Recording - June 2013 - 72

submarine U48 and over 250 people died, including Mr. Murphy, as well as
dozens of children who were being evacuated to Canada as part of a government-sponsored evacuation scheme.
But although the inventor was killed so early in his life, his idea lived on,
and engineers working at RCA in Princeton, NJ came up with a second
extremely ingenious idea. They found that, although using this circuit led
to considerable residual distortion, most of it was second harmonic distortion, and that by using two of these circuits in push-pull a lot of that could
be eliminated. This circuit, using 6N7 tubes, was adopted by the RCA 86
compressor in 1937, only two years later. This was probably the first actual
audio compressor and certainly the first one sold as a commercial product.
Some other designs out there used other gain-reducing devices besides
the variable-mu circuit (we use the Greek letter mu (µ) for gain, so the
notion of changing the gain of a tube is called a variable-mu circuit). Some

Guest editorial by Scott Dorsey

A Compression Story
Back in the late 1920s, an engineer working for Marconi Electronics in Chelmsford,
England, came up with a really ingenious
idea. Francis Money Graham Murphy was
born in 1903 as the son of a clergyman, was
educated a Cambridge, and by 1932 had
started working for Marconi.
He noticed that vacuum tubes don’t
amplify in a linear fashion, that the actual transfer function of them is exponential. Under normal circumstances this is
not an issue because the signal level in
most audio circuits is very low, so a very
small portion of that exponential curve
is being used, and if a DC bias voltage is
added that portion of the curve is in the
most linear point.
What Mr. Murphy at Marconi thought of,
though, is that you can adjust that DC bias
voltage, and you don’t have to adjust it so
that the most linear part of the curve is
being used. Not only that, you can adjust it
so that you’re using a portion of the curve
with a sharp slope for a lot of gain, or a portion of the curve with a flatter slope so the
gain is reduced. It’s a small step from having a circuit whose gain can be changed by
adjusting a voltage, to making that voltage
controlled by the signal level so the overall
gain is controlled by signal.
Now, Marconi back then had a patent
licensing agreement with RCA in New
York, so while Marconi patented the
idea in the UK, RCA patented it in
America. You can look up US patent
2,025.019 if you’re curious about the
original patent; it was finally approved
on December 17, 1935.
What happened to Mr. Murphy? He did
some other research and received a few
other patents before being moved into
management a few years later. In
September of 1940, he boarded the
steamship City Of Benares to travel to
Canada. On September 18, the ship was
torpedoed and sunk by the German
72

RECORDING June 2013

folks used variable resistors controlled mechanically by motors. Others used
light-dependent resistors controlled by a light bulb driven by signal, but all
of these other schemes are for a different article and a different time.
Incidentally, the first tube specifically designed to be used in this sort of
circuit was the 6386, which seems to have been introduced in 1953 and
which spawned a whole new generation of much higher-quality variablemu compressors.
Now, in the 1960s when J-FETs (junction field effect transistors) became
available, they had a similar nonlinear characteristic and many folks tried
using more or less the same circuit for solid-state variable-mu compressors.
Probably the most classic of the early examples is the 1967 Shure LevelLoc, which is a microphone preamplifier whose front end is a single J-FET
whose bias can be adjusted to control the gain over a very wide range of
levels. Shure doesn’t seem to have tried to patent this.
So... enter David Blackmer. Mr. Blackmer was born in 1927, at about the
time that Mr. Murphy was plotting out tube curves. He studied electronics
in the Navy and got the military electronics training that started so many
on the road to engineering, then worked in the electronics industry for
many years before starting dbx in 1971. In 1975, he came up with a refinement of the circuit design using a single gain control FET that dramatically reduced the distortion at the extreme ranges of the gain control.
Now, this gain control circuit, which was incorporated in the dbx 202 module, was never patented. But it had applications in everything from compressors and limiters to companding devices for tape noise reduction, to expanders
for exaggerating dynamics, all of which dbx made and sold happily.
Mr. Blackmer later sold dbx, went on to found Earthworks to make a
microphone design he had, and was active in the Audio Engineering
Society for many years. When he died in 2002 at age 75 he was still working to make better electronics for better audio, and he was always willing
to talk to someone who just wanted to stop by and talk about audio.
As the years have gone by, there have been a lot of refinements on this
circuit, by folks like Paul Buff at Allison Research and Harvey Rubens of
VCA Associates and somebody at Alesis. dbx themselves refined the
design and managed to make it possible to put a Blackmer Gain Cell on a
single integrated circuit, and the successors of those circuits are sold by
THAT Corporation.
But just about every analogue solid state compressor you will ever see
today, no matter whether it’s a voice-grade compressor in a cellphone or
the highest-grade studio compressor made today, uses a variation of
Blackmer’s gain cell, and Blackmer’s gain cell takes advantage of the variable-mu principle that Murphy invented nearly a century ago. There’s a
direct line of descent that I think is kind of cool.
Scott Dorsey (dorsey@recordingmag.com) is both an audio engineer and
an electrical engineer, giving him double the enjoyment of stories like this.

Recording - June 2013

Table of Contents for the Digital Edition of Recording - June 2013