Several people have wondered about Ampico expression response curves
and have written to me privately about them, in regard to what they
have heard or read. I probably haven't expressed myself very well
about these dissimilarities and the actual real-world response, so
let me simplify the two model Ampicos this way:
It may be simple to clamp the crescendo, raising pump pressure to
the maximum, and then plot the resultant dynamic curves, but it's
inconsequential since both models of Ampico were designed to operate
their own pump proportionally with their crescendo. The idea is
exactly the same as charting the performance of a Porche by laying a
brick on the accelerator and then logging its performance around the
track as you shift gears! What has been proved?
The only thing less meaningful would be then to compare the Ampico B
curves with its amplifier clamped to the Ampico A the same way.
Remember the basics -- if it were necessary that the Ampico B be
compatible with the model A, Ampico would have made sure that it was.
If, instead, it was necessary that the new Ampico be compatible with
the Ampico standard roll coding format, then it would have to be
compatible, else it could not compete very well with all the new
innovations from Aeolian, and there would be no reason to "trade up,"
or buy the new Ampico, would there?
I cannot answer the many, many reasons why certain Ampico B's don't
play A rolls well, but I can say this: if they didn't, originally,
Ampico would have never let them out the door, because the vast bulk of
everybody's collection at the time was still model A rolls, and either
they were to be delighted, or they would get their money back! Simple
economics demanded no less.
I have no recollection of anyone ever claiming that the Model A and B
were compatible, by the way. If someone here can ever recall anyone
writing that, I would appreciate them letting us know and the reasons
given for the comparison.
There seems to be three overriding reasons why a few souls believe that
if the two systems aren't compatible, then the B won't play A rolls
well. It isn't because someone has CLAIMED they are compatible. It is
because there are still some rebuilders who cannot understand HOW a
model B could ever authentically reproduce Ampico A rolls and not be
compatible with the A piano. That is the real sticking point.
So the "question of compatibility" begins and ends in the minds of
those who say, "either the pianos are compatible or they cannot play
equally well." If there is this perplexity then, it is within their own
mind. Here then are the three most misunderstood mechanical
differences between models and how it is so simply resolved:
1) The model A has two crescendos while the model B has only one.
2) The model A's intensities can also raise pump pressure, while the
model B cannot.
3) The model B's crescendos are twice as fast as the model A's.
1. The answer to this seeming disparity of two separate crescendos is
ultimately in the standard roll codes used. But many years prior to
the introduction of the new Ampico, Charles Stoddard himself admitted
that two crescendos aren't needed because they are both "joined at the
hip anyway" by the pump amplifier. Whenever stack pressures reached
15," the crescendos were, in effect, "connected together anyway through
Nuances created for the intensities are easily done together for both
bass and treble, and as Dr. Hickman said, "Well, Mr. Stoddard did that
(decided to have one crescendo instead of two) because you still had
the bass crescendo but it was tied into the other. In other words,
they were both on crescendo at the same time." Dr. Hickman also said,
"when in the devil is it necessary to crescendo the bass without the
treble. Why have it separate?" Of course, the answer is, "Never."
2. The next objection is in regard to the tendency of the model A to
raise pump pressures with its intensities alone -- which the model B
cannot do because its intensities are separate and not fed back to the
pump spill pressure. So what about this?
If rolls had been coded to raise pump pressures with intensities, then
they would not have those intensities available to use for accents,
because of the time required by the pump amplifier to react. Those
intensities would be required to just stand there for a considerable
length of time, musically, and be wasted. Intensities, therefore, were
to be used mainly at the point of accent and crescendos were to _both_
"nuance" and to "escalate" the pump, providing a platform to raise and
lower discrete intensities.
This is the _only_ way the A piano is able to have a "Normal" or
"Medium" setting, distinct from the "Brilliant" setting. It is
strictly the use of a crescendo which allows the Ampico a "Brilliant"
setting, since the "Medium" setting cannot utilize crescendos or
intensities to raise pump pressure over its nominal (idle) reroll
pressure. There would be no other way to do it!
(As an aside, however, the B was "doped out" to compensate the effect
the model A has when intensities are left on, which does raise the
pump. No question about it. That, however, has already been taken
care of in its design. Dr. Hickman alludes to this too, so it was
I have tried to show why the standard roll coding format uses the
crescendos and the intensities in the way it does, and why it didn't
use intensities alone, for example, to raise pump pressure -- since
you would have none left to accent with! We have also seen why two
crescendos are totally unnecessary, and the separate crescendos noticed
on rolls could be removed by re-editing and the most sensitive ear
could not hear the difference.
Remember, you would have to have a difference in length of 1.6" between
bass and treble slow crescendos at a tempo 80 to raise pump pressure
1." At a normal mf playing intensity, 1" is about a 6% difference.
And the fact is, even that tiny difference between bass and treble
occurs seldom on the oldest rolls, at that. That's why the model B
should have no trouble following them, either.
3. The third seeming inconsistency is the timing of the crescendos in
the model B Ampico. How can the piano possibly be compatible with an
old roll intended to crescendo at half the speed of the new instrument?
Answer: When the change it makes in its excursion, percentage-wise, is
half as much! The only thing "sacrificed" here is the length of time it
can react to super-long crescendos.
A slow crescendo hole lasting for 4 seconds (extremely rare) at a tempo
90 would be over 7 inches long! The model B would still react normally
to this, even though it lies far outside Ampico's standard roll coding
In summary, no one was interested in making the model B compatible
with the model A because it wasn't necessary. I have never even read
someone who claimed compatibility between these instruments, either.
To argue the point is to ignore the real reason for everything. It
isn't meaningful to depict either model of Ampico's expression
characteristics with a set of curves taken when the pumps are fully
amplified, since this is only a special case, occurring momentarily,
and certainly not representative of the Ampico family of expression
One other question was raised, as well. At the 6th intensity in the
model A, (that's when intensity bellows #6 and #4 are relaxed, but #2
is still under vacuum) there seems to be a percentage error measured
from what one would expect to have. The reason isn't caused by
"anomalies" created by the Bernoulli pressure effects on the regulator
valve. After all, at the 6th intensity, that valve is too far away
from the seat to be as strongly affected, anyway. If anything, one
would expect Bernoulli to have his irreverent say at the first
intensity, rather than the 6th. Sorry -- that just isn't the reason.
Nor would it more likely happen at the first intensity, since the first
intensity is your set point, and it just sits there, correcting for
every little change in stack pressure before the stack even recognizes
there was change made. Bernoulli is canceled out!
The real reason has to do with the fact that at the leverage position
of the #2 intensity bellows, it takes only about a plus or minus .18"
difference in bellows centering between the top and bottom leaves to
make this difference at the valve. So a plus or minus .09" error in
placement of both leaves could create this very easily, not to mention
the friction and angular pull of a bit of misalignment!
The approximately 8% difference discovered by one researcher therefore
is normal working tolerances when you are affecting a physical lever.
The other two intensity bellows aren't nearly as critical, because of
the lower ratio of leverage their tolerances would be multiplied by.