Note the conical ring of grey plastic where I removed the chrome plating with the drill bit.
In theory, doing this will void the TT-303's warranty.
However I can't imagine that any needed warranty support would be
refused if the work is done with care, since (as of early January 2013)
all current production TT-303s have their Volume knob modified in this way. We
will be doing this for all Devil Fish modified TB-303s and TT-303s in
the future. When pulling the knob off the shaft, it is best to
grip the knob with adhesive tape or rubber sheet - don't use pliers or
a screwdriver.
The remainder of this page is written as if the knob is unmodified
- so touching the knob means there is an electrical connection between
the user's body and the aluminium shaft of the Volume pot.
The noise only occurs with particular combinations
of circumstances, which, with care, can be avoided. Also, it is
possible that
a crucial part of the
noise process (high resistance between the shaft and bushing of the
Volume pot) may be greatly reduced by turning it a few dozen times.
If so, then this problem is largely a "teething trouble" of newly
manufactured TT-303s, and will be reduced with use or a few minutes
deliberate knob turning.
This problem, in principle, could affect Devil Fishes and any TB-303,
or probably TR-606 as well. Despite all the work I have done on
TB-303s to date, no-one had ever reported this problem, and I had not
noticed it myself.
The explanation (Gory details below) of the conditions which give rise to the noise are
necessarily complex and will only make sense to someone with a good
understanding of electronics. For those without the time,
interest or expertise to follow the explanation,
here are some things
you can do to minimize or eliminate the problem.
- Turn the Volume pot through its full range a few dozen times. Maybe this will reduce the problem by improving the electrical connectivity between the shaft and the bushing of the Volume pot.
- Run the machine from batteries. This will largely solve the
problem. If you want to run it from a power adaptor, read on.
- Don't touch the volume knob! If you must, wear non-conductive gloves, or read on.
- When touching the Volume knob, make sure you ground yourself by
touching some other grounded object, such as the Tempo knob.
- Even if you don't do any of the above, you may be able to reduce
the noise by making sure the audio out of the machine is plugged into
some other device which is grounded. Also, you may have less
trouble if you use an old-style (heavy and bulky) PSA-240, PSA-120 etc.
9 volt DC regulated power adaptor instead of the modern (lightweight,
compact) switch-mode adaptors such as those supplied with the TT-303,
and the current production Boss PSA-240S, PSA-120S etc.
Update 2012-12-21 with news from Cyclone Analogic:
I have been informed that future shipments of the TT-303 will be with old-style (heavier, simpler, bulkier) non-switch-mode power adaptors and that most distributors have, or will have, replacements for people who have already purchased a TT-303.
I haven't seen these adaptors but
I think it is likely that they will have less capacitive coupling than
switch-mode adaptors and so cause less trouble. The problem is
not really the adaptor - it is just that the pots, at least before they
are turned a lot, don't connect the shaft to ground properly and that
with a metal knob and the machine ungrounded and powered by an adaptor,
there can be a voltage on the pot shaft when someone touches it.
There could be other problems with pots in general, such as noise when
they are turned, or in a particular part of their
range. This should not happen with new
equipment. If it happens with old pots, I find it is generally
sufficient to clean the pot with isopropyl alcohol (IPA). This is
best done by smoldering it and dismantling it. However, in the
case of the many pots (other then the Volume pot, which I dismantle) of
the TR-808, I squirt in IPA, turn the pot a dozen times or so, and blow
the IPA out with compressed air while turning the pot some more.
So if your TB-303 or TR-606 has a noisy Volume pot in ways which are
different from what I describe below, you might find that squirting in
IPA and using compressed air, with dozens of turns, improves it and
perhaps fixes the problem for a decade or so.
Under no circumstances should anything other than isopropyl alcohol be squirted into pots or rotary switches. (Don't bother trying to clean tact switches - see the Devil Fish page for replacing these with sealed Omron switches.)
