[sdiy] Speculation on the idea of being on the Roland design team ca. 1992
rw at firstpr.com.au
Tue Jan 21 17:21:35 CET 2014
Was: Re: [sdiy] AH discussion: History: Don Lewis, Roland, TR-808,
popular electronic music
Regarding the last part of:
> Hi Robin, great story!
> It is fascinating to contemplate what Roland may have developed from
> your designs. I wonder how much resistance you would have got to the
> idea of making it all analogue.?
> Roland have never yet re-visited that technology, as far as I know.
It is an interesting hypothetical question. I would not have moved to
Japan. I would have had to sell my 1979 VW Kombi and give up living in
a place where solitude can be found with a 1 hour drive and a 10 hour
drive can take you to places where people are very scarce indeed.
Within a day's drive of Melbourne, there's a vast range of geography to
enjoy, and there is far more at greater distances - plenty I have never
experienced yet despite growing up here since the age of 6 in 1961.
Even if I don't go out of Melbourne from one month to the next, knowing
it there is important part of being alive.
I would have had to pack/sort/throw-out/store a house full of generally
useful or essential stuff. I would have had to close a thriving and
sociable electronic musical instrument modification business and go and
live in a tiny apartment in Japan, being dependent on a single employer
- albeit an illustrious one. Japanese is by all accounts devilishly
difficult to learn, and I am not good at such things. From all I know,
Westerners find it difficult to feel at home there.
If I had wanted to live there for a few years and if I did work as part
of the Roland design team, it would have been most interesting. I
really like they way they and other companies design and make things,
especially musical instruments - or at least they way they used to.
In principle, it might have been possible to influence the products to
the degree that some mass-market Roland gear embodied my thoughts on
generative music - provocative musical instruments which through their
own idiosyncrasies or explicit design features create sounds and timings
and musical processes in general which are not an exact reflection of
the user's intention or actions.
Likewise the idea that things should be capable of distortion,
overloading, cross-modulation, and messy and chaotic behaviour.
This is relatively simple to do in analogue, and with care it can sound
fantastic. I think it is much more challenging to do such complexity
and chaotic stuff, including "dirtiness", in the digital realm and have
it sound physically natural, without tiresome digital artefacts due to
aliasing, hard clipping, sample-based time-quantized delays rather than
sub-microsecond responses etc.
Maybe something ("drum machine", bass sequenced synthesizer, a
combination of the two - or why should there be such a distinction
between them?) along these lines could have been designed, manufactured
and sold with sufficient success to make it profitable, or at least to
lead to further developments which were financially successful.
However, there are many reasons why this might not have transpired.
Most musical instruments are made to serve the commands of the musician
as faithfully as possible. That's fine, but I am interested in making
instruments for people like me who firstly don't have the chops to play
notes on time or on pitch and secondly, and more importantly, don't
necessarily have a clear idea of what they want to do. People like me
want toys to play with - catalysts for larger processes out of which
comes at least fun, and perhaps something other people will enjoy
That's not what Roland were trying to do with the TR-808, TR-606 or
TB-303. As far as I know, this was their honest attempt, with the best
technology they had - this was **before** sampling was cost-effective
for them - to make a drum machine that sounded like real drums, or a
cheap electronic synthesizer with real musical expression capabilities
such as slide and accent, which sounded something like a bass guitar.
The beauty of these machines is how they *didn't* achieve this, because
no-one can achieve these goals with analogue circuitry. What they
created was musically expressive and dynamic, but it never sounded
really like drums, cymbals etc. - though the Cowbell was a remarkable
The 808 had lots of sounds, lots of controls, lots of outputs and the
Twin-T resonator circuits were naturally somewhat drum like, since they
are basically sinusoidal. There are strokes of genius - such as the Tom
circuits having back to back germanium diodes to increase the pitch at
higher sound levels, just as the skin of a drum gets tighter on average
when hit hard.
The hand-clap is brilliant. The snare noise filtering is too high and
short. The snare tones are too high. The bass drum has an odd click at
the start and is quite unnatural. I think it has its place, but I
prefer the 606, which has two non-clicky Twin-T resonators, starting in
phase with a big boomf, and then going their own ways due to their
Or course the TB-303 is totally inspired from the plain VCO onwards,
with its diode ladder filter and unique control systems for VCA and
filter, with Accent and the Accent Sweep Circuit, depending on the
position of the Resonance knob.
I think the TR-808 and TR-606 cymbals and hi-hat are utterly brilliant.
Six out of tune square wave oscillators (a single hex Schmidt trigger
chip, 6 resistors, 6 capacitors and 6 more resistors to mix them - and
two of the oscillators drive the Cow Bell) and then some totally artful
and inspired envelope driven mashing, with fixed high pass filters . . .
such a beautiful sound. It sounds even better with external signal
input to the mashing circuits. Once we tried a ride cymbal, softly
driven by constant mallet impacts, amplified up. Then the TR-808 really
sounded like real cymbals and hi-hat - but better still, it sounded like
nothing before in creation with high bright chord sounds from a
polyphonic keyboard. My favourite source, to this day, is the Organ
sound of the Casio M10 with its LPF bypassed. (This is a totally
complex weird looking waveform, with ~500kHz sampling rate - simply the
best they could do in 1980 because they didn't have the technology to
sample a real organ sound.)
