The MRP-1 Preamplifier.

Updated: 25 Oct 98
to The Project index
                       THE MRP1 PREAMPLIFIER. 
25 OCT 1998                                       MRP1.TXT

There never were any assembly notes written for the MRP-1.
This is my retrospective view of the whole business, which now
seems as if it all happened a very, very long time ago. It

This is the first preamplifier I designed, and the first of my
preamps to be published. It was designed in 1975 and appeared
in the Nov 1976 issue of Wireless World, (now Electronics
World) with some additions and extras appearing in Sept 1977.

There was essentially one aim when I designed this; to build
the best preamplifier in the world, bar none. This may seem
presumptuous, but in a world full of preamp designs, there
seemed to be no point in working on anything else.

The opamps available (at a long price) were the 709 and the
741. The 709 (in my hands at least) was hard to stabilise at
HF, and usually short-lived as it had no short-circuit
protection. The 741 was a great improvement for many
applications, but its poor distortion performance and low
slewrate meant that even in those days it only appeared in the
cheapest of equipment. There WAS a low-distortion opamp- I
think it was the National LM318H- but at that time the price
was truly prohibitive.

There were two ways to deal with this:

The first way was to purpose-design each discrete
amplification stage for its particular job; this was
successful because in most cases a differential amplifier was
not required. The familiar Baxandall tone-control stage
requires an inverting input, but this is assumed to be
referenced to ground. Similarily, filters, buffers, etc,
usually employ a unity-gain buffer, or emitter-follower, which
can be regarded as a differential amp where the inverting
input is always connected to the output. In preamps, true
differential amplifiers are rarely required. One exception is
in graphic equaliser stages, but a preamp with built-in
graphic today (1998) is about as fashionable as cauterising
wounds with red-hot irons.

The second way was to design your own discrete opamp. My first
adventures in this field were prompted by Daniel Meyer. [see
Wireless World July 1972, p309]
I soon discovered (this was during my first fortnight in the
hifi business) that the Meyer design had poor linearity.
Investigation showed that this was because the gain stage,
with a high collector impedance to get the NFB factor up, was
seriously degraded with even a light load. I added an emitter-
follower for buffering, and the THD performance was
transformed. A few more experiments with cascoding, improved
output buffering, etc, and I had an opamp with sub-0.002% THD
at 10 Vrms, which I reckoned was quite something; what was
encouraging was that it was so easy to get there. This sowed
an enthusiasm in me for low distortion that still endures;
it's so easy to get it beautifully linear- so why not do so?

My specific aims for this preamp were:

1) To demonstrate that stunning performance could be achieved
by using properly-designed discrete opamps. In particular I
was looking for a high disc overload margin, inspired by a
design I read about that used no less than +136 V to power the
disc input stage.
The use of discrete opamps meant that high supply rails could
be employed to maximise headroom throughout the preamp. In
fact, I used +/-24V as this was the highest I could get from
IC regulators, ie LM7824.

2) To introduce the Rumble Gate. This was a noise-gate
designed to suppress the hiss from an unemployed preamp, by
opening the output mute relay. The idea was that the subsonic
rumble from the run-in groove of even the flattest disc would
be more than enough to unmute the gate and ensure that not
even a microsecond of the first note would be lost.
This concept, as you may have noticed, utterly failed to catch
on, and I am still devastated.

3) To use separate input gain and output Volume controls.
The input gain controls set the signal level through the
preamp at the optimum level to strike a balance between noise
and headroom. The output volume potentiometer sets the actual
acoustic level.

There was much agonising over the architecture of the preamp,
heavily influenced by some of the early Cambridge Audio

1) Input headroom Disc stage. The input stage actually runs
between +24V and 0V, not between the +/-24V rails. This is 6
dB of headroom apparently thrown away. I did it because the
non-opamp, single-rail input stage was the quietest I could
design at the time, and I put noise first. The other point is
that the gain of this stage is so low that the input
capability is still enormous.

2) High supply rails. It will not have escaped your notice
that 24V is only 3 dB up on 17V (the standard opamp rails for
audio) and so this is a hard-work way to get more headroom. 

1) The discrete opamps all used LED biasing, as I had read in
some electronics magazine that if this was used to bias a
conventional current-source, it would give near-perfect
thermal compensation. I took this on trust at the time (and I
wish I could remember where I read it, because this was one of
those "seminal" experiences) but I have since tested this in
an environmental chamber (about 20 years later) and proved it
to be absolutely true. The model for a LED in this application
looks like a single silicon diode in series with a constant
voltage source. The LED colour, for red, yellow and green
anyway (I have not tried blue ones) seems unimportant.

In sober fact, this degree of precision in a discrete opamp is
completely unnecessary. None of the standing currents are in
any way critical. However, it does look jolly pretty on a
circuit board, and you must remember that in those far-off
days LEDs were relatively exotic (I saw my first LED in 1970,
at the Post Office Research Station) and a PCB that looked
like LA at night got many admiring comments.

                                        Douglas Self