Keep in mind before reading further that the only reason to change out
Duo-Art cross valves anyway is because over the years they become very
temperamental and leaky, changing sometimes from day to day, and playing
your favorites differently each time. Then if that piano is moved, the
chances are very good that it will not play at all, due to massively
combined leakage of all the valves, together.
In order to make this data meaningful to someone who may otherwise back
off with his index fingers making the "sign of the cross" like he has
just encountered the bogie man, I need to preface these figures and
facts with a summary of what I have already read in regard to some
pretty scary claims for round valves and let the reader decide
objectively how much substance they contain once all the facts are in.
That's elementary and fair. Otherwise, our minds sometimes tend to get
stuck in a previous misstatement or error, and despite all the facts
available to correct it, if we can't recall exactly why we were
reluctant to make a change, then despite the new facts, we still won't
change. So here are the main points mentioned so far against cross
1. As the area ratio between the lifter disk and the "working area" of
the poppet (defined by Bob Taylor as merely the hole in the valve
plate) approaches 1:1, the valve will cease to function (Bob Taylor).
2. Weight of the older "cross" valve poppet is too great (4.3 g) which
creates too much initial momentum, anyway. It should be changed to a
smaller valve, i.e. 2.3 g. as was done in the concert grand Steinway
collaboration with Mel Septon when these valves were changed to round
valves (Bob Taylor).
3. Cross valves are historic as well as superior in all respects and
should never be changed (Dave Saul).
4. Cross valves were still retained in only the top of the line
instruments for some reason (Dave Saul).
These points will now be answered.
1. An early Duo-Art pouch (1" dia. well) will lift 145 g without a
lifter present, at 25" of vacuum. So at a low vacuum, 5" of vacuum
with the standard lifter disk of 5/8" it exerts 50 g. of force. The
suction of the poppet against the plate at 5" vacuum pressure is about
5 g. (with one of the leathers tested) and the weight of the valve is
another 4.3 g (Taylor's figure). So with over 40 g. available to push
that valve, I don't think anyone is going to have a problem.
That's a force ratio of 5 to one, at least. At low vacuum, the suede
on the leather acts like lots of little springs (as will be proved),
partially supporting the valve, and preventing the valve from totally
sealing off, anyway. Therefore, I've discovered with the valve tester
that certain leathers actually give you a better pouch to valve force
ratio at low vacuum than at high vacuum.
One can then definitely say that even with no lifter disk at all, the
Duo-Art pouch has vastly more power to lift a poppet at any pressure
than that poppet can resist on either valve plate. Bob Taylor's
calculations, therefore, which presume that the moving of the poppet is
wholly dependent on the ratio of lifter area to valve hole area
difference is disproved. That is definitely incorrect as shown by
2. The mass of the poppet valve in the older instruments using cross
valves is too great for fast repetition? I would question that on two
grounds. First, both Bob Taylor and Dave Saul assured us that the
original valves, as designed, beat the socks off round valves and later
designs. How could they do this if they already had one strike against
them? That doesn't follow, logically.
Second, by decreasing the weight to 2.3 g, as was done (above), the
rebuilder now has a 2 g. advantage in the weight to pouch force ratio
which was already 5 to one. So I will leave it up to the reader as to
what they believe was actually accomplished by all that extra work,
ultimately. If you wrote an equation representing all the forces at
work on the poppet, and you have forces like 90 g. and 220 g. and
then an overall poppet mass of 2.3 g. instead of the original 4.3 g.
would you decide that you had found the deciding factor in optimizing
the valve design?
3 and 4. Cross valves are historic, and that's a fact of history.
Superior they aren't. As far as I know, I have been the only rebuilder
to publish data on them after researching them (beginning in 1986). A
cross valve hole, as I have already shown) requires more seating area
or hole perimeter length for the area uncorked than does a round hole.
That cuts down its volumetric efficiency considerably.
The cross valve also indents leather badly-- especially garment
leather. Unless you tan your own hides, you will never again buy
industrially staked leather goods-- an exceptionally strong,
solid-bodied leather .060-.070 thick, finely napped and extra tough, as
was used with cross valves. And which, we can all acknowledge, STILL
indented so badly that the valves, once rotated, totally incapacitated
the player. But garment leather is ten times worse!
So let's talk about history for a moment: Aeolian had purchased by
contract 50,000 Steinways in advance, and had laid up thousands of
cross valve stacks for these instruments, according to John DuParis, as
told to myself and Bill Singleton at the combined convention in St.
Paul. John was thoroughly familiar with the Aeolian factory and is
acknowledged to be the best source on this subject in the country
because of his first-hand information, according to Bill, who chose him
to settle this question, once and for all. John did!
When Aeolian sales of their big pianos didn't pan out, they asked
Steinway to be let out of their contract, and Steinway graciously
consented for Aeolian's sake (Steinway could have owned Aeolian,
otherwise). Aeolian continued producing pianos throughout the twenties
after that, but only the smaller pianos got the new stacks. The
Steinways and larger scaled pianos already had a stack laid up to
install. That is why these "flagship" model pianos retained the cross
valve, while the newer pianos got the round valves, instead.
In addition, Aeolian, historically, always had problems with these
valves, even when new! How do we know that for sure? By the factory's
own attempts to correct their tendencies to rotate! Several rebuilders
today have found species of this cross valve with slotted fiber insert
stem guides and pinched valve stems designed to prevent their rotation.
However, it didn't work because the clearances required at the stem
still allowed the valve too much rotation, anyway. The valve is a bad
design, even according to factory records. Neither their repetition,
their sensitivity, or their positive seating was as good as any round
valve you care to name. That is why Aeolian first tried to "fix" them,
and finally did away with them completely. Their much later Steinway
M's for example no longer used cross valves at all, and these were the
top of the line pianos designed to compete with Ampico's latest model
The proof of the pudding is certainly in the eating, and the proof of
the piano is in the hearing. Had cross valves been superior to round,
they would have made the concept work.
