conversations with Bob Harris, early '00
*) a short stack is optimal; after that all you can do is reduce losses

*) acoustic tuning only works when there is enough back pressure to keep 
   pressure in the manifold after the slug goes by
*) all of Smith's generalizations are on high back pressure passenger car
   systems
*) acoustic tuning is most effective during overlap
*) most acoustic effects are probably due to the carb or MAF being affected
   than by any scavenging
*) the usual "acoustic pulse kicks charge back into the cylinder" doesn't 
   work; the air flow is always outward, no matter which way the pressure wave
   is going

*) flow across the exhaust valve is sonic when the valve opens; the crack is
   the same thing that you get when you fire a gun
*) exhaust flow is nonlinear; indicator diagrams show most of blowdown 
   happening before the piston moves far from BDC.  The continuous 100PSI or
   so shown after that is probably from residual heat expansion and the
   compression of the piston
*) the exhaust gas is traveling at critical pressure; it behaves more like a
   liquid than a gas.  It is still a fluid.  The slug will pass an open branch
   without deviation; it will leap gaps without noticeable expansion; it stays
   in a coherent slug until some distance (many feet) from the exhaust valve

*) inertial tuning via the main exhaust slug is the major factor
*) as long as the primary pipe is long enough for the slug to still be in the
   pipe when the exhaust valve closes, there are no acoustical effects
*) as long as there is not enough back pressure to keep the branches of the
   header filled with gas, the header will run at lower-than-ambient pressure

*) some intake charge can be pulled into the exhaust by inertial vacuum.  
   During overlap, the pressure in the intake and exhaust are about the same;
   the cylinder is effectively drawing from two intake ports until the exhaust
   valve closes.
*) Ford uses drilled passages through the heads in their V8 and 3.8 V6 
   engines.  These air injection passages are always very clean, even on high
   mileage, heavily carboned engines.  This indicates the primary type of flow
   is fresh air, even though air injection is off most of the time.  At
   any given time three pipes are under vacuum, the fourth is only under 
   pressure part of the time.  Fresh charge is pulled into the drilling during
   the overlap cycle and distributed among the other ports.

*) a large exhaust box with a flat plate front or stingers will be more
   effective than a streamlined collector
*) collector size is very important.  Smith says 15x cylinder size.  Too small
   a collector will cause back pressure and hurt low and mid range torque.
*) most factory collectors are far too small
*) it'd be hard to make a collector too big
*) the Dr. Gas X crossovers are 3" pipe; they're basically collector 
   extensions.  Same for the Borla header mufflers
*) the ring mark inside the collector is where the slug has finally expanded
   enough to hit the ID of the collector.  Past that point the collector isn't
   doing much.  As long as the mark isn't more than six inches or so in the
   collector, cutting it off won't do anything; once you get close to the
   mark, it will recede ahead of the cut-off end.  3"?

*) mufflers should go at the extreme front of the exhaust system, where the
   noise amplitude is highest
*) free-flow catalytic convertors can be had with 3" inlets, are straight
   through flow, low restriction, and are substantially cheaper than 3"
   stainless mufflers
*) a catalytic convertor could work as a "flow diode" as it adds heat and
   velocity to the slug

*) primary pipe diameter is not of great importance as long as it's relatively
   close to the exhaust port size.  Too big is probably better than too small.
*) with a well-designed (low back pressure) exhaust system, each primary 
   functions completely independently of the others.  The vacuum in the 
   primaries prevents the pressure waves from going anywhere. 

*) almost any header design will work; the difference between a good design
   and a bad design is probably only a few hp.
*) the odd firing order per bank on V8s means one cylinder on each bank feeds
   an already-pressurized exhaust system as it follows the one 90 degrees 
   ahead of it.
*) with a high back pressure system, the usual four into one V8 header 
   actually works more like a three-into-one with an idler tube, acoustically. 
   In practice there doesn't seem to be any difference
*) a tri-Y design is actually a 270 degree even fire header
*) the angles and lengths on a tri-Y don't really matter as long as the pipes
   are big and long enough
*) the larger pipe used on the stem of most tri-Ys is effectively only a 
   simple stepped primary tube, though the step is larger than most 
*) all V8s with 90 degree cranks have one cylinder pair on each side firing
   only 90 degrees apart.  A small idler tube between the two trailing primary
   pipes, just ahead of the collectors, might help the situation.  Maybe 1"
   diameter, between #3 and #5 on a Ford V8.
*) you could probably do something useful by extending #3 and #5 primary pipes
   well into the collectors with stingers to help separate them from their
   leading cylinders

*) most of the stuff is in Smith; he's just opaque until you already 
   understand it

*) a two stroke expansion chamber works as an air spring; the volume of the
   chamber stores compressed air.  When it expands, it's easier to flow back
   into the cylinder than to exit via the stinger
*) two strokes are so loud because of the sharp pulse as the piston uncovers
   the exhaust port.  Same thing for Wankels.  The sharp pulse helps "square 
   up" the leading edge of the gas slug.

07/26/96
- looks like the more overlap a cam has, the more exhaust back pressure
  or scavenging becomes important