Things to Know When Repairing Vintage Synthesizers
Tech-tips by Scott Rider (aka the Old Crow)
(taken from http://www.oldcrows.net/~oldcrow/synth/tips.txt )
Disclaimer: the following article assumes the reader is familiar with circuit
board repairs and electrical soldering. While the information described here is
for everyone's benefit, in no way does the author assume any responsibility
for damage done to any equipment as a result of what is written here. If you
have a broken synth, pass this info on to your local keyboard tech.
As the now-vintage synthesizers of the 1970s and 1980s pass into old age
after having been around for 25 years or more, certain preventative
maintenance repairs begin to merit consideration. Case in point: integrated
circuits do not last forever, though some last longer than others. Capacitors,
too, also degrade over time. Keeping a handle on these two issues regarding
one's vintage synthesizer(s) can lead to many more years of useful life for the
machines.
Tip#1 -- Replacing the 4000-Series Chips
The 4000-series complementary metal-oxide semiconductor (CMOS) chips
created in the early 1970s were originally a 6 micron fabrication process (that
is, circuit elements were a minimum of 6 microns apart on the chip surface)
integrated circuit family designed to supply a full set of logic functions, basic
circuit building-blocks and a variety of analog signal switching functions.
Compared to today's chip fabrication which uses microcircuit densities down
to (currently) 0.07 micron, these old 6um chips must seem like bulky
dinosaurs. That fact is--they ARE! RCA CD4000 and Motorola MC14000
series (and all second sources) that made chips in the 70s and 80s from the
original A (unbuffered) and B (buffered) die masters have a life expectancy of
15 years. All 4000-series chips made in 6um process suffer from 'metal
migration', a slow reaction where ions from the metallized gate material of
each MOSFET (metal-oxide semiconductor field-effect transistor) in the chip
flow down to the substrate of the chip over a period of years. The result is a
short-circuit from one or more MOSFET gates to the substrate layer (think of
the substrate as an electrical ground). The end result is a useless chip.
Among the most popular of the 4000-series chips are the analog bilateral
switches (4016, 4066), the analog multiplexer / demultiplexers (4051,
4052,4053, 4067) and the very useful 4046 phase-locked loop (PLL). These
part numbers are used in large numbers in analogue synthesizers as they
allow the routing of analogue control signals either by front panel switches
controlling the chips directly or under microprocessor control as is done in
programmable synthesizers. The 4046 PLL became part of the versatile
oscillator designs in such machines as the OSCar.
The logic functions that define the majority of the 4000-series are also widely
used in vintage synthesizers. The Yamaha DX7, a purely digital synthesizer,
uses a fair number of 4000-series chips in its design. In the Yamaha CS80,
which has no microprocessor per se, over seventy 4000-series logic chips are
used to perform various tasks. This is in addition to the dozens of 4000-series
analogue switches the CS80 already uses.
Similarly, one can find a fair complement of 4000-series chips in about any
synthesizer from the 1975 to 1987 period. The Korg Polysix and Korg Trident
have at least 50 each. The SCI Prophet-5 has more than 40. And these only
scratch the surface when considering all the models of synthesizer made
between 1975 and 1987.
It is my suggestion that *every* 4000-series chip in a machine built before
1982 be replaced if a synthesizer exhibiting malfunctions is being considered
for repair. When replacing chips, I typically draw a diagram of the board(s) on
which the chips reside and label the diagram with the orientation and part
number of each chip. If the service manual is available for a given machine,
I'll photocopy the relevant board diagrams for the same purpose. This is to
keep track of the proper fit of the replacement chips.
I will use IC sockets for most repairs. Sockets allow for a much
simpler replacement of a given chip if it turns out to need replaced in the
future. Caveat: use high-quality machined-pin sockets. These are typically
40 cents each (versus the 5 cent cheap sockets), but they are much, much
less likely to contribute to problems down the road.
Just about all 4000-series ICs can be purchased from popular mail-order
places such as www.digikey.com, www.jameco.com, www.mouser.com,
Newark, Allied, FAI and so on. Typical prices on parts range from 10 cents
for something like a 4001 to $1.75 for a 4067. The best thing to do when
electing to replace all the 4000-series chips in a given machine is to make a
chip list with the quantity and part number of each chip required. Then go
place an order for them as well as the necessary number of machined-pin IC
sockets. Pay attention to the number of pins on the various parts--most of
them have either 14 or 16 pins, so order sockets accordingly.
