Griffon 6AS7G

Griffon 6AS7G

Home 6L6 Acousticals Griffon 20/20 Griffon 10/10-SE Griffon 6AS7G SRPP Pre-amp Line stage All-valve Theremin

 

Up

Background

I designed and built these particular amplifiers shortly after my original monoblocs. I had become fascinated with 6AS7G valves at this time, as these had two power triode sections in a single envelope. In addition to this unusual construction, they had (bearing in mind that they were designed for use in regulated power supplies) a phenomenally low anode impedance of about 280 Ohms. This meant that with a standard 2K5 Ohm load, a 6AS7G-based output stage would have an unbelievably low damping factor without the need of any kind of negative feedback loop. The only drawback to basing an amplifier around this valve was that extremely high gain was needed from the driver stage (around 250V P-P).

Since I had already enjoyed reasonable success with my 1:2+2 interstage transformers, I decided to use these again. I also chose to showcase the use of a 6AS7G in a more conventional form - as the power valve in a regulated power supply. Thus my 6AS7G monoblocs would feature two 6AS7Gs: the first in the power supply, and the second as a 'push-pull output stage in a single envelope'. The casework would then draw on my experiences with machined stainless steel.

Ironically, this had been  my original concept for a clever - but inherently simple - domestic amplifier. However, I had run into the problem of how to drive the very insensitive 6080 in push-pull, and switched my attention to 6L6s running as 'pseudo-triodes'. Now that I had a great, high-gain interstage transformer, I could see how to make it work.

 

Circuit description

The 6080 is a high-powered dual triode intended as a series pass valve in regulated power supplies. It has a phenomenally low anode impedance, meaning that a few 6080s in parallel can drive a loudspeaker load directly without the need for a transformer. However, this low anode impedance is bought at the cost of very low sensitivity: the 6080 has an amplification factor of two, and driving both halves of a 6080 in push-pull means a peak-to-peak driving voltage of 250V!

My way around this is through the use of a step-up phase-splitting transformer of a 1:2+2 ratio. Even a modest 6J5 with its amplification factor of 20 will drive this, as it only needs a drive voltage of about 3V for full power output. The 6080 itself can be driven to a little over 10W with both halves in push-pull, and I have further refined the circuit by using a second 6080 as part of a regulated power supply.

The above picture (left) is of the finished version of the amplifier. Note the laser-cut stainless steel chassis, and the matching transformer cover for the interstage transformer. The other transformers and inductors are standard Hammond items. The picture to the above right is of the breadboard I constructed to see if my idea of a driver stage would work. As can be seen from the pictures, I am currently using the 'coke bottle' version of the 6080 - namely the 6AS7G - as these are still in production in Russia. The Svetlana-made 6AS7Gs are particularly sweet-sounding, as my later listening tests indicate.

 

Audio circuit diagram

R1, R4, R5 900R, 2W C1 50uF, 300V
R2 50K, 1W C2, C4 100uF, 150V
R4 150K, 1/2W C3 1uF, 400V
R6, R7 1K2, 7W C5, C6 100uF, 150V
T1 10K, 1:2+2 V1 2×6J5 or 1×6SN7
T2 5K, 15W V2 6080

The Hammond 1615, the Sowter U002, and the Maplin "Millennium" output transformers are suitable for T2. The interstage, T1, is expected to be the Booth B41328, and so a custom Sowter would be needed, and Hammond do not currently make them.

 

Power supply circuit

Since the power supply is adjustable, different valves can be used (GZ34, 5AR4, 5V4G, GZ33 - all are possible substitutes). You could even go solid-state, just as long as the regulation circuit gets the ~350V it needs to work from. Ideally, the circuit should be given time to warm up before the HT is applied, so a delay circuit, thermal delay valve, or just a switch would be a good idea!

 

R1 68K, 1W V1 85A2 (5651A)
R2 1M, 1/2W V2 6SL7 (ECC83)
R3, R7 12K, 2W V3 6AS7G (6080)
R4, R5 1/2M, 1/2W V4 See below
R6 15K, 2W L1 10H, 180mA choke
C1 0.1uF, 400V T1 See below
C2 220uf, 450V VR1 10K linear variable resistor
C3 16uF, 500V

 

In the above diagram, both sections of the 6SL7 are shown separately. The 6080/6AS7G is shown as a single triode, but in actuality, both halves need to be used in parallel.

Please note that the centre of the 300-0-300V winding needs to be earthed for the HT to work. Ideally, a 150mA fuse should be used for this to protect the transformer winding in case a valve fails, and a switch or automatic relay should also be used to manually delay the HT (i.e. the switch should be "off" until the 6080/6AS7G heaters are fully lit, then the circuit may be made - use a plastic one for safety that can handle 440VAC).

The mains transformer, T1, needs to supply:

bullet300-0-300V @ 150mA
bullet5V @ 3A
bullet6.3V @ 5A
bullet6.3V @ 1A

The 6.3V @ 5A winding should be left floating and not be earthed, as it is needed for the heaters of both the power supply and audio 6080/6AS7G. As it is unlikely that a mains transformer has just these windings (a custom design from Sowter could do), I would suggest using a second transformer just to handle the 6.3V @ 1A for the small signal valves (the 6SL7 and two 6J5s/one 6SN7). This second winding should have either a side or a centre-tap earthed to reduce hum.

Suitable transformers include the Hammond 372FX mains transformer (used together with the 166J6 low-voltage transformer) and the Sowter M038N (it has a second 6.3V @ 2A winding). Suitable chokes for L1 are the Hammond 193D and the Sowter CA15.

