It's probably about ten years ago that I read a letter in
Wireless World magazine from the late, and great, Scroggie
deploring the term "current dumping" to describe the amplifier
topology invented and exploited by the Acoustical Manufacturing
Company in the Quad 405. Scroggie acknowledged that the term
pertained to the relatively straightforward task undertaken by
the output transistors in such an amplifier. However, he argued,
since the current the output transistors passed was that used
directly to energise the loudspeaker and thus produce sound, the
act these transistors played could hardly be considered
"dumping" suggesting - as the word does - that they were
involved in disposing of a useless commodity. Times change and
words with it. Now dumping is an everyday word for storing
something precious - computer data. Perhaps it is only the
ever-Victorian, scatological British who could regard
responsible "dumping" as the hallmark of a accountable
professional life. Woe will surely befall a man who fails to
dump his data in the appropriate way and at the appropriate
time! Not that Scroggie was criticising the development of the
Quad 405. I would say that the belief is generally held that
Quad's current-dumping idea is a clever conception. I've owned a
Quad 405 as a studio amplifier for many years and it's a
wonderful product. I choose to ignore the engineers and
academics who consign it to the waste-paper basket of history,
they - blinded by mathematics - prefer their own (faulty)
reasoning to the evidence of their own senses. Technically
current-dumping does work - believe me I've measured it. It
works well and it goes on working, year after year. But how
exactly does it work?
In order to illustrate this,
I've drawn Fig. 1; a simple power amplifier with a voltage
amplification stage (A1) and a complementary bipolar transistor
output-stage. In many ways this looks like any other
complementary power amplifier except that there is no static
bias applied to the output transistors. Negative feedback
encloses the whole and attempts to maintain linearity despite
the large transfer-characteristic discontinuity introduced by
the unbiased output stage. If you breadboard something like my
Fig. 1, I guarantee you'll be surprised. It doesn't sound that
bad. Provided the gain/bandwidth of A1 is large enough, the
output signal is remarkably undistorted, especially at low
frequencies. Essentially this arrangement would be entirely
acceptable were it possible to construct the amplifier A1
perfectly, so that it "slewed" infinitely quickly across the
crossover "deadband". I've tried to show the way A1 behaves by
sketching the signal waveform at the bases of the two
transistors. Of course, it isn't possible to construct a perfect
amplifier for A1 and, in practice, as frequency increases,
crossover distortion starts to make itself heard. The essence of
the idea behind current-dumping is illustrated by the inclusion
of Rd (shown with dotted connections in Fig. 1). Rd feeds
current directly to the load during the proportion of the output
cycle when both the output transistors are off. In effect, Rd
reduces the "gap" A1 is required to slew across and permits a
practical amplifier to be used instead of a mythological,
perfect amplifier. Stripped of their duty during the essential
and fragile crossover region, the output transistors are only
called upon to deliver drive into the load at powers above about
100mW whilst A1 does all the clever bit in between - hence the
term "current-dumping" to describe the uncomplicated job they
perform. In Quad's final circuit, the resistor is augmented by a
capacitor and inductor which act in a reactance-bridge
arrangement to enhance the effect of Rd across the whole audio
bandwidth.
Despite the undeniable engineering ingenuity displayed in the
405 amplifier, Quad never seem to have earned creme-de-la-creme,
audiophile status for this product nor indeed its heirs.
However, comments that amplifiers of this stable lack, "slam" or
"punch", probably derive from Quad's prudent decision to employ
extensive current-protection and sub-bass roll-off (features
which have many times saved my 405's life during its spell in my
studio). Importantly, it is almost certainly not the action of
the amplifier topology itself which accounts for this apparent
lack of subjective impact, but the effect of these other
ancillary features and design decisions. Is it, perhaps, because
of this poor subjective reaction to the Quad amplifiers that
other manufacturers have not trodden the current-dumping route?
Unlikely, more probable is that Quad's patents have prevented a
flood of carbon-copy, current-dumping amplifiers.
That brings me to the nub of this month's column, does patenting
an invention like current-dumping really do anybody any good in
the long-run? Granted, Quad have kept the technology to
themselves but have they ultimately profited from this? Have
they sold more amplifiers by virtue of their innovation? I doubt
it. Quad's undoubted reputation of reliability and solid
engineering would hardly have fallen apart if they had (like
everybody else) built upon their success with the 303 and
continued with the development of the class A/B audio power
amplifier. True the designers, freed of their usual
commercially-rooted secrecy by the sanctions of the law, were
able to write articles in learned periodicals explaining the
technique in full and bask in the glory of their own cleverness.
But did Quad really hope that other manufacturers, wowed by the
brilliance of their idea, would opt to pay Quad a royalty for
every power amplifier sold? Surely not. After all, putting the
invention in perspective, what was done in the 405 was to
engineer a power amplifier minus the usual requirement for a
quiescent bias adjustment and that simply just does not confer a
big enough commercial advantage for others to follow. It's not
as if they invented an amplifier without the need for a mains
transformer! Sadly, I have to conclude that protecting the
current-dumping invention has done neither Quad, nor its
inventors, nor- most importantly - the idea itself any favours.
Interestingly, Quad's patents for current-dumping were published
in 1975 which means the protection expires next year. So perhaps
current dumping could yet see its full flowering and become
household bywords for high-quality amplification. Sorry Mr.
Scroggie.
I've drawn Fig. 1; a simple power amplifier with a voltage
amplification stage (A1) and a complementary bipolar transistor
output-stage. In many ways this looks like any other
complementary power amplifier except that there is no static
bias applied to the output transistors. Negative feedback
encloses the whole and attempts to maintain linearity despite
the large transfer-characteristic discontinuity introduced by
the unbiased output stage. If you breadboard something like my
Fig. 1, I guarantee you'll be surprised. It doesn't sound that
bad. Provided the gain/bandwidth of A1 is large enough, the
output signal is remarkably undistorted, especially at low
frequencies. Essentially this arrangement would be entirely
acceptable were it possible to construct the amplifier A1
perfectly, so that it "slewed" infinitely quickly across the
crossover "deadband". I've tried to show the way A1 behaves by
sketching the signal waveform at the bases of the two
transistors. Of course, it isn't possible to construct a perfect
amplifier for A1 and, in practice, as frequency increases,
crossover distortion starts to make itself heard. The essence of
the idea behind current-dumping is illustrated by the inclusion
of Rd (shown with dotted connections in Fig. 1). Rd feeds
current directly to the load during the proportion of the output
cycle when both the output transistors are off. In effect, Rd
reduces the "gap" A1 is required to slew across and permits a
practical amplifier to be used instead of a mythological,
perfect amplifier. Stripped of their duty during the essential
and fragile crossover region, the output transistors are only
called upon to deliver drive into the load at powers above about
100mW whilst A1 does all the clever bit in between - hence the
term "current-dumping" to describe the uncomplicated job they
perform. In Quad's final circuit, the resistor is augmented by a
capacitor and inductor which act in a reactance-bridge
arrangement to enhance the effect of Rd across the whole audio
bandwidth.
