Mercury-free Switching in Red Welte Equipment

In "Red" Welte pianos and Vorsetzers (at least in early ones), the
switch for the pump motor uses a metal cup containing mercury, and
a metal rod.  When a lever in the spool box is pressed down against a
spring, the rod enters the mercury, completing the circuit and starting
the pump.  A catch holds the rod down in the mercury.

In the take-up spool there is a notch, and this notch has a radial face
which is covered with two layers of felt.  Under the spool there is
a pawl, which presses up gently against the spool.  During play and
reroll, the notch is covered with the roll paper, so the pawl cannot
enter.  At the end of reroll, however, the notch is uncovered and the
pawl enters.

As the spool continues to turn, the radial face impacts the pawl,
forcing it towards the rear of the instrument.  Through a system of
levers, this motion accomplishes two purposes.  First, it releases the
above-mentioned catch, allowing the spring to withdraw the rod from the
mercury, interrupting the circuit and stopping the pump.  Second, it
shifts the transmission into neutral.

Here we must digress a moment.  Yes, on these instruments the
transmission has a detented neutral position, between "Spielen" (play)
and "Retour" (rewind), and this position engages automatically at the
end of reroll.  That makes attaching the roll's leader easier, since
both the roll and the take-up spool can be turned without turning the
wind motor.

This arrangement also has a more important function.  After the radial
face of the notch has impacted the pawl and moved it as far as the
travel of the levers allows, the take-up spool must stop very abruptly.
However, when this happens the transmission is already in neutral, so
the wind motor does not have to stop abruptly.  Since on these
instruments the wind motor turns a flywheel, and since the reservoir
stores energy for a while after the pump is switched off, this system
avoids undue stress on the mechanism and on the roll's leader.  It also
avoids the flapping noise made by many players at the conclusion of
reroll.

Back to the switch: there is also a small rod extending into the spool
box, permitting release of the catch, and stopping the motor, at any
time.

Due to the inductance of the motor windings, an arc will be drawn at
the surface of the mercury when the switch opens.  Being neither a
physicist nor a physician, I am not qualified to comment on any
mechanism by which this could cause a minute amount of the mercury to
enter the atmosphere, or on whether this might be sufficient to
endanger the health of the listeners to the music.  However, there are
those who will worry that it might.  Also, it is possible to touch live
parts of the switch, so there is a shock hazard.

In a modern restoration, it would be possible to eliminate the shock
hazard, and greatly reduce the arcing, by using a low voltage on the
switch, and then using a relay to switch the motor.  Still, mercury is
a toxic substance, and some would prefer not to have an open container
of it in their home.  I am told that some restorers have installed a
modern switch which closes when the rod comes down.  However, this
would detract from an interesting, antique mechanism.

A better idea, in my opinion, is to place powdered graphite in the
cup.  The graphite sold as a dry lubricant for locks and other small
mechanisms is suitable.  However, the level to which you bring the
graphite is critical.  Unlike mercury, graphite is not a liquid, and
if you put too much in, it will not be possible to push the lever down
until the catch holds it.  If, on the other hand, when the lever is
latched down, you can put your finger on it and feel that it goes down
further, then more graphite must be added.  Otherwise, when you take
your finger away and the spring moves the lever back up to the catch,
the rod will lose contact with the graphite.

I would suggest connecting an analogue ohmmeter to the switch when
adjusting the level.  Once the meter can measure some resistance when
the lever is latched down, push it down and release it several times
after each addition of graphite.  You will find that the resistance
varies greatly with each closure.  When I finished adjusting the level,
it varied from about 200 ohms to about 750, but these values are not
critical; much higher resistances are acceptable.  You will also find
that, when the switch is closed and nothing is touched, there will be
minor variations in the resistance with time.  That is why an analogue
ohmmeter should be used; a digital meter may not settle down.

How to switch the pump motor when the switch resistance is so great?
Modern electronics solves this problem.  From a surplus house, I
purchased a small switching 5-volt power supply of the plug-in type
(sometimes called "wall wart").  Across its output I connected the
switch, in series with a 200K resistor.  The switch-resistor junction
was connected to one input of a LM311 integrated circuit comparator,
which was powered by the same supply.

A comparator is a device that compares two voltages.  If the first is
greater than the second, it puts out a voltage.  If the opposite is the
case, it puts out a different (in this case much smaller) voltage.  The
other input of the comparator was connected to a 2 volt reference
voltage, obtained from a voltage divider across the 5 volt supply.

When the switch is closed, even if its resistance is as much as
200,000 ohms, the voltage at the switch-resistor junction will exceed
the reference, so the comparator puts out the larger voltage.  This is
connected to the control input of a solid state relay (also from a
surplus house), causing the relay to close.

When the switch opens, the lower voltage from the comparator is
insufficient to keep the relay closed, so it opens.  The solid state
relay could directly switch power to an AC pump motor.  However, in my
application it switches AC power to a Minarik drive, which supplies
direct current to, and controls the speed of, a DC pump motor.  As
explained in my MMD postings 2001.11.03.03 and 2001.11.06.06, this is
done to switch between the two pump speeds used in the Red Welte
system.

So far I have had the switch in the Welte, the circuitry described
here, the Minarik drive, and the motor described in my MMD posting of
06 November 2001 ["Pump Motor Speed Control"], all working together,
and everything works as expected..  Since the Vorsetzer is not
restored, the motor has only operated without load, but I see no reason
to anticipate problems when it is connected to the pump.

The circuit does draw a small amount of power when the switch is open.
I measured 1.2 ma from the 5-volt supply.  (It would be even less if an
LP311 comparator were used, but so far I have not been able to obtain
one.)  This small drain will not increase the power draw of the "wall
wart" appreciably above its no-load value, and since the unit does not
get noticeably warm after hours of operation, not much power is wasted.
However, if desired, a switch could be used to cut the power to the
"wall wart" when the instrument will not be played for an extended
period.

Some may argue that all this electronics has no place in an antique
instrument.  However, if you do not have the original two-slider
rheostat (as I do not), then there is nothing of much interest in the
compartment to the left of the spool box, and I see no harm in putting
a box containing electronics there, provided it is done in such away
that a future owner can remove it without trace should he so desire.

If you do have the rheostat, then the circuitry described here could be
mounted in a small box, and concealed somewhere in the instrument.  The
object, of course, is to keep undisturbed the compartment to the right
of the spool box, which contains the radial wind motor, the flywheel,
the transmission, and the switch.

Musically,

Cecil Grace, in Western North Carolina