(The following is written with the
assumption that the knob is unmodified - so touching the knob connects
the user to the aluminium shaft of the Volume pot. It is also
written with the assumption that the Tempo knob is unmodified and that
that pot's shaft is securely grounded, which is not absolutely
guaranteed, since it may have similar high resistance to that of the
Volume pot, depending on how it is pressed, how much it has been turned
etc.)
There's no easy way to describe what I
have observed. This does not necessarily explain problems in
machines I haven't looked at, but if you understand the following you
will probably be able to establish whether the problems you are
observing are explained by the mechanisms I present here.
Unless otherwise stated, what I write below applies to TB-303, TR-606
and TT-303. I have not dismantled a TT-303 Volume pot, but it
looks much the same as the ALPS Volume pots used in the TB-303
and replacement pots supplied by Technology Transplant which are
made by some company other than ALPS. I am confident that the
TT-303 Volume pot has the same general construction details as the
original ALPS pots. While there may be overall differences in the
occurrence of problems with these pots, my initial experience of
observing the same in-principle (if not degree) problem with a
brand-new TT-303 and with a 30 year old TB-303 indicate to me that the
problems are general to this design of pot, in all the circumstances of
how they are used, which is primarily a matter concerning power
adaptors, grounding of the TB-303, TT-303 etc. and to what extent the
user is grounded or subject to capacitive coupling with the mains.
The
Volume pot has several components of interest:
- A metal case and mounting bracket, comprised of several pressed
pieces and a die-cast metal bushing. The bracket is soldered to
the main PCB and is securely grounded there. This means the
entire case and bushing is securely grounded too.
- An aluminium shaft. This is supported by the die-cast
bushing and the bearing between the two is lubricated with
grease. This grease is presumably not intended to be electrically
conductive, but if it was consistently electrically conductive to a
high degree, then the problems mentioned on this page would not occur
to any significant degree.
- Inside the pot is a phenolic board with three metal terminals,
conductive tracks and a resistive track. The resistive track runs
from the anticlockwise (ACW) terminal (via a short conductive track) to
the conductive track which connects to the clockwise (CW)
terminal. The centre terminal goes to two metal pieces which rub
against the ring of the rotating wiper as described below. The
TB-303's Volume pot is 50k "log" pot - "logarithmic" being an imprecise
shorthand for the "audio taper" arrangement where small changes in
resistance happen initially, with larger changes towards the CW
end. I haven't measured the resistance of the TT-303 pot but will
assume it is the same 50k. The "log" curve is created by screen
printing two or so extra layers of conductive material towards the ACW
end.
- Mounted on the bottom end of the shaft is a molded plastic piece (nylon?) which holds the metal wiper arrangement.
This has an inner ring which the abovementioned two metal pieces rub
against with a sliding motion as the pot is turned. This ring
connects to the wipers themselves which are fine and flexible.
These are moved from ACW to CW ends of the resistive track. At
each end of the movement, the wiper is touching the conductive track,
rather than the resistive track, so the resistance from the wiper to
the CW or ACW terminal is very low.
The ACW terminal is ground. The CW terminal is driven by a relatively low impedance audio signal.
The wiper drives the output amplifier - for Audio Out and to drive the Headphone Amplifier.
The AC and DC impedance of the wiper (ignoring the impedance of the amplifier it is driving) depends on where the pot is turned:
- When ACW, it is theoretically connected to ground. However,
the real impedance would be composed of the wiper-to-track resistance
plus the two-metal-pieces to ring resistance. For our purposes we
will assume these add to zero, or to such a low impedance that we can
regard it as zero.
- Similarly, when CW, there is an ideally zero ohm connection
between the wiper and the CW terminal, which is driven by a low
impedance audio signal.
- Anywhere else, the wiper is touching the resistive track.
The impedance will be the resistance to the left to the ACW terminal in
parallel with the resistance to the right - to the CW terminal.
