TB Sequencer Replication by Software (was another @#!! TB thread)

[media at mail1.nai.net]

At 11:13 AM -0700 04/29/98, Sean Costello wrote:
>P.S.  I haven't worked with a 303 in real life, so these are just guesses.
>Perhaps the division of the note isn't by 32nds, but is by some other factor
>that would contribute to the "feel" of the slide.  Or maybe I am completely
>wrong.

The division of the notes isn't 32nds.  The slide is a simple  RC
integrator so the slide time is fixed with an exponential curve.

So, just before someone posts more guesses to this list that are completely
wrong, I'd thought I'd forward this post that I made to another list in
attempt to end the thread here.

You know, cause I'd rather talk about music ;)

PEACE OUT :)
MARK

~~~~~~~~

I don't think this is quite on topic, so I have to mention that I have
played 303 at parties in nyc :)  That being said, I, the greatest living
authority on the Roland TB-303 in this hemisphere, may now entertain your
question :) :)

The following is mostly things I have cut and pasted from posts I have made
to other mailing lists, sorry if it might seem a bit choppy:

The first edition of the service documentation is dated Feb. 19, 1982.  The
TB-303 was made from 1982 to 1984.  It was designed and marketed as a bass
machine to be used along with a TR-606 or TR-808 drum machine in order to
create a complete rhythm section.  As such, it was a dismal failure.  By
1985 dealers were selling them for as low as $50.

By 1995 used originals were selling for as much as $2000.  I think they are
"only" about $1000 now.

While it is true that many of the exact identical parts are out of
production, this should only be a repair issue if its computer breaks.  The
rest of the TB-303 is built from discrete parts and a few generic IC's.
Functional replacements for these parts are still available.

However, one might have trouble finding parts that are just as bad.  They
weren't manufactured to very high tolerances.  TB-303's are almost like
snowflakes.  Of the two I own and the dozen I've played with, no two sound
exactly alike :)

So, while it most likely impossible to build an exact replica, the
availability of parts isn't an issue for building a better clone.  I
believe the primary issue is the cost of replicating its CPU.  There is a
very limited market already crowded with a dozen small European competitors
all trying to undersell the price of a used TB-303.

For purposes of discussion, the 303 can be broken down into two parts: the
sequencer, and the synthesizer.

The 303 synthesizer section is a single oscillator monosynth that contains
a special combination of gimmicks and bad engineering.  As it was intended
only to create bass sounds, this section contains very limited controls
compared to a typical synthesizer.  One unique feature, intended for
emulating slap bass, is that the initial filter cutoff is offset by the
amount of envelope modulation.  It gets its characteristic squawk from its
wonky "18db diode ladder" filter that not will not self-oscillate
regardless of the resonance setting.

However, I severely doubt any of these features are patented, especially
considering that they have been copied rather well in some commercial units
such as the Braintec Transistorbass-3 and the less commercial TBS-303.

The sequencer tells the synth what to play through four values.  It outputs
a gate that determines the length of note and a voltage corresponding to
the pitch of the note played (the higher the voltage, the higher the note).
The sequencer also produces two other values which are either on or off for
a given note: slide, and accent.

The slide switches in an analog circuit (an RC integrator) that creates a
unique portamento such that the time it takes to slide from one note to the
next seems the same regardless of the distance between the notes (ie. it
has the effect of a faster rate for further distances).  This feature
differs from the SH-101 and MC-202 whose portamento times are set by a
front panel knob (ie. the rate is constant for any given knob setting).

The accent has three functions: 1) it opens up the VCA (making the notes
louder)  2) it raises the VCF frequency such that the effect of repeated
accents are partially cumulative  3) it shortens the envelope decay.

The problem with all of the analogue clones is that none of them have a
built-in sequencer.  Therefore, they can not reproduce the unique slide or
accent functions which are very important to the TB-303's sound.  While
most of the knock-offs have tried to implement these features via MIDI
control, none of them have succeeded.  Note that out of a dozen
manufactures none of them bothered to try to include any sort of sequencer
whatsoever.  That's like cloning a pig without the pork!!

At the heart of any sequencer is a computer.  The computers found in modern
TV's and microwave ovens are more powerful than the one in a TB-303.  It is
a uPD650C-133  #15179176.  It is virtually identical to the uPD650C-nnn in
the TR-606 and TR-808, except it has the instructions on "how to run a 303"
burned into it, instead of some other set of instructions.

I'd imagine they'd all do really fucked up things if you swapped the CPU's
around :)

If you were to rebuild it today, you probably put the CPU, memory, and DAC
all on one chip.  The problem is that the cost of mass producing such a
chip is prohibitive to a small company.  No one wants to manufacture an IC
in such small numbers.

Beyond that, Roland is a big company that can still sue whomever they want.
They want to sell their MC-303's or whatever else instead.   Although, I do
not know of any patents, they still might be able to claim that the
instruction set for the sequencer is "intellectual property."

However, it is not a good sequencer in the practical sense.  It is very
limited and difficult to program.  The fact that it is so bad is what gives
a TB-303 much of its unique charm.

The technical limitations of an instrument greatly determine its sound and
how it is played.  For example, a guitar gets its characteristic phrasing
partly because the notes in any chord sound in the order in which it is
strummed.  If it had a keyboard, guitar composition would be vastly
different.  The mode jumps of a trumpet are a distinctive part of its sound
-- if you added a slide it would lose that quality and cease to be a
trumpet.  The same would hold true for adding frets to a violin, or a
reliable interface to a TB-303.

So, even building a complete TB-303 replica would not be enough.  It would
also need a cryptic manual that more than half the users wouldn't receive,
and switches that soon clog with dust and become unreliable, that would
force the user into random styles of programming :)

In the case of the Novation Basstation, I am sure claiming 303 emulation is
merely a marketing gimmick.  I used to own a Basstation.  They could have
easily made it a much better 303 emulator -- I don't even think they were
trying.  Out of all the "clones" it sounds the least like a TB-303.

The most successful clones, Steinberg's Rebirth and the Roland MC-303, are
all digital.  Note that they both contain sequencers.

Remember, the TB-303 was a failure for Roland 13 years ago.  Even if it
wasn't, if you were some ambitious, young Japanese engineer, would you
propose that the company should start selling them again??

As you could imagine, there is quite a bit of stuff on the internet about
the 303.  Check out this website:  http://www.teknet.ch/tb303.htm   It has
grown much in the last two years, when it first started all it had were
quotes taken from posts I had made to Todd Sines' Analogue Heaven.

There are also TB-303 and x0xbox mailing lists.  I'm not on any of them but
I'm sure you could do a web search.

Now, there is even a 303 newsgroup:  alt.music.synth.roland.tb303  (well
that, and alt.urban.legend ;)

Oh, and another thing, jungle sucks :P

PEACE OUT :)
MARK


--
Whole essays could develop from this, in terms of how a few simple
passive components in the TB-303 mimic the distress cry of a living,
sentient creature.
                                        -- Robin Whittle