Due to the adjustability of this power supply, the possibility of using any of the following rectifier valves: 5U4GB, 5AR4, 5R4, 5V4, GZ32, GZ33, GZ34. As it is set for a given output voltage by means of the variable resistor VR1, changing the rectifier valve should not change anything. VR1 is, if anything, permanent, but is included by me - alongside a built-in voltmeter - to experiment with different voltages. I will confirm my list of possible alternative rectifiers and give the values of a pair of fixed resistors to replace VR1 (and thus fix the output voltage of this power supply at 250V) in due course.

 

A note on components

Since I'm using valves, I'm planning on a full set of matching transformers. Hammond have a good range - recently available in the UK. I'm going for a "300" series mains transformer, a choke, and a 15W 5K output transformer, per channel. These are a very cheap option. Alternatively, you could buy from Audio Note, Sowter, Woodside Electronics, or even use the Maplins set intended for their "Millennium" amplifier (the 350V high power transformer needs a solid-state bridge rectifier, but handles all the current you need, and the matching 6K6 20W ultra-linear is close enough as an output transformer). I'm using my own interstage transformer for the phase-splitting, but Audio Note and Sowter have equivalents in their ranges.

The most noticeable valves I'm using are Svetlana 6AS7Gs for both the power output double-triodes and the regulated power supply. I'm choosing them for both quality and looks, but I'm initially going to use my amps with my old Brimar 6080 pulls. The 6080 is a rugged version of the 6AS7G, looking very much like a 6550. They're both common types, usually cheap as they're not considered "premium" audio ones like 2A3s or 6L6GCs, and are plentiful as both NOS and current manufacture. You could spend heavily on Tung-Sol 6AS7GAs, or similar types like the 5998 - which has twice the gain - or the impressive Western Electric 421A (note that you'd need different cathode bias resistors for these!), but I'm working with what I've got.

The rest of the circuit calls for a 6SL7, a 6SN7, a standard voltage reference, and a suitable rectifier. 6SL7s and 6SN7s are both in current production, as well as being plentiful as NOS, so which ones you use here are purely up to you. Since only the 6SN7 is used for audio, this is the only one you'd need to worry about. I'm going to use what I've got about - any hobbyist worth his or her salt should have some somewhere! You might prefer to use a pair of 6J5s instead, or even 6C5s, the basic circuit is identical for any of these. The rectifier will hopefully be a 5U4GB, again a standard type, but if the circuit works out, you should be able to choose from a variety e.g. a 5V4 or 5AR4 should work as well.

The voltage reference I'm using calls for a standard 5651A / 85A2, which is a very common type. I'll probably use up a pair of older 85A1s, very similar, but a locktal version, but I suggest that anyone building this chooses the easier-to-find, improved 5651A, which will be cheap if you don't already have some.

As for the casework, I'm planning on using two monobloc chassis here made by Radshape. Just like I did with my 6L6 amps, I'm ordering a pair of custom ones made in stainless steel. This time around, I'm trying out a few different ideas. You could make your own using Maplins chassis, or even buy in some of the Hammond ones. This is one of those "how skilled are you" things - I'm not, so I'm prepared to pay to have mine made for me. If you don't mind paying for precision work, then you might like to ask a local sheet-metal worker for a quote.

The rest of the components are fairly standard - valve sockets, plugs, leads, resistors, capacitors. These are all things you can get from Maplins and RS, or buy in premium types. Just keep to the ratings in my circuits. I'll probably be using a mixture of modern, ex-equipment, and NOS parts to keep costs down here, but I'm deliberately going for things that you can buy over the counter. And, as always, there's nothing to stop you for going for premium parts, e.g. gold-plated ceramic sockets instead of my 8-pin relays, paper-in-oil capacitors instead of my polypropylenes, or tantalum resistors instead of my carbons.

 

Assembly notes

To begin with, the stainless steel case needed to be thoroughly washed with a detergent. Although the outside was beautifully polished, the inside was coated in soot and burned plastic from the laser cutting! Once the case was dry, the transformers were bolted on and the various connectors fixed to it. The next step was the slow creation of the wiring loom, taking care to run signal cables well away from the heater strings. Note the central heater transformer: this is a small unit that provides either 12V or 6V to the 6SL7 and 6SN7/6J5 valves, giving somewhat more flexibility in the choice of these. 

After the last cable run was hooked up, the first resistors and capacitors were added. Most of the power supply control circuit can be seen to be mounted on a breadboard above the rectifier valve. The two bias resistors for the audio 6AS7G/6080 had to be carefully fixed so that there was enough space for a good air flow, otherwise there would be too much heat in too confined a space. It is possible to add bias capacitors as well, but I have found the sound quality to be better if they are omitted. 

 

The finished product!

At long last, all of this work has resulted in a matching pair of monobloc amplifiers, capable of giving 10W of Class A power and capable of using a variety of rectifier valves (what's to hand, really).

The sound quality is very "tubey", as my American friends might say. The treble is relaxed and non-intrusive. It sounds somewhat dull with my Quad ESL-57s, but for that is an excellent partner for my Tannoys, which have a tendency to sound harsh and strident. The midrange is smooth and liquid, perhaps not as detailed on the ESLs as that of the Quad IIs (an unfair comparison given the legendary Quad synergy), but the bass was both rich and taut. On the Tannoys, this bass was shown to go as deep as the music called for. Overall, I must say that I preferred these amplifiers with my Tannoys, so clearly they're better matched with horn-loaded loudspeakers with large bass units. I fear the problems with the ESLs may have stemmed more from the large capacitive load interfering with the high frequency response, and a tetrode heavy feedback amplifier would be far preferable!