One more note: some circuit boards in synthesizers amaze me. There is a
28-chip circuit in the Yamaha CS80 comprised of all 4000-series chips and
not a single decoupling capacitor (the TKC board for those who have tinkered
with the CS80). The CS80 is a wonderful machine, but frankly this sort of
engineering is unacceptable. Other boards in the CS80 have decoupling
(though not as much as I'd like to see), but to completely omit it from a circuit
board is bad engineering. When replacing all 28 chips, I will install a 0.1uF
capacitor (see below) across the Vss and Vdd pins of each chip, and a 100uF
aluminium electrolytic across each power rail input to the common circuit
ground. My replacement circuit boards for the
SH and TKC circuit boards "feature" proper decoupling.
The TKC board in the CS80 is responsible for operating the polyphonic
aftertouch (it is the scanning circuit) and 'promoting' a given note's key
pressure value to the proper array of VCAs that comprise the modulation
routing for a given voice. And people wonder why the polyphonic aftertouch is
the first thing to go on the CS80--because the chips have no decoupling,
that’s why!
Tip #2 -- Replacing Old Capacitors
Capacitors do all sorts of things in analog circuits. The most mundane
function, power-supply decoupling, is also one of the most important:
decoupling keeps 'glitch' noise under control that occurs as a result of rapidly-
switching transistors in the logic gates, it prevents high-gain circuit elements
such as operational amplifiers from self-oscillating (a bad thing!), and it keeps
the power rail voltages cleaner overall. The dV/dt aspect of capacitors is put
to use all over an analogue synth: integrators for oscillators and filters,
envelope generators, DC ripple filtering in power supplies--the list is immense.
Capacitors, however, do not last forever. In fact, many capacitors are made
of perishable materials like paper. Over time, the plastic used to seal many
capacitors withers and cracks, at which point the external environment affects
the original value and tolerances of the part. For large-value capacitors such
as aluminium electrolytics, the oil-impregnated paper that forms the insulating
dielectric between the aluminium foil 'plates' dries out, drastically affecting the
capabilities of the part.
This leads to several problems that need attention. First, old filter capacitors
in a power supply may not filter the pulsating DC into smooth DC properly.
These same filter capacitors also absorb AC 'surges'; an old capacitor may
not be able to do this effectively and the surge energy might end up damaging
chips and transistors.
Next, decoupling capacitors may no longer actually be doing their intended
task. Their capacitance has changed (almost always to a lesser value) and
frequencies that were once decoupled may be leaking into the power supply
rails. Less critical but noticeable would be the apparent changes in oscillators
that no longer seem to be stable or have the range they once did, or filters
that distort or seem less musical.
Another place old capacitors can affect performance is in their frequent role
as the voltage 'hold' capacitor in a sample/hold circuit. Sample/hold (S/H)
circuits are used in many synthesizers for the storage of control voltages that
define the parameters in many programmable polyphonic (and some
monophonic) keyboards. If these capacitors degrade, the voltages they are
expected to store reliably begin to drift or droop, which often results in an
audible degradation of the sound of the instrument. Pitches drift, etc.
Replacing capacitors is easier than replacing chips. The typical rules to
follow are to replace a part with the same value part and *usually* the same
type of part. There are two types of parts to avoid, however: mylar capacitors
and tantalum capacitors. There are an abundance of these two capacitor
types in old synthesizers, and they should be replaced first among all the
other capacitors. Replace mylar capacitors with polystyrene or polypropylene
capacitors. Use polypropylene capacitors in the most tolerance-critical areas:
oscillator timing capacitors, tuned filter capacitive elements, and so on.
Polystyrene capacitors can be used most everywhere else for things such as
replacement S/H hold capacitors, op-amp circuitry, etc.
For places where tantalum polarized capacitors were used (The Prophet-5
had lots of these), replace them with aluminium electrolytic parts of the same
value. Tantalum capacitors have a very low WVDC and are the first parts to
die if voltage spikes make it past the power supply defences. In general,
replace old aluminium electrolytics with new ones. Modern capacitor
manufacturing techniques have yielded much better parts, so take advantage
of this. For power supply decoupling on circuit boards, replace the usually
0.1uF ceramic disc capacitors with monolithic dipped-ceramic parts. Do the
same for other values of decoupling capacitors in the circuit.
Lastly--and importantly--replace the filter capacitors in the power
supply. These take a lot of abuse from the AC main line and should be
replaced at least once in the life of a synthesizer.