Almost all the sounds of the TR-808, and all of the TR-606, respond to
the trigger level in a totally physically musically meaningful way - the
whole nature of the sound changes in much the same way as that of a
physical object does, depending on how hard it is struck.
Part of the charm of the 808, 606 and 303 is how they are really
*trying* to sound like a real physical instrument, getting quite a lot
of the way, achieving a great deal of musicality, but ultimately
failing, despite their valiant efforts. "Their" could be the designers,
the machine itself or the musicians who chose to use the machine because
they preferred it to a real drummer, or couldn't afford a real drummer.
(Q: What's the best argument for drum machines? A: Drummers. But only
if you are happy with a potentially lively mechanical flavour.)
This is the same kind of thing which makes a half-way decent watercolour
painting of a sunset worth looking at. It is the effort, the striving,
the partial success in spite of the difficulties and really the
*impossibility* of depicting a real sunset, that makes it valuable - in
the contest that an artist (or a drum machine manufacturer) was doing
their utmost to turn their paint or their circuitry into something more
elevated and evocative than just a bunch of paint or an electronic
A child's painting of the sunset or whatever can still be touching and
inspiring - if we know it is the work of a child (or perhaps a naive and
partially disabled adult) *striving*, rather than an idiot adult
sploshing paint carelessly.
I think this is a big thing behind art. Strive to depict or express
something in a way which is impossible, and if you get some of the way,
people appreciate the effort and the charm of the partial success mixed
with failure. Brave and hopeful effort in the face of impossibility.
Also, they appreciate the *sensibility* the artist was striving for,
since it is usually not an attempt to photographically reproduce a
scene, or to produce a sound which is *exactly* like the original by way
of being a recording of a particular such sound.
Paintings which look like uninspiring photographs are tedious and not at
all an addition to the world. Likewise boring photographs which convey
no insight or view. Likewise sampling, unless it is used with great
flair. (Think Yello, The Art of Noise, the remix CD which came with the
Australian release of Deep Forest's first CD . . . I think this is rare,
and only done by people who have a deep sense of the physical world and
how their photocopy techniques are bound to fail in the longer-term
musical sense, in terms of timing and veracity of physical instruments
never sounding the same from one note to the next. So they use brief
snippets of sound, copy and pasted time and again, ra - ra - ra - ra
like a cash register cranking over, in ways which could never be played
manually, to create something which is bound to fail in terms of
replicating physical playing - and yet create something fresh and
memorable at the same time.)
But look what happens when you give the designers digital sampling.
The Yamaha RX7 and RX9, the Alesis drum machines etc. etc. etc. The
TR-909 . . . I think these are terribly boring machines.
The 909 is offensive to the ear by using triangle waves and VCAs with
envelopes, hardly even exponential envelopes, to make something
supposedly like a tom sound. It sounds like crap to me. The way these
fail to achieve their goal is attractive to some folks, but I think the
attraction for them is a kind of mechanical irony - a buzzy annoying not
even half-way attractive or passable imitation, combined with a very
predictable, solid, kick sound
There's a new Roland video "Talk about TR-909":
where they say they started with the idea of analogue, but they
preferred to sample the cymbals.
Of all things, the delicate, different-every-stroke, cymbal and hi-hat
sounds should not be reduced to a single dead, frozen, lifeless and
instantly grating and annoying recording of a single cymbal sound.
What were they thinking?
Why do people like this stuff, except for the ironic sound of
pretending, sort of, to be having cymbal sound with copy and paste rapid
repetition of *exactly* the same sample, hundreds of times in a single
piece? The ear tells you it is the same sample, the second time it
plays. (Or do people really not notice?) So it is not so much like
hearing a sound process, it is like ripping your ears back to a time in
the past when a cymbal was hit, and then ripping your ears back to that
time again, and again, and again and again, as if this was something we
want to hear - which is never the case, I think, unless we are
intentionally doing photocopy music for the accentuated monotony of it,
or because we are really brain dead and can't tell the difference.
In 1992 the TR-909 was old already, but celebrated by techno folks - I
think those who favoured hard, repetitive, mechanical, boring, photocopy
Maybe I could have interested Roland in something juicy, complex,
analogue (apart from perhaps some digital reverb which could be sent
back into the guts of the analogue circuitry to really mess things up).
Maybe a major drum machine manufacturer could have been convinced to
make a machine which doesn't attempt to, or pretend to, slavishly
replicate physical drums. Why name the damn things Bass Drum, Snare
etc. etc. It type-casts the circuitry and does not prod the musicians
to do anything but the reductionist replication of other people's
glories, or at best the first idea which comes to their minds.
