1970s Design Indulgence

However, the ability to sell a half percent THD solid-state amplifier might elude me...

And to be perfectly honest it doesn't really sound right. I'd be kidding myself as well as you if I said so.

Reading between the lines - and I've done quite a lot of reading - high frequency distortion seems to have been a problem with many a 70s amplifier design.

The ones which didn't suffer appear to be those using regulated power supplies, which are quite busy circuits to say the least: a design by JLH had more parts than the amplifier it was powering.

Latter circuits tend to "float" everything between current sources and current mirrors and all but the output stage is well-isolated. This tends to illustrate the problem.

The availability of "cheap" but good regulated switch mode power supplies exists, but not in the voltages required for much above 30 watts. 48 volts seems to be the cut-off point for manufacturers.

If it were possible to make full use of 48 volts you might be able to swing 16 volts rms and obtain 32 watts into 8 ohms. However, if we look back 10 years at the topic which led to the Proprius, I think you will find I'd been there already, and the results were average at the best.

What emerged was the Proprius which wastes a few volts due to its asymmetry, and in doing so can only swing 14.4 volts rms - just short of 26 watts into 8 ohms - but has resulted in "a fine, deeply satisfying amplifier" (Stereo Times).

And I so wanted to produce a deeply satisfying integrated amplifier. Why you may ask? Well, today is about one box solutions as evidenced by this quote from a post on here: "I'm only selling them because I'm simplyfing my setup to 1 box".

Simplification leads to complication or compromise as I am finding. It was the reason why hi-fi separates emerged and became so popular. Get it right and it's deeply satisfying.

On the other hand I can see how untidy a power amplifier consisting of 4 boxes is. Putting them in one box raises a number of problems; one being the legality and functional safety of putting two desktop 48 volt switch mode power supplies into the box. I could make it look like a storage slot and write in the instructions that they should be pulled-out and used remotely (which the user could ignore), but otherwise wouldn't two IEC leads powering one box look weird?

There wouldn't be sufficient power left over for a motor driven volume control: these consume up to 200mA at 5V, which at 1 watt doesn't seem much, but a voltage dropper from 48 volts would result in nearly 10 watts of waste. Then which power supply do I take that off: the left channel or the right channel?

There would be sufficient power left for dual mono preamps, but who'd buy a manual integrated in these microchip convenience days?

I'm 64. They put the state pension age up to 66...

When I was a lad, lads got excited about multiple boxes arranged to serve a purpose. British motorbikes (unit construction) and high fidelity systems were made like that. Fiddling with controls was all the rage. If it had knobs, sliders, switches, levers and pedals it was cool.

The first remote controls I saw were tethered to the equipment they served by multi-core cables, and with the exception of slide projectors, we understood these remote controls were for the elderly and disabled.

If this and other writings on amplifier design, or its created evolution, should ever be looked back on by a far distant future generation, we'll all look like a bunch of amateurs.

The problem has always been backward compatibility: competing with valves.

The transistor attempts to be a valve but because of scientific reasons never will, and this was ably demonstrated by early transistor amplifiers being low powered compared to a medium output valve amplifier. Here I'm thinking 50 watts. Early transformer-less transistor amplifiers could not achieve 50 watts.

The transistor (of the BJT type) is a low-impedance device, but has to be "tricked" into having a high input impedance. It is a low current gain device added to which is its high capacitance which makes its high frequency performance useless.

It can only compete with the valve amplifier at low power, but the urge to own at least 100 watts of transistor amplifier has made any chance of it competing with the valve impossible..

Lower powered transistor amplifiers don't sell and the proof is the demise of the once available devices.

Julian Vereker knew this only too well. The NAITs could only do less than 25 watts - I seem to remember they were nearer 21 watts - and so Mr Vereker didn't publish any specifications. Quite wise.

We can read the works of JLH on transistor amplifiers and see the folly of damping factor: to make the transistor amplifier sound more "human" like the valve amplifier he used an output resistor - and so did the NAITs!

At the same time the industry was harping on about damping factor - aiming for the stars - yet the sound was getting worse.

Why? Because a transistor can never be a valve!

3 valves and you're there. 16 transistors and you might be lucky. And for what? Power! We are asking too much of the transistor.

The maximum sustaining voltages of the earliest transistors up to the late 1960s would be a maximum of 60 volts. On a wing and a prayer some would try 80 volt HT hoping that none of the transistors would venture outside their 60 volt zone, but faults happened and they blew the lot.

60 volts might result in 38 watts; 80 volts might result in 60 watts - at the onset of clipping. With the full global negative feedback in place and the "right" test set filters, and usually at half power 0.01% THD might have been measured at 1kHz, but I can categorically state that at the onset of clipping the distortion is nothing like. More like 1% or more.

But they could sound nice. The lower sustaining voltages would not suffer the Early effect problems high voltage transistors face. As we ask for more power, we have to have more volts to drive the valve-era 8 ohms load.

What we should have done IMO is to convert to 4 ohm speakers and stay with low voltage transistors that we might obtain 75 watts of nice sound.

What is apparent in all the designs or incarnations of the last year is that the effect of the Early voltage messes up linearity to the degree of increasing activity between the audio spectrum and the crossing frequency.

By adding high frequency noise the wanted frequencies are modulated and will produce false harmonics, making for a brighter sound.

The heavy filtering of a switch mode power supply along with its tight regulation prevents this modulation to some degree.

The problem is the lack of high voltage switch mode power supplies, so we have gone full circle again, and the solution still points to lower voltages and the resultant lower powers.

The only solution from a purely measurement point of view is massive open-loop gain coupled with a large amount of negative feedback, which Mr Self and Mr Cordell seem to promote. That might be OK for a disco amplifier or in a loud demo room at a hifi show, but is absolutely useless for home listening.

My thoughts turned to transistors such as BC141/161's; TIP33's and perhaps a BC327 for the input stage. I can get BC327's...