annoys me about modern valve amplifiers, be they kits or factory-made, is that designers today don't like taking risks. As often as not, their circuits are classics. There's nothing wrong with this per se, as the Williamson, GEC, and Mullard-type circuits are well thought out and straightforward to build, but these are just the tip of the iceberg. It is indeed ironic that one of the most popular vintage amplifiers in the UK - the Quad II - features a rare and clever circuit, and, despite there being a wealth of knowledge about Quad IIs on the Internet, is rarely imitated by hobbyists (the exceptions to this rule are frequently icons of DIY design, but more of this later!).

As you've doubtlessly guessed, this circuit is a particular favourite of mine, and the design I'm going to walk you through here, has been inspired by not only the Quad II, but the Wireless World tradition of simple, elegant "Quality Amplifiers".

These two monoblocs may be considered to be the predecessors to my Griffon 20/20 amplifiers. Both monoblocs are, indeed, identical, and served as the test bed for my ideas on transformer-coupling and acoustical output stages. I've learned a lot from these over the years, and therefore my later amplifiers embody what seems to be the best I have achieved.

For example, the sheer rigidity and beauty of the polished stainless steel chassis here made me set myself the tough standard of complete, polished stainless steel chassis from then on. The impressive quality of my own-design output transformers and coupling transformers led me to keep the basic design, but tweak it slightly (to give solder tags and good loudspeaker matching possibilities).

The basic circuit is fairly simple, and just involves a lone 6J5 triode as the driver stage and a pair of 6L6s as the output stage. The power supply is a conventional valve-rectified pi-filter, but with an additional VR150 stabiliser neon for the driver stage.

Overview

Transformers are the heart and soul of this amplifier. The output transformer, for example, features a distributed load arrangement, in which the standard 5K primary winding is split 80%/20% between the cathodes and the anodes of the output valves. 

In addition to this, I've chosen to use what is called an interstage transformer to both split the phase (drive the output valves in push-pull) and couple the driver valve (a 6J5) to the 807s. Needless to say, these transformers are my own design, and have been lovingly, wonderfully wound for me by a firm by the name of Booth Transformers, of Burton-on-Trent, England.

Already it can be seen that I've made some detours from the original Quad design (see below). For a start, I've avoided using a global feedback loop - the local feedback inherent to the output stage should be quite sufficient. I've also chosen to replace the pair of EF86 driver/phase-splitter valves with my transformer-coupled 6J5, and gone for a pair of Russian 5881s instead of the Quad's famous KT66s. I've also taken the opportunity to clean up the power supply somewhat, adding another filter stage to the circuit (i.e. a GZ34 running into a 16uF-10H-100uF arrangement for the output stage, then with a second 10H choke and a 220uF capacitor to add extra filtering for the screen grids, and finally a VR150 to provide a rock-solid 150V for the 6J5 driver). 

So why am I doing this? Well, for a start I side with the single-ended triode enthusiasts who disdain heavy negative feedback and valve phase-splitters. This amplifier is therefore a statement of how I believe these may be avoided in creating a push-pull amplifier, and intend to demonstrate in the only way possible that the "single-ended triode sound" can be achieved by other means. It should go without saying that nothing but the best should do regarding such crucial choices as the audio transformers and power supply components.

Choosing the parts

Okay, I'm sure that people are going to be a bit put off by my using custom-wound transformers here. Don't be. Not only should you be able to buy off-the-shelf alternatives from the likes of Sowter, but you should never be afraid to ask a transformer winder to make you something special. You'll be surprised at how many companies will make a one-off for the same price as a transformer from one of its standard lines. If you have any difficulty finding suitable transformers or transformer winders, just e-mail me - this is one of my ongoing crusades!

The rest of the parts are fairly standard, and you should be able to get most of these from a mail order electronics catalogue. Make sure that you choose resistors and capacitors of the right ratings, of course, but don't feel that you need to spend inordinately much. Although you might prefer tantalum resistors and paper-in-oil capacitors, I've deliberately used standard wirewound resistors and polypropylene or electrolytic capacitors in these designs. Again, there's nothing to stop you using solid silver hookup wire, although I've settled for heavy copper.

The valves themselves should be easy enough to locate. I'm using a set of NOS Brimar 6J5s as the driver valves, and doubt that they are greatly different from any other make. 6J5s are standard triodes, after all, and there are some nice ones available from Russia or China if you prefer modern production. Alternatively, you could always use a single 6SN7 valve for both channels (a 6SN7 being essentially two 6J5s in a single envelope). I chose Russian 5881s as the power valves, as these are a variant of the popular 6L6GC. You can build this circuit with either 1625s, 807s, 6L6GCs, or KT66s without changing anything other than the heater windings. I'm using a VR150 because I'm rather fond of neon stabiliser valves, but you could always substitute the smaller 0A2, or even recalculate the dropping resistor to use a decoupling capacitor instead (I've marked the necessary changes on the circuit diagram).

Building a case may be a little trickier, as this requires some metalworking skills. Bear in mind that you'll need something heavy enough to take all of the transformers, but still be light enough to work by hand. Hammond, for example, sells standard cases for DIY use. I chose to go down a different route by having a local metalworking firm laser-cut a pair of stainless steel chassis to my specification (stainless steel looks wonderful and is supremely tough, but is nigh on impossible to work at home!).These would be 2mm thick, laser cut and welded. 

