[ Editor's note:
[ The Digest, being a daily forum, enables give-and-take discussion
[ which is almost impossible in a monthly or quarterly journal. I am
[ monitoring the technical thread about the Duo-Art valve with
[ interest, because (1) it's something for every piano tech to think
[ about, and (2) I hope that new experiments will help us to
[ understand how the valves really work. If the right data can
[ be gathered from the experiments, we can re-discover the science
[ underlying valve design.
[ For the moment I want to minimize the discussion about theory
[ and concentrate on gathering new data from bench experiments, and
[ from rebuilders experiences. Craig Brougher describes a test bench
[ setup in his article, and we hope others will try to duplicate his
[ results. Later, with solid data to analyze, we can postulate how
[ and why everything works like it does.
[ -- Robbie
I've made some interesting measurements regarding Duo-Art Cross valves
versus round valves. First, let me describe my test setup.
On a board, I mounted three valve plates -- two were round (one from
PPC and the other from Bob Streicher), and one was the Duo-Art original
cross valve plate. These three valves in turn were supplied by a 5/8"
hose from a regulated supply set to exactly 25" vacuum pressure as
monitored by a (0-40" vac) Haenni precision Swiss gauge.
The board was then clamped upside down and supplied with vacuum. The
valves used were turned off a wooden dowel on the lathe and supplied
with tiny eyelets, from which were hung a 6" fine-wire spring. At the
end of the spring was a Dixie-cup with a wire bail, into which I would
pour small lead shot. As soon as I started to hear the seal
"sputtering"-- indicating that it was about to break, I weighed the
shot and spring. I did this for several different kinds of leather and
different diameters of leather, and came up with some interesting
conclusions that I was not aware of, frankly.
The cross valve's clamp on a very solid, thin, fine napped leather
(.030) was low-- about 55-60 grams at 25" of water vacuum. That
follows right along with what we've been hearing-- that these valve
plates actually release sooner than their round counterparts. But
I then tried a fine white organ leather of about .060 thick, and the
clamping force was 80 grams. Now, that was interesting! About 50%
greater than the thinner leather (that Duo-Art never used, anyway.)
So I next tried that same white leather on the round valve plates from
Bob Streicher, and discovered an interesting fact. It's release point
was consistently 70 grams. Ten grams less than the cross valve. Hmm.
That doesn't necessarily prove that I've been right, however. Because
the valve leather I've been using on the Duo-Arts is the thinner
leather and stuck tightly to my round valve plate until I reached 90
grams. Now what can I say? Did I have sluggish valves after all?
By sticking all kinds of leathers to those valves and trying each one,
I began to realize that there is more variables in the leather than
there ever was in the geometry! That still doesn't answer the question,
"Why did my valves work so well, if they were 80% harder to actuate
than the slightly rougher and thicker suede?" Obviously, the problem is
not as everyone thought-- "geometry." There is another factor present.
So I went to the pouch and bleed. But before I tell you what I
discovered there, let me describe the characteristic of the two valve
plates to release.
The cross valve plate is easier to disconnect from because it is not
uniform in shape. Therefore, a little bump from the pouch will break
the seal on one side, thus releasing the valve earlier. If the valve
were highly precision, then we wouldn't have that effect. It would
release all together, but require more initial force to do it that way.
Instead, the cross valve releases "begrudgingly," but at a lower force.
The round valve releases all at once, but at a higher force. What you
may gain in initial release time by the cross valve plate, you lose,
because the extra force required to release the round valve also shoots
the valve upward (potential to kinetic energy release). The slower but
easier release may give the valve an inevitable tip to the side, which
must then be corrected at the top seat before the valve can seal and
close the pneumatic. So- what it seemed we had gained, we actually
The Duo-Art pouches have 1" pouch wells. The test setup I designed is
a 1" dia. pouch well with a 5/32" nipple going directly to the supply.
Through the bottom of the well is a 1" long x 5/32" brass nipple acting
as a bushing, through which a 1.5" long brass welding rod 1/8" dia.
passes. The rod has a 5/8" dia lifter attached to its end under the
pouch entrapping it, but it does not glue to the pouch, and is operated
by the pouch in reverse. The pouch is dipped about 1/8", or factory
dip. 5/8" is the diameter of the factory lifter disks of the
particular early Duo-Art that I have measured.
The advantage of doing it this way is the precision with which the
pouch must exert its force on the stem, and the way that I was able to
also operate the same pouch by vacuuming a chamber above it instead.
This way, we are not questioning whether there was a difference between
pouches. The same pouch works for both, without having to remove
anything or modify something.
Above this pouch then is an evacuated chamber that I can alternately
vacuum down instead. It also has a rod and bushing stem system, but
has a 1/2" lifter instead of a 5/8" lifter. By setting a cup of shot
on the end of this stem (or the other stem underneath the pouch by
flipping it over), I was able to measure the power of a 1" sealed pouch
to lift weight at 25" of vacuum.
With the 1/2" lifter disk, the pouch exerted 190 grams. With the 5/8"
disk, it exerted 220 grams. Both, you can see, will easily lift the
valve whose adhesion to the valve plate ranged from 55 to 95 grams,
depending on the leather used. With margins like that, (far in excess
of Bob Taylor's ratio between lifter disk and valve diameter) I really
foresee no lifting problems by such a pouch, so I have to believe that
the problems encountered using round valve plates are probably caused
by something other than the ratio of lifter disk to valve poppet, or
the insufficient diameter of the pouch.
The scale I used is a Pelouse P-10 electronic postal scale.
If anyone has some ideas of other tests they would like to perform
using this already built setup, please describe in detail what you
would like to try. And, if you have any ideas concerning the anomalies
encountered with these valves, please speak up. I'm learning about