Paddy doesn't have to worry about "starting a war," because I don't
fight them. I understand the power of opinions. They are usually
resolute and irrevocable, but ultimately the facts prevail over time,
so we will let this all sift down, and I am prepared to accede to
British superiority every time I find it, as in this reply.
Some years back when I was single, I traveled around the country and
"deeply" repaired (even restored) many Duo-Arts right in the customer's
own home (about 27 in all, I think). I lived there until I was
finished, sometime for up to 5 months at a time and many pianos, and I
encouraged anyone who wished to watch me ( All of these instruments had
been "fully restored" just a few years earlier and delivered as "new"
instruments). I got a lot of interesting and knowledgeable questions
from the owners, and they learned things because I often could
demonstrate what I was claiming with gauges and an actual demo, and I
didn't back off and demure from any loaded question. I liked them. I
But quite often my answers and hands-on demonstrations didn't jibe with
what they believed. A few weeks later, I would get the same question
as an objection. So I'm used to this. It's normal. If I got mad
about it, then I would have the problem, because there is no disputing
the facts of the matter. Only the acceptance of the basic principles
which determine the facts of the matter.
Let's look at this from a new perspective, which John Tuttle mentioned.
His dad used to be a plant engineer. Suppose he ordered in several
hundred pressure regulators which were supposed to follow a square law
exponential curve (or some vague approximation), but they were designed
around a large, linear spring.
However, he was told, "Not to worry, Mr. Tuttle. We know you need a
deeply concave curve and this box is basically linear, but you can get
it by changing the levers and things in each individual unit. For
instance, changing the accordion lever angles, away from the box.
Pulling out the felt and remodeling it or using nameboard felt,
readjust the accordion travel, tighten up the accordion lever rods,
change the place the spill knife covers it's hole, and there's spring
tensions to change, shorten up the regulator springs and increase the
spill spring tension. By using different combinations of these
adjustments, any box can be made to curve its response.
"So even though the box is designed linear, and even though we both
agree that whatever the natural rate of the spring (linear) is also the
natural rate of the box (linear) and that's a basic principle, and even
though the accordions move the knife valve in linear, equal-sized steps
and gradually close the spill valve the same percentage, we feel that
the ideal curve you need can be regulated into each box differently and
separately, with enough time and effort."
Who thinks that Mr. Tuttle would have put in his order for these
regulators? He would think they were nuts! And no factory is going to
build tens of thousands of square-law output regulators using, as their
basis, flat, linear springs and telling the customer that by getting
them all out of regulation they will be able to approximate the curve
they want to get. Yet we are asked to believe it. For instance, how
long would it take the factory to regulate each piano off the line if
they really wanted it to be exact, and each instrument achieve that
subtle, desirable curve? (In this case, not a square-law curve but a
linear curve with a bend in it?)
Remember the principle: The rate of the box is primarily determined by
the rate of the spring, and the accordions merely tap off a portion of
the pressure available at a specific step in that curve. They do _not_
change the curve or vary anything else except the position of the knife
valve. (Yes, you can bend it by changing a half-dozen different
Now because the accordions cascade in binary fashion (each doubling the
effect of the preceding one for a total of 15/16th of an inch), the
travel they give the knife valve is such that the overall pressure of
the box is very deliberately linear. That _is_ a fact and not
contestable. It is a physically provable response and any small
curvature noticed in one box is often negated by the opposite tendency
in the next one, meaning that these are normal variances, not to be
confused with design criteria.
If Duo-Art wanted all their boxes uniform with a basically curved ideal
response, that would not be hard to achieve. One of many ways: You
just make the regulators smaller with the same or stronger spring
tension so that as the spring is stretched the regulator cannot close
commensurately as quickly. Or you change the spill valve speed at the
lowest steps. By rescaling, there would be a basic upward curve to
every box output that could not be re-regulated to linear, and Paddy's
ideal curve would then be in stone, inherent in the box's design. You
see, the concave curve is easier to achieve and cheaper to build.
linear is more difficult. The reason this box is linear is because the
spring is linear-- and-- the regulators large enough to take advantage
of it's full range without losing power. They are sized to the spring
strength itself. Paddy said:
> The gaps between successive levels increase throughout, markedly so
> above power 10 providing that the pump has sufficient tension and
> flow to minimize the inherent 'droop'. Power 10 is approximately mf,
> so there are more gradations below mf than above.
