Caillan Taylor: A 35WPC stereo amplifier

A few weeks ago I was lucky enough to come across a gumtree listing in my local area for a number of vintage amplifier parts. I picked up a handful of really great components for a bargain!

A small selection of vintage switches, transformers and valves - many still brand new in their original packaging!

I stumbled across what appeared to be a complete set of power and output transformers pulled from a valve stereo amplifier, the secondary windings had been conveniently labeled so it was not too difficult to determine a suitable circuit topology.

Two output transformers (L+R) and a power transformer (C). Note the three primary taps, indicating the transformers are built for class AB (push-pull) operation.

I applied a small test signal using my function generator to the 8 ohm taps of the output transformers and using the relationship:

where Zp and Zs are the primary and secondary transformer impedances respectively. α is the transformer turns ratio, which can be calculated:

I calculated the output transformer's primary impedances to be around 4k Ohms, which was a good match for a set of EL34 valves I had also picked up - the EL34 pentode valves are specified for an anode resistance of 3-4k Ohms in class AB operation.

I had decided I would try to build a stereo amplifier sourcing as much as I could for the build from stock I had on hand (being a student meant that cost was a significant influence on the design). I decided I would base my design on the Dynaco ST-70 stereo amplifier - which was a very popular kit amp which was first sold in 1959. I will attach a link to the schematic here.

Dynaco ST-70 - Image credit: http://www.radiolaguy.com/images/audio/Dynaco70-f.jpg

My first design challenge was adapting the power transformer I had to suit the purpose of this amplifier. The table below compares my unbranded power transformer with the model PA060 transformer used in the ST-70:

Transformer	PA060	unbranded
Output	360-0-360V, 300mA	330V
Filament	2 x 6.3V, 5A 1 x 5V, 4A	1 x 6.3V
Bias	55V	6.3V

Table comparing the PA060 transformer with my unbranded transformer.

Current ratings for my transformer are unknown therefore are not included.

Some problems were more easily addressed than others - In my design I replaced the valve rectifier with a full bridge diode rectifier to eliminate the requirement of a 5V secondary tap. As an added benefit, silicon diodes are cheaper and have a much lower forward voltage drop than the original tube rectifier, which will help compensate in my power transformer's lower output voltage.

My next issue was sourcing a suitable power supply for the 4 x EL34 and two preamp valve filaments. The EL34 datasheet specifies the heater current to be around 1.5A per valve, so the output section alone requires 6A of heater current to operate correctly!

I had two NOS 2 x 6V, 1.7A transformers in my workshop. According to the manufacturer's datasheet the secondary windings can be wired in parallel for a total 3.4A current supply at 6V. I substituted the rare 7199 preamp valves for a pair of 6u8, which I was able to obtain for a fraction of the price. The 6u8 datasheet specifies the heater current as 450mA, I have used the 6.3V Filament winding of the HT transformer to supply the preamp section with 6.3V at 900mA.

Schematic diagram of the filament wiring

The original ST-70 design implemented a fixed-bias output stage with a combined plate current of 100mA per channel (Ip = 50mA). A look at the EL34 datasheet revealed the characteristic curves specifying a corresponding grid voltage in the region of -20 to -30V. The bias winding on the HT transformer would supply a maximum of -8.8V using a standard diode rectifier, therefore a voltage multiplier circuit was required to achieve the desired bias voltage.

Using 3 Villard cascade multiplier sections I could achieve a bias voltage of approximately -50VDC. A potentiometer was then placed across the output to allow adjustment of bias voltage when replacing the output valves.

Simulation of the voltage multiplier in LTspice

The amplifier is now operational, I have spent a few hours listening to it and overall I'm quite happy with the result.

Under-side of chassis.

Finished product with NAD 1155 preamp

I have taken some measurements using my Rigol DS1052E digital oscilloscope, the binary .wfm files were imported into MATLAB using Paul Wagenaars' excellent script. I was having trouble with MATLABs FFT function, so a friend helped me out with a simple python tool which yielded some nice informative graphs. Below are waveforms for sine and square wave into an 8 ohm load at the maximum power output and corresponding plots of the frequency content.