Here is a design for a valve based, stereo audio-amplifier, not using valves types which are expensive to procure because of their esteemed nature in performance and purity of sound for music reproduction, guitar amplifiers etc.
e.g. EL84, 6L6, KT66, EL34, 6550A, 6SN7 types etc which typically command high prices when offered for sale.

A Stereo 17 watt per channel Ultralinear, push-pull Amplifier using 6CM5/EL36 output valves.
The circuit design uses 6CM5/EL36 power pentodes in a push-pull Ultra-linear configuration, with a fixed - 50 volt DC grid bias.
Anode current of each o/p valve is a conservative 25 mA approx, and maximum undistorted output power is about 17 Watt/channel.
The input stage is based on the Mullard 5-20 but I have used a 6AU6 instead of the popular EF86 pentode. (see later)
The phase splitter is of the well known cathode coupled configuration using a 12AX7 twin triode, being as in the LEAK TL 12+ amplifier.

I acknowledge inspiration for my design came heavily from the following web pages;

http://home.alphal ~cambie/EL36.htm#John_Hunter



The 6CM5/EL36 pentode was originally intended for use both as a “line output or horizontal deflection” valve in television receivers, and (somewhat rarely) as a Class B audio power amplifier.
The 6CM5 valve was once plentifully available (as a service spare for B/W TV receivers) and because it is not a highly prized audio tube, by valve collectors and audio buffs (like the EL84, KT66, 6SN7, EF86, 12AX7's etc) you should be able to obtain some cheaply from valve vendors.
The Soviet 6pi31 (6P31) is a direct equivalent, and the well regarded Sveltana branded type can be purchased cheap and in quantity.
6π31S (a.k.a. = 6P31C, 6p31C, 6pi31C) = 6n31C (Analogue of EL36)

The input stage is based on the Mullard 5-20, however instead of the popular EF86 I have used a 6AU6 which is cheap and easy to find.

An interesting comment is sited in: " VALVE AMPLIFIERS: MORGAN JONES. NEWNES (1995) pp224 "

The EF86 pentode has a heater structure as a helix and an electrostatic shield around the anode (mesh) both designed to reduce hum in audio circuits. The EF86 pentode, is responsible for high sensitivity but poor noise performance.

So the humble 6AU6 should be OK.

The screen decoupling capacitor is connected from G2 to cathode (rather than hard to ground) because the cathode is not at AC ground on account of the negative feedback into the cathode. Grounding the screen decoupling cap would produce positive feedback to G2 and hence instability.
I believe negative feedback gives the amplifier a substantially flatter frequency response, than otherwise would be.
Only the DC biassing cathode resistor is decoupled for audio variations by a 22uF electrolytic cap'.
A (12AX7) cathode coupled phase splitter is used, where V1 (first triode) is a voltage amplifier and V2 (second triode) can be considered as a grounded grid amplifier fed from the commoned cathodes of V1 & V2.
This circuit is well proven and is exactly as in the LEAK TL 12+ amplifier.
Unfortunately the 12AX7 is highly revered by audiophiles and some brand names attract unusually high prices when sold at on-line auction.
It is still in current manufacture (Soviets and the Chinese) so look around and you will find some budget priced units, I'm sure.


The total valve filament current requirement is :

Valve Current Total
6CM5 1.25 X 4 5
12AX7 0.45 X 2 0.9
6AU6 0.3 X 2 0.6
ALL 6.5 A

The power transformer I've used was salvaged from a c1980's japanese Color TV as was some winding wire (from de-gaussing coil).
It was more than suitable, because it was fitted with an external copper electrostatic shield and also fully metal clad .
The transformer (E & I laminations) were disassembled, the existing outer 6.3v filament winding removed and new windings installed, to provide enough current to supply the heater current needs of 6.5 amp. The existing mains primary and 110v secondary were retained.
I didn't have any thick enough gauge wire to satisfy the above amperage specification, so I made 8 (hand wound) windings ( yes 8 ! ) of #24 gauge wire producing 3.5 volt each (no load).
These were polarity series'd and paralled (2 x 4) to give an effective 7v centre tapped filament winding (see diagram).
The earthed center tap is good practice as it reduces induced AC hum in an audio amplifier.
In operation with the voltage drop caused by the 6.5 amp load, filament voltage measured was correct at 6.3 Volt.
The winding wire was salved from the de-gaussing coil of the same donor color TV and there was more than enough for the rewind.

I did consider single-point earthing as a hum reduction measure but with tagstrip component mountings each earth lug carries as many earthy connections as practical.
I took particular care to bond the earth lugs of tag strips solidly to the metal chassis . Good construction practice was observed, particularily with filament wiring (see diagram) and also extra care taken with the arrangement of all components so that the underside is aesthetically pleasing !