Although these added to the overall cost, I felt that they gave that extra edge of quality. In America, it would be standard practice to have a top plate laser cut (as I did for my Theremin), and then mount this on a wooden frame. However, Birmingham is a city of machine tools, not forests - a metal chassis would be much easier to fabricate. I carefully laid out the transformers to minimize hum and keep the signal paths short. To keep things easy for Radshape to manufacture, I gave them the hole sizes as co-ordinates - having learnt the hard way from the Theremin chassis that diagrams were easy to misread. These would now be micrometer precise.

Aside from the obvious mounting holes for the transformers, there needed to be holes for the valve bases (six in all - two power valves, one signal, one rectifier, two neons), cable runs, and for the other parts. The finished result was better than I'd expected - with any luck, the amps should look truly professional!

Building the First Channel

I managed to follow the above more or less in building the first channel. Sourcing some components proved to be problematic (Maplins had without warning withdrawn a lot of their standard stock!). Assembly wasn't much harder than bolting together the sockets, transformers and connectors. The actual soldering was fairly straightforward. The initial amp probably wasn't so great a layout (although hum was inaudible), but I did need to work out how to integrate the components. The second one will have sub-assembly boards to be connected in place.

Aside from the obvious mounting holes for the transformers, there needed to be holes for the valve bases (six in all - two power valves, one signal, one rectifier, two neons - the second neon being used for the screen grids in an experiment with 6AV5s), cable runs, and for the other parts. The finished result was better than I'd expected - with any luck, the amps should look truly professional!

The valve sockets were Russian-made and bought from RS - I found these to be not only cheaper than McMurdos (or ceramic Chinese) but of a superb build quality. I chose to use solid-core "Bell wire" from Maplins for the internal wiring. Since most of the wiring would be exposed, I could have something tough and colour-coded. Signal carrying cables would be high-quality shielded audio. The speaker binding posts would accept both bare cable and banana plugs, and a gold-plated teflon-insulated phono socket would add that touch of class. The power socket would be an old-fashioned Bulgin "kettle plug", like that used in the Quad ESLs (and most of the lab equipment at Aston!). All the bolts and screws would be standard M5 or 4BA. One little inclusion was to be a long binding post for the earth wires - I planned on neatly crimping earth cables in a star-earth arrangement.

The heavy stainless steel chassis and a good few bolts both looked good and ensured the transformers were fixed in as stable a way as possible.

Assembly

I prefer to start building an amplifier by fixing all of the hardware onto the chassis. In this instance, I've begun by bolting the various valve bases, sockets, and switches onto the chassis. My next step is to bolt the transformers in place, using a cardboard box to help prop the underside up for easy access.

Now that everything is in place, I'm ready to begin soldering the components in place. I prefer to start with the various cable runs here - note that I am carefully twisting the heater cables together to minimise hum, and am careful to keep signal cables clear of these and the HT lines.

I personally find it easier to use a bus bar as the earth: one end of this thick copper wire runs to the external earth connector, the other is connected to both of the input sockets. A good rule to follow with an earth bus, is to solder all of the earth connections from a single stage (grid leak, bias, decoupling capacitor) to a single spot on the bar. Similarly, keep high-current earthing points like the output stage close to the earth connector and as far away from the earlier audio stages as possible.

Feel free to use as many tag boards as you like in building an amplifier. They will simplify the wiring and make it easier to replace components. Given that this particular amplifier has next to no components (a very unusual case!), I am making the unusual move of wiring components directly into the circuit (this is called point-to-point). In a conventional amplifier, this approach would soon lead to a rat's nest of wiring!

The power supply

Given that the power supply is extremely simple, I am not bothering with drawing out a circuit diagram for the moment. It can be envisaged in a very linear fashion:

The idea here is to provide a heavily-filtered power supply for each monobloc amplifier, with additional filtering or stabilisation being used for the more critical areas - the screen grids of the output stage and the 6J5 driver valve. 

The finished amplifier

The finished monoblocs are somewhat rough and ready in appearance (with people tactfully calling them "British industrial"!), but nevertheless demonstrate the circuit very well. The bass is taut, wooden, and extended; the midrange and treble are liquid and detailed. All in all, this '6L6 Acoustical' design is a success!

So far, these monoblocs have gone head-to-head with the original Quad IIs, a Pye Mozart, a pair of EMIs using KT66s, a modern Yamaha MOSFET home cinema system, and even a pair of Krells - all over loudspeakers ranging from B&W bookshelves to Tannoy Monitor Golds, Klipschorns, and Quad ESL-57s. 

In my home system, I certainly prefer these amplifiers to my trusty Quad IIs. Although the Quad IIs are magical amplifiers, particularly when partnered with the original Quad ESL-57 electrostatic loudspeakers, my own-design monoblocs appear to have more detail and atmosphere. For some reason, smaller speakers, such as bookshelves, appear to sound much 'bigger'. Most noticeably, a 'wooden', natural-sounding taut bass has become noticeable as very characteristic of these monoblocs.

Unusually, these monoblocs appear to be extremely sensitive with regard to pre-amplifiers and cable runs. From my own experiences, I would recommend using either a basic low-impedance 'buffer' pre-amplifier (such as a cathode follower) if not a passive volume control. 

The 6L6 Acousticals and the Griffon 20/20

Given that I had first designed these amplifiers several years ago, and had spent a couple of years tweaking the circuits, I felt that the time was right to revisit this project, and build a 'production version' of these amplifiers. My goal was to create something that not only embodied all of the improvements I had tried on this circuit, but would - unlike the original monoblocs - be built to a far higher standard and take the original concept to its ultimate conclusion.