Briarsfield Hi-Fi - Akai M-8 restoration

The year is 1963. Transistor technology is still in its infancy, large format tape is still king; Philips haven't even invented the cassette tape yet.

The Akai M-8 was a full-size reel to reel tape recorder, they were also sold under the Roberts brand. Built in Tokyo in Japan, these were made in the era before Japan really got a name for quality audio equipment.
Audio came from a pair of small valve amplifiers, consisting of a two stage pre-amplifier using a single ECC83 and a single ended EL84 power amplifier capable of around 5 watts RMS. The phono/microphone stage were powered by a single EF86 with a simple passive filter network to approximate correct equalization; the pre-amp also doubled as a record amplifier when recording. There was also a valve oscillator on the left channel to power the deck's erase head.

All stages of the circuit were cathode biased, the whole thing was completely unremarkable as far as valves go but single ended valve amplifiers can always sound good if set up correctly. The strength of this type of circuit is its purity, aside from the valves there's very little else needed to make it work - the signal line is very clean.

Akai M-8 in original condition (Photo courtesy of Stuart Lester).

They came to me in a box with many other parts. In the box was also a design brief with the major goals:

The first part of the plan would be to merely get the amps running. This seemed relatively straightforward, just a matter of spending a couple of hours connecting the various runs of wiring which formed part of the play/record circuit, making sure everything was safe before power was applied. Then came the first opportunity to actually hear them - there was a definite lack of bass and the treble was extremely glassy with lots of ringing on the midrange, not at all valve-like; but what was there sounded promising.

Powered up for the first time, possibly the worst photo I've ever taken..

It was clear that the sound problems were coming from the small coupling capacitors (the lack of bass) and the filter components (the ringing and glassy treble), the distinct roll off at the high end probably made worse by stray capacitance from all the old wiring and switches - a large portion of the circuitry could indeed be removed from each amplifier and would improve things greatly. The components used originally were old and never really intended for audio, changing these would also yield a large improvement to the sound.

The original schematic. Most of what's in there can be removed. Original schematic courtesy of Akai.

With the amps themselves working, the next thing was to begin working out other aspects of the design. The first thing on the list would be the tone circuits, because these would ideally need to be added to each channel as it was rebuilt. I chose a basic Mullard passive tone network, I had built these before and they worked well - one reason these appealed to me was that they could easily be built around an existing valve stage, they were relatively insensitive to impedances of the components around them and also the filter frequencies could be changed at will merely by changing a couple of components per channel. The huge downside to these is the huge amount of gain I would lose, around 20dB. This is effectively cutting the signal level by 10, doing some basic math on the available gain and signal levels told me that this wouldn't work unless I could increase the signal levels by enough to compensate. the only attenuation network was before the first ECC83 stage, after plotting the loadline on paper it seemed that I could afford to add another 20dB there. I was worried about what the audible effect might be if I drove the valve with a signal that high, in practice though it worked well.

The VU meters might prove a challenge, a brief test showed that they needed around 3 volts to drive them into the +3dB red band and there was nowhere near that kind of voltage in the signal circuits and in any case the extra load would degrade the sound quality. So these would each need to be fitted with a dedicated op-amp to provide the voltage swing they needed without loading the sensitive valves.

A small amplifier for each VU would be ideal, an op-amp could easily provide the voltage swing and could be designed with enough gain to be able to use as much shunt resistance as possible between it and the audio, to reduce any interaction and hence distortion as far as possible. To keep things simple, I would try to power these from the heater windings on the main transformers, in practice though I would run into problems with this idea and would end up fitting a separate transformer for them.

Originally the VUs were fitted with twin lamps, these were blackened with age so would be replaced with new lamps. Also a dimmer switch would be fitted, this would need some kind of adjustable voltage regulator - a simple LM317 would be ideal if I could keep it cool enough.

The pre-amp section seemed simple enough at first glance, merely a selector switch and a few runs of audio cable, the selector switch being fitted in place of two of the input jacks on the right channel's amplifier. However, the shape of the chassis of the amps and the desired place for the selector would cause problems and mean long runs of thin cable, not something I'd want on a quality audio system - besides, it would become a nightmare to connect everything together. Also, a good quality wafer switch would prove far too large to fit into the space where the switch was meant to go.

I decided to switch the inputs via signal relays, this meant more cost and complexity but meant that all audio switching could now be done on the rear panel, with only a minimal length of wiring to run through. Also, we could now use a bog standard rotary switch - which would work fine and cope easily switching the 5 volts to each relay coil - without it having any effect on the audio. Nothing wrong with a standard rotary switch, they give a nice click as you turn them, this along with the switchover click of the relays makes it sound quite chunky. Also they're fairly rugged and should last a very long time.

Tape record outs wouldn't be a problem, as on most equipment I'd drive them straight from the line inputs. The pre-outs might cause issues however, as modern transistor amplifiers tend to be more of a load than valves are used to. I decided to run the pre-outs straight from the output of the second ECC83, really there was no other way. Not the perfect solution but in practice it worked well.

The cabinet wouldn't be a major problem, from the start I was planning on building a basic cabinet from sheet MDF, something which would be strong enough and which could be covered in some kind of material. Covering material - a dark fabric - had already been sent, I wasn't sure how difficult this would be to cover but I decided to worry about that later.

So, we finally had the makings of a plan. There was a lot more to do but things were slowly starting to come together.