If the pot really is 50k (it could easily be 45 to 55k - these
resistances are not critical or tightly controlled) then at some point,
to the right of top centre (since it is a log pot) there will be a
point where the resistance on either side is 25k, so the resistance
(AKA impedance - I am using the terms loosely and interchangeably)
would be 12.5k ohms to a combination of ground (ACW) and low AC
impedance (CW). Elsewhere, the impedance will be less than 12.5k.
In what I observed, all the noise resulted from the following basic chain of events:
- There was an AC voltage between the shaft of the Volume pot and
its bushing, which means that with respect to the ACW terminal (the
ground of the machine) it had an AC voltage on it which was big enough
to cause audible trouble according to the following:
- There was capacitive coupling (and perhaps some resistive
coupling, but the resistance should be hundreds of megohms, since this
is a piece of plastic) between the shaft of the Volume pot and its
wiper. This is primarily or solely through the (nylon?) molded
plastic part which supports the wiper mechanically and isolates it
electrically from the shaft.
- This coupled an AC noise current to the wiper, which created an
AC noise voltage on the wiper as long as the pot was not at its ACW or
CW ends. The greatest noise voltage for a given current would
occur when the wiper is somewhere to the right of centre, where it has
about 25k ohms to the left and 25 k ohms to the right.
- We hear this noise voltage after it is sent to the Headphone Amplifier and/or the Audio Out socket.
Points 2, 3 and 4 above follow naturally from the occurrence of point 1
above, as long as there is some connection to the pot shaft which
imposes a voltage on it with respect to the "ground" of the machine
(the ACW terminal). There's no way of making pots in general, or
these kinds of pot in particular, without capacitive coupling from the
shaft to the wiper.
The shaft is lubricated by grease which is not guaranteed to connect it
electrically to the bushing. I have never seen a pot where there
was an explicitly designed arrangement to electrically connect the
metal shaft to the bushing.
It had never occurred to me that there might be a voltage on the pot
shaft with respect to the ground of the machine, but this is what I am
observing.
With our TT-303, the noise problem only occurs (written in the present
tense, but see below) when I press downwards on the knob. This
must disconnect the shaft electrically, by altering the conditions
where the aluminium shaft rubs against the die-cast bushing. With
our particular pot I found the DC resistance was typically 0.15 to 0.3
ohms or so, but would exceed 10 megohms and sometimes exceed 20 megohms
(the limit of my digital multimeter) simply by pressing down on the
shaft, quite lightly. This switch to near infinite resistance was
easily repeatable and didn't seem to be altered by the rotational
position of the shaft. If I pressed firmly, the resistance would
decline, usually to a few ohms or a fraction of an ohm.
This Volume pot had only been turned a handful of times.
I turned it quickly through about 30 full cycles.
Now I find it much harder to get a high resistance!
Now, in some, most or all turning positions, I find it difficult or
impossible to get a resistance above a few ohms no matter how hard I
press or wiggle the shaft. I can get it to go high sometimes by
pulling the shaft upwards.
On this admittedly limited experimental basis, I tentatively conclude
that this high-resistance problem with the TT-303 Volume pot is likely
to become much less of a problem after the pot has been turned
repeatedly. Presumably the grease is more thinly spread and/or
there are metal particles in the grease which make it more likely to
electrically connect the shaft with the bushing as it is rotated and
moved with pressure from various directions.
I tried the same experiment with a 30 year old TB-303. It was
more difficult to get a higher than fractional ohm resistance, since
the resistance depended more on pressure and rotational position than
with the TT-303 pot. However, I was at times able to get
resistances of several k ohms, since the meter would sometimes read
2.53 k ohms, or similar, for a single measurement cycle.
This pot is from ALPS (I assume the TT-303 pot is by another
manufacturer). Both pots have in principle the same
problem. The TB-303 pot has it to a minor degree. The
TT-303
had it to a much larger degree. After turning it a few dozen times, the TT-303 pot has the problem to a
much lesser degree than before these turns.
