1970s Design Indulgence

I wish I had more readers who understand how the basics of transistor circuits work. I'm pretty sure most who read these pages have little grasp of electronics. I know one or two do, but maybe don't let on.

So, for those who'd really like to understand, a good book to start with is Ray Marston's Diode, Transistor and FET Circuits Manual.

First published in 1991, the FET circuits can by now be forgotten, as you'll be lucky to find any FETs other than those intended to switch.

Chapter 3 gives a great introduction into how transistors really work, making the understanding of them really clear.

If you can do Ohms law, you're already set up to grasp how a lot of the circuits, both in Marston's book, and on here, do their job.

I think we need to learn to ignore the marketing myths and instead, learn!

Attributed to Michi (Rotel) by one in the hifi press was the work of many others. This was some time ago when I read it - possibly near on 25 years.

It basically said:

The output current of the amplifier stage must be capable of charging the equalisation capacitors.

Charging them to what?

To the ever changing voltages dictated by the musical flow, which means both in the positive and the negative direction, and by the incremental differences of the levels of the music, which can be a transient, and at all the frequencies contained in the music, but also the noise gain of the amplifier stage.

So first you have to understand what a musical waveform looks like.

And also here you need to grasp that it is talking about a conventional negative feedback phono stage, not the passive interstage method created by RCA as a way of selling a double triode, that the mainstream flock to.

The output current and the capacitive load presented by the EQ capacitors determine the speed at which the stage can operate, otherwise known as the slew rate.

This is not a new thing!

The fed back signal adds to the input signal negatively (negative feedback) and must be 180 degrees out of phase to comply with that law.

Then, if the signals do not match the input stage adds or subtracts so the output matches the input signal, and that reduces the distortion.

Except, the harmonics of the distortion, without negative feedback are simple, but when fed back, the output generates more complex harmonics, even though they are not as great in size as if without negative feedback.

This is not a new thing either!

But, the symmetry wasn't understood, and I remember what an AES paper claimed was the best, because it had the drive current to be able to charge the capacitors. But, a single ended amplifier cannot drive on both half cycles, so it wasn't the best solution.

However, these became the "go to" design for a good part of the latter 70's/early 80's. It became known in Elektor as the Preco. It was an advancement on the two transistor circuits of earlier years.

What came next was the op-amp circuits, and these provided the symmetry, but not all of them.

Back in 1972 Otala described overshoot due to propagation delay, and very soon afterwards certain members of the AES "disproved" it. Except that a respected op-amp designer Jung, went on to develop the ideas of slew induced distortion, which wasn't far away from what Otala meant (IMO).

Horowitz and Hill went on the explain it quite well in "The Art of Electronics", that once the input stage becomes saturated then the signal "meets a brick wall" - the low input impedance voltage amplifier stage (of the op-amp).

The mathematics of slew rate with capacitive loading and available output current are nothing new either.

Neither is stability margin which is as old as Nyquist and Bode.

And neither is the effect of capacitive load on negative feedback - as old as the hills - divide the output and take that as the negative feedback - a way to reduce the value of NFB resistor. So, if the divider is a capacitance, the gain increases at some high frequency until?

So, why are the same mistakes made over and over?

It's as if nobody can hear the difference in sound.

An unstable circuit sounds dry. It is very stop and start, which means the music doesn't flow, but what it does have is "signature" which might suit some stop-start music.

Where does the instability exist? It exists between the op-amp's (or discrete circuit's) frequency response (beta >> 1) and the transition frequency (beta = 1).

Is that audible? No.

However, its effect is audible.

A simplistic argument says it wastes power. What it does is put asymmetry on the power supply, then one channel pulls ground current through the other. Do you know how a stereo width controller works?

The hifi press do not know. Best means projecting the image way beyond the outside edges of the loudspeakers, which is what a stereo width controller set to max does. It is negative crosstalk (now, which audio science website promotes that?)

Some therefore use tricks to impress, a lot more than you'd think.

There is a lot of self-righteousness in audio electronics design. Recognising that, I have tried to do my job in real righteousness.

So, back to overshoot and propagation delays. It is argued that the speed (slew rate) required is so slow for 20kHz, that propagation cannot come into it.

So, what could trigger instability between beta>>1 and beta 1?

And why is op-amp true gain called noise gain?

Nothing is absolutely immune. So should I cater for frequencies way beyond what is audible? Well, if they didn't do it with power amplifiers, they would burn up very quickly.

So, what is the difference between a small signal amplifier and a power amplifier in that respect? Nothing, except that the small signal amplifier goes on and on without burning out.

So, what can cause the propagation delay between output and NFB node?

Capacitors?

The layman says how can that be? The layman does not comprehend that capacitors have inductance (and resistance - and impedance anyway). So how can a capacitor result in propagation delays? (rhetorical question)

So, we have components that suit a task and components that don't. Or should I say, components that properly implemented lead to a natural flowing musical sound, no matter what music you throw at it.

In my case I was rather unfair to the ‘7eventy’/‘Nostalgia’ in that I had it replace a pair of Proprii and so all my comparisons relate to that situation. This will probably sound rather negative, but it really shouldn’t, because I’m comparing it to the best I know.

Ian who ran the relay leg before me said he thought there was less bass than the Proprii. I agree; a track we particularly noticed it on was Zombie by the Cranberries, but yesterday evening listening to Bowie’s Blackstar there seemed to be loads of bass. Relative to the Proprii: less bass, less depth and width of soundstage, less involving, less warmth and occasionally a bit fatiguing. We much preferred it with vinyl than we did streaming from NAS via RPi/HiFiBerry. One vinyl track that really didn’t work for us was Supertramp’s Sister Moonshine; the piccolo sounded rather piercing.

If I had had more time with it I would have liked to have compared to my ageing, recently serviced, Technics SUV-303.

Thanks to Graham for the opportunity to try out this amplifier.