Because the physical appearance of these "naked" hand crafted amplifiers, is a consideration, as well as their electrical performance;
I have included the OC3 Voltage Regulator tube, which glows purple when operating! The input voltage amplifier is hence fed from 108 volt regulated HT supply.
While this is not strictly needed, rather than have the gas-filled VR tube purely for appearance, I used it to supply stabilized B+ volts for the 6AU6 input stage.


The output transformers are sold as 100v line output speaker line transformers, but astute observation by experimenters has revealed that these transformer have the centre tap and the 43% screen tap (Ultralinear operation), and have an overall primary to secondary ratio of 8000 : 8 ohm

These transformers are sold by Altronics in Australia are very cheap and rated at 15 watt.
They proved ideal for the application.

A well written analysis is available at:
Application of the M1115 100V line, 15W Speaker Transformer to Valve Push-Pull Output
The on-line catalogue listing and pricing of the M1115 transformer is :
Similar transformers may be readily available worldwide , I haven't bothered to check.
8 Ohm to 1.25W - 15W PA Speaker Transformer
15 W 100V line speaker transformer with multi-tapped primary and secondary to suit 8 Ohm speakers.
High quality, high performance 100V line audio transformers.
All models are wound on professional grade grain orientated steel core for extended frequency response.
Designed for use in high grade constant voltage 100V and 70V PA Systems.
Frequency Response: 30Hz - 20kHz ±3dB
Secondary Tap: 8 Ohm
Power taps: 1.25W, 2.5W, 5W, 10W, 15W

To quote from some learned text's :

In a push-pull amplifier (compared to a parallel pair of valves): 2nd Harmonic distortion produced by either valve is cancelled by equal and opposite distortion from the other.
Two triodes in Push-pull therefore give greater undistorted output than if connected in parallel.
3rd harmonic distortion does not cancel in this way; pentodes whose output is limited by 3rd harmonics therefore do not share in this particular advantage.
They can be made to do so by connecting their screen grids to suitable tapping’s on the primary of the output transformer, in what is rather absurdly called the "Ultra-Linear" system.


Ultra-Linear operation provides a compromise between the high efficiency of tetrodes and the low distortion and uncritical load impedance of triodes.
There is an undoubted advantage in modulator service because the matching of the modulator to the RF stage is not so critical and the screen (DC) supply is obtained "automatically" from the transformer.
Screen grid tappings: 43% gives minimum distortion 20% gives maximum power.

RSGB handbook 4th Edition (1968) pp9.19

Patent 496,883 dated 5 June 1937 is one which is considered to be among the most important, which later became known as the "Ultra - Linear" amplifier.
With so many of Blumlein’s circuits the design is deceptively simple.
It shows a pentode output stage of a single ended audio frequency amplifier with a tap on the primary winding of the output transformer.
This tap provided feedback to the second grid to improve the linearity of the amplifier.
Blumlein realized that if the tap was placed at the Anode end of the primary winding, the valve would then be connected as a triode, and if the tap was at the supply end, it would be a pure pentode.
It was well known that when a pentode was connected as a triode it would be far less efficient and therefore provide a lower power output. If however the tap was placed at a distance 15-20% down from the supply end of the supply end of the transformer, the valve would combine positive features of both the triode and the pentode.
The Ultra-Linear circuit in its push-pull form became widely used in high-fidelity valve amplifiers during the second world war, though it was a good many years before general practice caught up with Blumlein’s thinking.
While Blumlein did not consider the problems associated with distortion resulting from feedback to the screen, he regarded it mainly as a convenient alternative to the control-grid feedback for reducing the undesirably high impedance of pentodes while retaining their efficiency.
He did suggest lowering the output impedance of the valve to approximately equal the optimum load. This would have been done in order to prevent damage to the valve or the output transformer if, the load became disconnected.



Final stage fixed-bias allows slightly higher anode-to-cathode voltage for the output valves than using cathode bias.
The -50v bias voltage is obtained by zener regulation from a seperately rectified output from mains transformer secondary.

A 10 Ohm resistor in each 6CM5 cathode earth return is used to monitor cathode current (by Ohms Law the easily measured voltage across this resistor, divided by 10, yields the value of standing cathode current through the valve, this in turn being the sum of principally the anode current plus a lesser amount of screen current) and acts as a fuse under overload conditions.
(Max power of 17W was measured by applying a 1kHz sinewave to the amp and measuring the peak to peak voltage across an 8 Ohm dummy load at the point of clipping. The RMS volts = pp volts divided by 2.88 (twice the square root of 2). Power then = Vrms2/8 Ohm.
The power supply is straightforward; voltage doubler B+ and seperate negative bias supply.
The front panel is uncomplicated; stereo input jack, ganged dual audio gain control, and a 3 position switch OFF, STBY (valve heaters) and RUN pos'n.

In summary I can state that the amplifier gives a good account of itself, certainly loud enough for my needs, and has the rich, distortion free mellow audio, that (as we are told) only vacuum tubes can produce!

Frank Hughes VK6FH December 2008