Circumstances leading to voltage between the pot shaft and the ground of the machine
If the shaft was made of plastic, or the knob was non-conductive
plastic, none of the problems discussed here could occur. (The
only other pot which handles audio in the TB-303 or TT-303 is the
Resonance pot. It has a plastic shaft, so these problems cannot
occur.) However, the shaft is made of metal and in both the
TB-303 and TT-303, the knobs are chrome-plated ABS plastic - so they
are conductive too. (BTW, it would have been difficult for both
Roland and Cyclone Analogic to get such a great job done of this chrome
plating. The Roland knobs and buttons frequently look great after
decades of use. Hopefully the same will be true of the Cyclone
Analogic buttons and knobs.)
None of these problems will occur
until the user touches the Volume knob with their skin - or in some
other way which forms an electrical connection.
None of these problems will occur if
the Volume pot's shaft has a low resistance, such as less then 50 ohms
or similar, to the bushing.
None of these problems will occur if
the user's skin contact with the knob causes no voltage on the shaft
due to their body being at the same voltage as the machine itself.
Before considering the various circumstances which may lead to trouble, here is the
bad news about power adaptors in general.
(This does not apply to the power adaptors used for laptops, since some
of them have a ground terminal and it is possible that their outputs
are either directly or at least largely capacitively coupled to ground.)
In general, power adaptors do not have a ground pin. Therefore
their output (I am assuming 9 volt DC regulated adaptors, which is what
we must use for TB-303s, TR-606s and TT-303s) is capacitively coupled,
to some degree at least, to both the Active and the Neutral
lines. In the TB-303 and TR-606, and I guess in the TT-303,
the 9 volt regulated voltage becomes -3 volts and +6 volts with respect
to the ground of the machine.
The following is written for the nominally 240 volts 50 Hz mains supply
in Australia. However, I think it is actually specified to be 230
volts or so, even though it may be more. In the USA the mains
voltage is 120V or so, at 60 Hz, so the same things apply in principle,
to a lesser degree.
If the output was equally capacitively coupled to both the Active and
the Neutral, then with no other loading (resistive or capacitive) then
the voltage capacitively coupled to the output leads (which become -3
and +6 volts) will be exactly 120 volts AC - a 50Hz signal.
In the USA this would be about 60 volts AC at 60 Hz.
In an old-style (non-switch-mode) adaptor, the capacitive coupling was
purely between the primary and secondary windings of the 50 or 60Hz
mains transformer, with the metal core of the transformer possibly
being an intermediate in this coupling. There would be no shield
between the two windings, since there is no ground to connect it to and
it cannot be assumed that one mains pin or the other is neutral.
In modern switch-mode adaptors, there would be a similar coupling,
however, the primary is connected to a rectified and smoothed version
of the mains and the primary itself is being turned on and off at
somewhere between 20 and 60kHz. The exact frequency may fluctuate
and the drive may even turn on and off, as part of the
regulation. The duty cycle and the frequency may be unstable and
would change with load and probably with the 100, 50, 120 or 60Hz
variations in the rectified and somewhat smoothed mains voltage.
This is a very intense form of AC coupling - such high voltages at such
high frequencies. I guess that some, many or all adaptors are designed
with extra capacitors to reduce this high frequency coupling, as
described in the next paragraph.
It is likely or at least possible (I don't feel like pulling apart adaptors) that modern switch-mode adaptors have
a capacitor from the output lead (say the negative lead, which has a large capacitor to the positive lead anyway) to
both of the mains terminals.
Such capacitors would reduce the drive to the output wires from the 20
to 60kHz high frequency switching. Since that switching is not
sinusoidal, and would have many higher harmonics, it would not surprise
me if the designers of these switch mode adaptors would need to include
such output to mains-pins capacitors in order to meet electromagnetic
noise emission standards.
