Some guesses on the self-tuning piano:

> Hi All,  As a dealer of QRS products for over two decades, I am
> quite interested in the new "Self-Tuning" piano. However, a few
> questions do come to mind.
>
> What happens if 'little Johnny' happens to drop one of his metal
> play toys on the strings?

I suspect that there's a routine in the microprocessor that times the
tuning process.  If everything isn't tuned right after, say, 60
seconds, the tuning routine shuts down and lights a warning light
somewhere.

> What happens if someone spills a drink in the piano?

Probably nothing more than what would happen in a conventional piano.
Most drinks aren't very conductive anyway, and the voltages necessary
to heat piano wires are quite low.

> What happens if 'Puss', the cat, happens to jump inside?

I was wondering about roasted hamsters, myself.  Again, the voltages
and temperatures involved aren't very high.  The NY Times article states
that the piano is initially tuned at the factory at 95 degrees.  While
this must be a real treat for the piano tuner, it's hardly damaging to
living tissue.

> These are not uncommon occurrences.

> What happens if a string breaks and 'shorts out' a dozen or more
> other strings?

Again, I sure hope that the programmer included a time-out feature in
the microprocessor's ROM.

> How "hot" is hot?  And is there a point at which the whole system
> shuts down because of an overload??

Ninety-five degrees (presumably F) or so.  The Times article doesn't
say anything about disasters.  However, it's pretty easy to put a
current monitor in the circuitry.  Must have been a big treat to
arrange all those current switches.  The current would have to be on
the order of ten amperes, and there are a lot of strings in a piano.
Electronic switches that can handle substantial currents are not cheap,
but they'd need one for each string.  My guess is that they also use a
current pulse through the magnetic pickup to vibrate the string during
the tuning process.  That'd take some extra switching, too.

> Who repairs the system?  A computer technician or a piano tuner?
> Will a piano tuner who works on these pianos have to be cross-trained
> to work on computers?

God only knows.  If they were smart, they'd have included a lot of
self-diagnosis tools in the system.  One question is whether they're
going to make the diagnostic codes, etc, a big secret or allow mere
mortals to understand them.

For what it's worth, I don't think that the computer part of these
pianos amounts to anything more than a one-chip microcontroller plus
some sort of readout and perhaps some test buttons.  All of the
software would be written into the microcontrollers integrated circuit.
There's nothing much to repair in such a system, which is both good and
bad: you cannot do anything for the instrument except to replace a
defective microcontroller, and that involves expense and delay.  Then
again, such systems are extraordinarily reliable.

If there was going to be trouble, I'd anticipate that it would be in
the power supplies that supply current to the strings.  Another
likelihood would be problems with the contacts between the power supply
and each end of the string.

> Can I leave the piano in my home at the shore during the winter
> when the heat gets turned off?  Can I unplug the piano and expect
> it to 'retune' itself once the piano is plugged in again?

The firm seems to imply that you can.

> Who is the tuner who will establish the information that goes into
> the tuning [memory] chip?

Initially, that guy at the piano factory with the tuning wrench and the
swim trunks.

> If the system can only be as good as the information that goes into
> the chip, how much better can it be than a regular piano that is well
maintained?

Well, it wouldn't be.  The idea is that you won't have to tune the
thing on a regular basis.  If you'd be tuning the instrument regularly
anyway, it wouldn't be a useful purchase.

> And what about the customer who likes his/her piano tuned a certain
> way?  When it comes to tuning is there really a "one size that
> fits all"?

I think the article states that any number of tuning schemes can be
accommodated by the system.  This seems to imply that there must be some
way of writing a new tuning into some sort of a memory.  Though I can't
imagine how the hell you're supposed to tune the thing in someone's
home at 95 degrees.

> And finally, is all of this exactness really worth the complexity or
> the cost?  Sure, it might be very novel, but there have been lots of
> novel products that did not survive the marketplace.  I sure hope QRS
> isn't wasting their stockholder's money on a "Tucker".

There do seem to be similarities.  There is something very 1885 about
the idea of a self-tuning piano.  I can almost see the Victorian
woodcuts in the advertisements.  I don't have any idea whether it would
sell these days or not.  If the seller is smart, he'll keep the price
reasonably low.

> A closing thought: If tonal purity is so important, why not just
> make a piano with one string per note?  It can't get any more "pure"
> than that!

I think that the instrument tunes itself like a real human tuner would:
one string at a time.  Thus the two or three strings that share a
hammer would not necessarily be tuned to the same pitch.

Here's a bit of speculation on how the instrument would be arranged
(I have no more specific knowledge than anyone else, but this is how
I might design such a device)

For what it's worth, I rather doubt that the added electronics will be
either bulky or intrusive.  The most noticeable feature would probably
be a set of guitar pickups that listen to each string in the
instrument.  I imagine that the pickups would be connected with
computer-style ribbon cable which would run to a small circuit board
somewhere.

On the assumption that the tuning pins are insulated from the frame,
there would have to be a rotary contact on each pin such that the
string could be connected to a fairly heavy wire.  These wires would
run back to a power-supply board probably located near the
microcontroller circuit board.  The frame would form a common contact
with all the non-pinned (?) ends of the strings, so there would be a
cable running from it to the power supply as well.

To tune a string, the microcontroller (i.e., the computer) sends a
pulse of current to a magnetic pickup (just a coil of wire located
close to the string).  This vibrates the string by giving it a bit of a
tug.  The microprocessor then listens to the waveform produced by the
pickup and compares it to a waveform for that note that is stored in
its memory.

If the frequencies are unequal, it connects a high-current supply to
the string and sends a certain amount of electric current through.
Then it pulses the pickup and listens once again, adjusting the current
each time to bring the string waveform, as derived from the pickup,
closer to that which is stored in memory.

The computer then remembers the amount of current that it needed to
bring the string into tune, and then goes on to the next string.

The trick here is that the current must be maintained through the
strings while the instrument is being played.  This is probably done by
supplying the current in the form of fixed-current, fixed-time pulses
of varying lengths.  The longer the pulse, the greater is the heating
of the string.  (It is the length of the pulse that would be stored in
the computer memory.)

Thus a single power supply could be switched onto each string in
succession to supply it with the appropriate heating current for as
long as necessary.  The power supply would be able to service each
string frequently enough to prevent it from cooling down significantly.

This all sounds complicated, but once you've figured out what must be
done, it's fairly straightforward to program into a computer memory.

Mark Kinsler