Now that these objections are clarified, here are some test results
using different leathers and different pressures. You will also notice
that it is actually the leather used that seems to make the most
difference in the readings taken on these valves.
This fact leads one to the conclusion that probably what rebuilders
have experienced all along cannot be pinned down precisely because of
the different leathers they will all tend to use. Since the pouch well
of these pianos are admittedly a bit undersized then, unless the
rebuilder is determined to tear a stack back down as many times as
necessary in order to get it right, he will have no choice but to let
pianos go out the door with sluggish or problematic valves.
The following tests were made with three leathers, A, B, and C.
Leather A is .030 thick, hard-bodied calfskin with an ultra-fine nap (I
call it 'the pink stuff'). Leather B is a softer handed deerskin with
a medium (average) nap and is .060 thick. Leather C is also deerskin,
soft, medium nap, and .070 thick (from a different skin).
The weights recorded are on a Pelouse P-10 scale which is accurate to
5 g. So if a reading fell between, I would add two pennies. If it
went up to the next 5 gram increment, I assumed it was a bit over its
stated value, and if it did not move, then it was either a bit under,
or exactly on that value. However, 5 g. plus or minus is meaningless
in these measurements. The leather, as you will soon realize, is the
determining factor, altogether. The leather you pick means the valve
will either play well or not! There are also other sub-factors which we
will touch on at the conclusion of this article.
The Valve plates are designated (O) and (X) respectively, below. Three
vacuum pressures were recorded, 5, 10, and 15. Three leathers were
used, A, B, and C. Leather A is .030 extra fine nap calf, very solid.
Leather B is soft deerskin, .060, medium nap. Leather C is soft
deerskin, .070, fine nap. All leathers are very nice material, and
[ The first column, 'L', is Leather sample. ]
L | 5" vac. | 10" vac. | 15" vac.
| | |
A | (X) 15g (O) 15g | (X) 35-40g (O) 40-45 | (X) 45-50g (O) 65-70
B | (X) 15g (O) 15g | (X) 35g (O) 45-50 | (X) 50-60g (O) 60-65
C | (X) 15-20 (O) 20g | (X) 30g (O) 40-45 | (X) 155*g (O) 60-65
While these readings can vary notably even depending on the individual
pieces of leather used right out of the same skin, one strange reading
popped up consistently, after many multiple attempts to check it *, I
had to include it. It does show what I had suspected from thick
leather that is also softly tanned-- it wraps itself around the sharp
corners of the cross valve and sticks really tight! That is what
finally causes the cross shape to become imbedded in the leather.
Notice that there was no problem with the round valve and this same
These readings also show something else, too. Since pressure readings
in inches of water are linear with lbs/sq. ft. pressure, 5" should be
half the force of 10" and so forth. But notice that these clamping
pressures of the poppets do more than double as the pressure goes up.
The reason can only be that the leather nap surfaces act somehow like
little springs, tightening a little more with pressure.
What this also proves is that there has to be a pressure gradient
developed under the seating area of the valve, or footprint.
Therefore, you cannot ignore the sealed area. Makes sense, since air
molecules are many orders of magnitude smaller than the fibers we
intend to use to seal them with!
So while all these readings differ, and while the one "out of sight"
reading was also consistent, none of them answer the question as to why
some rebuilders have bad luck with cross valve replacements, since the
pouches still exceed, by many grams, the strongest forces applied to
the poppets. Yes, even pouches that don't have a lifter disk are still
able to operate the valve. It shows that all attempts to evaluate
these valve characteristics statically are really invalid.
I think, from my studies thus far, that the problem possibly arises
either from the leather chosen, the pouches used, the sealant used on
the pouches, or the bleed sizes. I know that by resizing the bleeds,
we are also able to simulate some of the same effects, and conversely,
eliminate some of those effects.
I have also discovered that many old, unrestored Duo-Arts have more
than simply one bleed hole size used in their bleed strips. So it
makes sense to check on that once the valve and pouch have been
replaced and regapped.
In summary then, the Duo-Art valve hole-to-lifter-disk-ratio is not a
deciding factor at all, statically. Neither is the pouch diameter to
valve hole, since it's easily adequate to operate either form of the
valve. Using a larger or smaller lifter will effect the static and
dynamic properties of a valve, of course, but mainly it's repetition.
(Since the return speed of a valve is many times slower than its
actuation speed, too large a lifter slows the returning valve too
One roll I have always used in my stack of rolls to initially test the
Duo-Art is Hofmann's rendition of Caprice Espagnol. But much harder to
play rolls than that one exist. For example, Otto Cesana's "Negro
Heaven." It uses entire handfulls of huge chords played staccato at
the lowest levels constantly. Some rebuilders think that once they
have their piano set to play Home Sweet Home or Afternoon of a Faun
softly enough, that the same piano should then not be able to play
Cesana. That isn't true, but Negro Heaven is a scary, tough test to
make. Try it on your own Duo-Art and see how it fares.
Like Dave Saul said, "nothing can take the place of listening to a
large number of rolls containing many different kinds of musical
passages." And I would add to that, regulate each valve to its
environment first. Test them all before you reinstall the stack.
And then be prepared to remove the stack and re-regulate the valves
if they aren't up to standards, if you want a first-class job.
[ Editor's note:
[ Along with the test data Craig has presented his conclusions.
[ I hope that others will build a similar test setup and try to
[ duplicate Craig's results, and perhaps propose further tests.
[ We can discuss "theory and practice" after the experimental data
[ is sufficient. What additional parameters might be measured?
[ -- Robbie