The adaptor supplied with the TT-303 is a modern switch-mode
design. The one we received has two round pins matching the
Europa's standard:
http://en.wikipedia.org/wiki/Europlug. Here is a picture of it:
It is an LAY-186, made in China by an unknown manufacturer. Its
input is "100 - 240AC" and its output is 9 volts DC, with a maximum
current of 300mA.
There's nothing wrong with this adaptor which isn't also wrong with
most or all other modern switch-mode adaptors. This and most or
all other such adaptors have a high degree of capacitive coupling from
the mains pins to the output. Here is a scope trace of the output
(the positive terminal, but the negative would be the same) with a 10
megohm probe:
This is 50 volts per division. These are 150 volt peaks - 300 volts peak-to-peak!
Loading this down by touching it with my finger slightly reduced
it. Without the 10 megohm load and the scope prober's (~20pf I
guess) capacitance, the voltage would be higher still.
Exactly why the signal is this shape I have not investigated in
detail. Suffice to say the adaptor rectifies the mains in some
way and so it doesn't surprise me to see a non-sine signal on its
output. Note that there is no visible 20 to 60kHz signal.
There would be (I think) if not for the capacitors which I hypothesise
exist between the output and the two mains pins.
I don't have a PSA-240 here, but I tried some other old-style
non-switch-mode adaptor and the voltage was a pure sine wave about 70
volts peak-to-peak.
Here are various scenarios, from the worst to the least problematic, in
which there could be noise problems. These noise problems will
only occur if (as I observed above) the high shaft-bushing resistance
problem is occurring. If this problem has been reduced by the
Volume pot having been turned a lot, then these problems will generally
be worse with new machines and get better with time - or they may be
greatly reduced by turning the pot a few dozen times. However, I
can't reliably predict the behaviour of pots I haven't seen - and I
have only observed this with one pot.
Scenario 1: switch-mode adaptor, ungrounded machine
Assuming (as seems reasonable) that the
switch-mode adaptors have a stronger capacitive coupling between the
mains pins and their output, then the worst scenario is when one of
these is used and the machine is not connected electrically to any
other systems. (MIDI In or Out doesn't count, since MIDI is or at
least should be, electrically isolated, by the use of an opto-coupler
at the receiving end.)
I assume the user is listening with headphones and that their body is
not electrically grounded. However, if it is, then the noise
would be worse, as I explain below.
The adaptor drives the machine up and down with a voltage similar to
that shown above, which could easily be 50V RMS or more. This is
driving the capacitance of the machine in space (generally to ground,
but also so some degree to mains wires), and the headphones leads and
headphones, as capacitively coupled to the body of the user. The
user's body is capacitively coupled to ground and also to some degree
to mains wires.
When the user touches the knob, current will flow through the shaft due
to the voltage of the machine being such a high (50V RMS or more) AC 50
or 60 Hz signal and the body of the user being (if it was not touching
the Volume knob) at a lower voltage.
So current is coupled from the active mains wire, through the adaptor,
to the "ground" of the machine, and through the volume pot's shaft to
the user - where it is capacitively coupled to space in general, which
is primarily ground. There is a variable resistive coupling
between the shaft and the bushing, and the problem only occurs when
this is high enough resistance to allow a significant voltage to appear
between bushing and shaft as this current flows into the knob and the
user's body. The voltage between the shaft and the bushing would
be lowered firstly by a low resistance and secondly by a high
capacitance. But the capacitance of the greased bearing gap is
likely to be only a few tens of picofarads - a small value compared to
the capacitance of the user's body, which is likely to be in the
hundreds of picofarads.
To the extent that current does flow into the user's body from this
imposed 50 or 60Hz signal, and to the extent that this develops a
voltage between the shaft and the bushing (due to high resistance and
low capacitance between the two) then we have a problem with the shaft
of the pot having a noise voltage on it with respect to the "ground" of
the machine. This is capacitively coupled to the wiper as noted
above and the user hears it.
This problem would be worse if the user is grounded in some way -
because then their body's voltage would not float with the imposed
current. This would increase the current and so increase the
voltage between the wiper and bushing.
