I saw Andy Taylor's post and could not hold this in any longer.  Sorry
if it's a bit heavy, but I have simplified as much as possible,
missing some techie bits out!

 [ Andy wrote in Digest 980721: ]

> A solenoid is the weakest at it's "open" (down) position; [it]
> reaches its maximum power when the plunger is _even_ with
> the electromagnetic field, which is usually all the way up.

There is a way of improving this electronically; many dot matrix
printers (look how fast they are!) use switch-mode solenoid driver IC's
which run off, say, a 50 vdc supply to a 6-volt solenoid.

At turn-on the full 50v is applied to the solenoid (I am talking micro-
seconds now).  The current ramps up (at a rate determined by the
solenoids winding inductance, resistance and applied voltage) at about
8 times normal rate, and when a pre-selected current is reached (e.g.,
4 amperes) the IC turns off for a few microseconds, and then on again
until the current ramps up to 4 amps, and so on.

This, then, controls the average current to the solenoid.  Left like
this for long it will burn out, so the 4 amp "set point" is reduced
with time (over a few milliseconds) so that it ramps down to a current
just sufficient to hold the solenoid closed.  All this is complex, but
the advantages are :-

1.  When solenoid is open a "burst of power" gets over the large magnetic
gap and system momentum, causing a much faster and powerful response;

2.  Heating in the solenoid can be lower because once the "power burst"
is over the power drops to a fraction of normal, just enough to hold it
shut.

3.  On an existing system (e.g., pipe organ) valves can be opened up a
bit, allowing easier air movement and hence being faster still.

4.  If the ramp rate is set correctly the solenoid will come to rest
gently (or variable for piano/drum) and make less noise.

More info on a typical ICs at

     http://www.farnell.com/datasheets/5144001.pdf

> The primary problem with solenoids is, after a key strikes, there
> is  some residue of magnetism still in the solenoid that makes the
>  key a little sluggish to return.

I think it may be the control, not the solenoid.

If you take a solenoid and put the wires on and off a battery by hand,
when you disconnect a wire a small spark can be seen and you may get a
shock!

This is caused by back electromotive force (back EMF).  Simply put, at
the instant of turn-off there is a magnetic field holding the solenoid
shut; this has to "collapse"  _before_ the solenoid can start to open.

Now, this magnetic field is energy, and as it collapses it generates
electricity in the windings of the solenoid and comes out as (a) high
voltage, low current and sparks at switch contact,  _or_ (b) low
voltage high(er) current and goes though a diode across it and dissi-
pates as heat in the windings, (a) or(b) according to the whim of the
driver circuit designer.

The magnetism in a solenoid is caused by current flowing through the
turns, and since a fixed amount of energy is in the solenoid (same for
(a) or (b)) and energy = amps * volts * time, something has to give
between (a) and (b).

To make this clearer let's put some example numbers in and see :-

(a)  0.0024 = 0.012A * 400V * 0.0005 sec
(b)  0.0024 = 0.04 A *   2V * 0.03 sec

As you can see (a) is quite a lot faster (25 ms) than (b)!  And remember
that a musician can apparently hear a timing error of 8 milliseconds.

So why do we now have diodes across our solenoids ?

The problems with (a) (sparking contacts at 400 volts) are switch
contact erosion, radio interference, and coils breaking down due to high
voltages.  When semiconductors become available most transistors were
low voltage and could not cope with the back EMF, so a diode was
used.  It cured all four problems and become the de facto standard.

I have done a number of tests using solenoids breaking light-beam
switches driven by 1200v rated transistors (IGBT's), with the results
stored and plotted out from an oscilloscope.  Off-delay differences of
up to 40 ms were noted and, like most things, a compromise is needed
to cure it.

Throw away that diode and put a resistor across the coil: low wattage,
500 to 2k ohm area, with the value determined by the peak voltage your
driving circuit can stand.  Change your driving transistor for a modern
400v type (about 1.5 times cost 50v type) and this will make all the
difference.

If speed is really important, I use a little circuit which discharges a
capacitor into the coil first and then applies a holding voltage.   (It
needs two supplies, one R, one C and a diode.)

Let's see how fast rolls CAN be punched!

Andy Saunders
Ipswich Suffolk

 [ Or, in Andy Taylor's situation, let's see how fast the solenoids
 [ will operate the piano keys!  Does anyone have typical voltage
 [ _and_ current oscillograms from a solenoid piano which could be
 [ placed at the MMD web site?  -- Robbie