Again this is basic arithmetic. A binary cascade doesn't mean that
levels increase markedly above power 10 (20" of water). If they do,
something is wrong -- the knife valve is leaking and no longer
controlling the pressure in the stack. The very _definition_ of a
linear cascading effect is to begin with 1/16" equal-sized incremental
steps. In no chart will one even see the point at which the spill
completely closes (the 10th step) because it does so gradually at a
rate determined on a sliding scale. Nor does the vacuum then "increase
markedly throughout," in the American Duo-Art. I don't know about the
20" is nominal pressure in all reproducers and refers to the threshold
of forte. Actually, 15" is in most players, mezzo-forte. We can say
this because most pneumatics are roughly the same size and most valves
play them all the same way. So 20" plays them all about the same
loudness. All common upright player pianos built after 1920 will play
at a normal mf volume on about 15". But it depends on how many notes
are playing at a time, and their duration.
For example, If you were to watch some of the huge chords playing on
Ravel's "Bolero," you would be surprised -- as your piano shakes the
walls of your room, that at best, with everything on including the
crash valve, that your Duo-Art may be barely getting above 20-25" of
vacuum at that particular point (certainly not 40"). Ideally you could
say it should be at least 40 inches, but no. Why does it drop?
Because the pump gasses out. The pump is sized to be an integral part
of the expression capacity of a baby grand and designed to limit the
capability. Even its belt was designed to slip above a certain
pressure. Paddy hasn't mentioned this about British Aeolian. I again
defer to his knowledge of that system.
> On British Duo-Arts the crash valve was not done away with: it was
> never envisaged, but its absence in no way means that ffff passages
> are not reproduced. British boxes in practice give negligible
> constriction and the selected regulator springs match the full
> expression curve/regulator span closely to the natural dynamic range
> of the instrument, so a sudden step-change crash is superfluous and
> musically an affront.
Regarding the superfluous crash valve, many people don't like it.
Granted, it's touchy. But when it's set right, over the course of
two dozen rolls or so, it is definitely needed (that is, in the
questionable American models), especially in a large room playing heavy
classics realistically. On the other hand, it is easily disabled, and
in British fashion, the Weber and Steinway uprights didn't use them at
all. The Theme regulator snaps completely closed during a crash, so
to say that British boxes do not need crash valves because their
regulators don't limit crash potential and maximum power can be
delivered serially through the regulator, is to say they are built
considerably different than the American design. Again, what can we
say, but that we possibly have a very inferior model, with the
exceptions of our uprights.
No American made box can provide a true crash, Paddy, so I defer to
your different equipment and to British superiority. No American
Duo-Art can provide a thunderous crash to the full capability of the
baby or parlor grand by accordion movement alone because the regulators
would throttle and block off the stack. When the crash valve operates,
the Theme regulator goes completely flat. That would choke the piano.
This is why the crash valve operates a third dump valve within the box,
bypassing the regulators and ducted by the respective Theme valve
directly into the stack. To say the valve is unnecessary in our style
of box may casts aspersions on the wisdom of American Aeolian but you
might be right. I feel it is better said that "American pianos had
them and British pianos didn't, because in a superior design they are
not needed." To give Aeolian their due however, it was, after all, the
American Steinway Duo-Art which debuted at Carnegie Hall and received
the music critics' highest acclaim.
> I may add that I have yet to hear a Duo-Art in the US that plays as
> loud, even using a crash valve, as a number of similar instruments
> here in the UK while still being able to match their ravishing
To this, all I can say is, maybe 40-50" of water vacuum in the United
States just doesn't work as well as 40-50" across the Atlantic? And
the same goes in reverse for ravishing pianissimos? Maybe American
Duo-Arts aren't as good as British Duo-Arts? Or perhaps the British
Duo-Arts just ravish longer-- say, up to the tenth step or so? We
apparently are not getting our money's worth from all the inches of
vacuum ravishing we buy :-).
Since none of this talk proves anything to those who don't believe it,
my last suggestion is to go to my web page and download this RealPlayer
zip file. I cannot promise ravishing, but rather some objectivity.
I can't recommend streaming files, as their musical data is greatly
compressed. Just download it to your desktop and then double-click
to open. What you will be listening to is a Duo-Art with a linear
expression response curve -- for what that's worth.