Grounding the machine would stop the AC drive to the user via the
Volume pot knob. However, the user's body (if not otherwise
grounded) would still probably be coupled to the mains wires in the
room - see Scenario 3.
Scenario 2: non-switch-mode adaptor, ungrounded machine
The same as above would occur, but
probably to a lesser extent, assuming the old non-switch-mode adaptors
had a lower capacitive coupling between the mains pins and their output
leads.
Scenario 3: Batteries, switch-mode adaptor or non-switch-mode adaptor - grounded machine
In this scenario there is either no AC
drive to the "ground" of the machine (when operating from batteries) or
all such drive (from an adaptor) does not alter the voltage of the
"ground" of the machine, because it is connected to real ground (some
presumably grounded audio system) by one or more cables. The most
likely one is the Audio Out lead. However CV cables or Audio In
will do the same job.
In this case, there's no AC drive of the user's body from the "ground"
of the machine, since the machine's "ground" voltage really is ground.
However, the problem can still occur due to the user's body not being
grounded and being capacitively coupled (through space) to one or more
mains wires. In this scenario, the pot shaft can have a noise
voltage imposed on it by the normally noisy and otherwise floating
(ungrounded) voltage of the user's body. This is what happens
when we touch an input to an audio amplifier with our finger - unless
we are grounded, and assuming this is in a mains setting (not running
from a car battery far from mains, inverters etc.) - we hear a loud
hum, which is from our body being capacitively connected to one or more
mains wires through space, or worse still if we wrap our hands around
the insulated outer part of a mains cable. The capacitive
coupling of noise from the mains wires to the user's body drives a
current onto the Volume pot shaft, which (if there is a high resistance
between shaft and bushing) is capacitively coupled to the wiper, so we
hear it.
Scenario 4: Batteries and ungrounded machine
This is the least likely scenario for
noise problems to occur. Still, they could in theory occur due to
different capacitive couplings between the machine with respect to
ground and mains - imposing voltage X on the machine - and between the
user's body and ground an mains - imposing voltage Y. To the
extent that these couplings differ, with their consequently different
voltages X and Y, when the user is connected to the Volume pot shaft, a
current my flow and this may involve the pot shaft having a noise
voltage with respect to the ground of the machine, which is
capacitively coupled to the wiper.
Speculation . . .
Maybe the TT-303 Volume pots have poorer electrical connection between
shaft and bushing than the ALPS pots in the TB-303. There's no
way of telling for sure, since it appears that at least for the TT-303
pots, this problem may diminish with multiple rotations. There's
no pristine TB-303 Volume pot to test.
However, maybe there is no significant initial difference between these
two types of pot - and likewise no significant difference between their
much improved (lower) resistance between shaft and bushing once they
have been turned quite a lot.
This problem was never, to my knowledge, reported by TB-303
users. Two users have reported it to me - or at least reported
something which may have the same characteristics and causes of what I
observed in our TT-303.
Even if the TT-303 pots are fundamentally the same as those in the
TB-303 in respect of their pattern of resistance between shaft and
bushing, as it varies with use, force, rotation, side-to-side pressure
etc. then this recent flurry of reports, (in the absence of any such
flurry for the newly introduced TB-303 as far as I recall) could be
explained by some combination of the following:
- In the 1980s and most of the 1990s, all power adaptors were
non-switch-mode and would have had generally low capacitive coupling of
the mains to their outputs.
- The people who bought TB-303s were generally not as fussy about
sound quality as those who are now buying TT-303s. Since then we have had
nearly 30 years of CDs and other digital audio systems as a
reference for "good quality".
- The TB-303 was not initially used much for recording, whereas
virtually everyone with a TT-303 wants for recording.
- I knew only a handful of TB-303 owners then. Now, via email
and in particular mailing lists, I am likely to hear of difficulties
encountered by any one of dozens or hundreds of new purchasers or
TT-303s.