It would be better to have circuits which could, generally, do these
somewhat realistic sounds but could then be altered with knobs and
switches - and with external sound inputs - to do totally fresh and new
things which don't sound like any instrument at all. They would use the
original acoustic physical drum sounds and dynamic responses as
*inspiration* for doing a pleasing imitation, and then going well beyond
that into new territory beyond the realm of the responses of existing or
practical physical objects.
Then the sound generation channels could be named in some way which
breaks the ties with physical drums. Name them after wildlife, fish,
opera singers, rivers, 17th century American Puritans, Old Testament
figures or whatever.
I doubt that anyone from inside Japan or outside could convince those at
the helm of the Good Ship Roland, ca. 1992, to steer some of their
energies into such radical, playful, equipment. There's a good chance
that the sales would have been great, since the 1990s saw an explosion
of interest in totally synthetic synthesized music. I think the world
would be richer and more interesting with large numbers of
mass-produced, good quality, Japanese gear with all these quirks and
Even if Roland had wanted to to this, and if a great design was done -
these inherently analogue devices (it would be much harder, even now, to
get the richness, clarity, messiness and complexity with digital
techniques) would have presented perhaps insurmountable problems for a
company such as Roland.
These would not be prohibitive problems, or problems at all, for a niche
Every TR-808 and TR-606 has a different set of 6 frequencies for its
cymbal and hi-hat oscillators. So every one sounds different from the
next. (With an input socket, it is possible to play the mix from one
machine - taken by not pushing the plug in so far as it opens the NC
contact - and to record it or send it live into another machine. Then
they sound the same.)
Some people would be fussed that their machine didn't sound just like
[fill in the big star's name]'s machine - or their friend's machine.
This would generate service calls, returns to shops and customers
auditioning machines from the shop stock until they got one which
sounded right to their ears. Such variability is pretty much
intolerable. Even for a small manufacturer, those fixed differences in
oscillator frequencies could cause unhappiness for some customers and so
for retailers and themselves. There's no easy answer - even if there
were 6 trimpots, customers and even technicians wouldn't know how to
trim their way in a 6 dimensional space to get any given sound they wanted.
TR-808 kick drums have various decay lengths. I think owners frequently
ask technicians to fix this, but there's no spec for what the decay
should be, and the actual decay depends on the precise relationship
between some resistor values, with the resistors being 5% tolerance.
So I would have wanted to make instruments full of the idiosyncrasies
which are anathema to the Japanese aesthetic of mass production and to
their very real needs in terms of being able to efficiently sell them.
It is hard for me to imagine my hypothetical design relationship with
Roland working out well, whether or not I was based in Japan.
A further problem for Roland, or any other large manufacturer with a
global marketing chain and whose customers expect good technical
support, is that analogue gear is hard to maintain. It is hard to
document and hard to fix. It frequently involves trimpots and it is
hard to get anyone to calibrate a machine with more than a few trim pots.
Roland have to make gear which mainly goes out and satisfies customers,
with very few failures. Their profit margins are slim and the cost of a
warranty repair - even if there was no fault and nothing to repair -
could easily wipe out the profit from the sales of multiple instruments.
There are huge development, set-up, marketing, distribution and service
/ spare parts support costs in every product they make, in proportion to
how complex it is and how much it differs from previous products.
As time goes on and the number of people going into electronics in
general declines (due to the vast growth of IT, programming etc.
alternatives) and while fewer and fewer want to do service work, with
more and more time taken up by digital stuff, firmware, etc. it is
harder to get people with the interest and skills to properly debug and
fix analogue electronics. Surface mount makes it harder still.
I think it is kookily heroic and entirely admirable that Korg have
remade the MS-20 in all knobular glory.
However, in general, for large manufacturers, I think the chances of
them making the heroic large analogue gear of yesteryear are over.
(Likewise steam locomotives, Zeppelins and motor vehicles which bark and
throb and are not full of electronics and airbags.)
They never would have made that stuff which we treasure now if they
could have done it digitally in the first place. So we are lucky there
was a window of a decade or so when analogue was possible but digital
was not, which forced them to create analogue gear of elegance and
sometimes of breathtaking complexity. I am thinking especially of the
JP-8, the TR-808, the Prophet 5, OBxA etc and some much rarer
polysynths, especially the CS-80, which is choc-a-block full of Yamaha
custom chips that must have cost millions of dollars in development and
manufacturing set-up costs.
Smaller companies can, in principle, make analogue gear with its costs
and idiosyncrasies - but who among them has the resources to design,
manufacture and properly market really complex gear? Once they do, they
need to find simpler ways of making gear in order to keep the revenues
up and stop driving themselves crazy.
I think a lot of the gear we value is the result of folly, great
inventiveness, desperation (not being able to copy the sounds with a
sampler) and other factors which somehow lead to a sonically attractive,
provocative, unique musical instrument - when the main impetus of the
industry is to crank out easy-to-maintain, easy-to-manufacture,
predictable equipment which does not need much in the way of user
documentation. This is gear which I and other denizens of this list and
of Analogue Heaven generally find not to be inspiring, suggestive or
especially memorable. But such instruments sell well - and people who
can really play, such as Don Lewis, can make great music with them.
More information about the Synth-diy