I've looked at the archive discussion on this topic, and the recent
answers.  All good stuff, and useful advances to the understanding of
one of the most puzzling aspects of the hobby: Duo-Art regulation!
I'd like to describe a technique I've found useful and have not seen
discussed before.

As we know, Duo-Art suction levels drop as the amount of air passing
through the system increases (because the knife valve has to open a
little wider to let the extra air through, and so stretches the
regulator spring less).

If different notes admit different amounts of air when they play, it
will result in varying suction levels for the same coding, which can
cause notes to drop out at low levels.  I'd like to describe a quick
way to test for this.

To set up the test in an intact piano, uncover the tracker bar and
disconnect a note (I use the lowest bass and highest treble, doing each
side of the stack separately).  Connect the chosen tube to a suction
gauge - which will need to be sensitive to a fraction of an inch.  This
will give a reading of the stack suction.  It will take a little time
to stabilise, because you are connected via the bleed, but will give a
correct reading as long as the gauge does not leak.

Now put on a test roll which plays each note at zero level.  First,
read the suction for zero level when no note is playing.  Now play each
note in turn.  The suction level in the stack will drop a little as
each note plays, due to air admitted via the tracker bar and any
leakage in the valve.  Measure the change in suction level for each
note.  Ideally the change should be near-identical for each one.  To
compensate for the effect of the wind motor, either play the roll at
constant speed or stop before each measurement.  The absolute values
will differ either way, but the differentials will show up just the
same, which is all we want.

 [ Disengage the chain and run the motor at Tempo 80.  -- Robbie

If any note shows a significantly greater suction drop when it plays,
then either the valve or pneumatic is leaking or its bleed is too big.
These notes will either drop out themselves, or (worse still) will
cause chords containing them to drop out.  Fix them so all the notes
admit the same amount of air! If you don't, the Duo-Art will *never*
work properly, no matter how well everything else works.

If you have a fast-acting gauge, and play the roll right through, you
can look for a greater 'blip' when a note plays, which may indicate
transient leakage across the valve due to sluggishness or poor seating.
It's not definitive, but can give a quick feel for the problem.

The advantage of this test is that it can be performed on intact
instruments quickly before committing yourself to taking the stack
apart.  It's also useful to confirm that everything's been put together
right after rebuilding!

This tests only note-playing air flow.  As mentioned by others, the
Duo-Art also requires a very small 'static' air flow to allow it to
control suction when no notes are playing.  The static air flow is
compensated for during regulation so won't cause drop-out, but it will
affect the overall response curve.  One of the obvious points is that
it must be the same for both halves of the stack.

A controlled bleed as described by Phil Dayson is a good way of
balancing the two sides of the stack.  It also seems a perfectly
reasonable thing to use to manipulate the response curve, if that's
what you want to do (it doesn't damage the piano, so what's the harm in
trying?).  However, I can't help thinking that Duo-Arts must be capable
of working as intended without such fixes.  I've been trying some
regulation ideas which I'll report later if they work!

Julian Dyer