Pete Knobloch wrote regarding the chart for his loud duo-art. I
don't think that an accompaniment level of 20 wci is very loud for
level 15 but the lower levels might be what cause it to seem loud. It
is the low end (4 to 10 wci) that gives sensitivity to a reproducing
piano. You first mention a 0 level of 4.5 wci but later change it to 5
wci.  This is actually a big difference for the zero level.  Will the
piano play without dropping notes at 4 or 4.5 wci? The levels you list
in this range are high and could be even higher if the gauge is
inaccurate. If it is a cheap (less than $300 new) modern gauge it could
be way off. The only way to get meaningful vacuum figures is with a
water column or with a gauge callibrated with a water column. I don't
like to hook up a water gauge (even with a trap) to a piano since I can
be somewhat klutzy now and then. I have a few very good gauges which I
check against a water column once in a while and they are always very
close. For the low end I use a 0 to 10 wci gauge.      There might be
some way that your dynamic (while playing a roll) levels are higher
than the static (tested ones), but I can't think of anything that would
cause this. Craig? Dynamic levels on a duo-art are lower due the lack
of a regulator.  If you need to change the accordian spacing as much as
you did to get 20 wci at 15 there is definitely something wrong with
the expression box like leaking knife valves (which are probably
sticking too.) The need to make any radical changes from factory
settings on a duo-art indicates that something is wrong.  Duo-arts are
hard to patch together since everything is so interdependant.  You
probably need to rebuild the stack and expression box if they haven't
been done properly in a long time.  They way the duo-art attains the
required playing levels is very dependent on the stack valves leaking
the right amount at the right time. The coding is heavily weighted for
chord size and natural playing will never happen if this isn't right.
Unfortunately the cross valves together with modern leather are a bad
combination.  One other thing to be careful about.  Make sure that you
retain the binary relationship between the accordian spacings or the
dynamic (loudness) curve can get strange. I didn't plot your curve but
it does look a little steep in the beginning and flat on top.  I don't
see the exponential increase you previously mentioned, I guess that was
before the changes.