Regarding Bernt Damm's question: A few thoughts come to mind, but
this is a complex topic and there is more to say!

The setting of the Duo-Art accordion travel is determined by the
geometry of the knife valve assembly.  The intention is to move the
knife valve over its full travel, not too much or too little.  Because
the rest of the geometry doesn't change, there is no call to change the
setting of the accordion travel either.  However, it is important to
get the setting of the various connecting rods right as these can
affect the knife valve movement.

The suction level obtained from a particular accordion setting is
determined by the tension in the regulator spring after it has been
stretched as the regulator closes.  If a note or chord does not play at
a particular setting, then the spring tension is too low.  (Assuming
the stack is airtight, etc.) If the 'won't play' test chords all play,
then the spring tension is too high.

Springs have two characteristics which determine their tension at a
particular extension: the tension when not extended, and the spring
rate.  If a spring is not the same in both regards as the original Duo-
Art spring, then the effect of stretching it will not be the same and
the test roll will not come out quite right.  To quiet a Duo-Art
overall, a weaker spring should be fitted, rather than mucking up the
expression geometry.

An observation I have heard made about new springs is that their
unextended tension is too low, so they need to be stretched further
to achieve the zero level (which reduces the working length of the
spring so further increasing tension as it is extended and making the
instrument too loud).  Does anybody have any measurements of these
parameters of old and new springs? New made-to-measure Duo-Art springs
are not readily available, as far as I know.

I have noted that some UK-made Duo-Art boxes have suspect geometry --
the knife valve 'tops out' before the accordions have fully closed.
This flattens off the top of the response curve.  This could be solved
by reducing accordion travel, but is probably better handled by making
slightly longer connecting arms to reduce the effect of the accordion
movement, thus maintaining the timing of the expression level changes.

Julian Dyer