Since I've been raving about how those $%#&* hardware guys
keep screwing up the coin switches, I guess it's my turn to eat crow.
The reported problem with mech-multipliers on Blasteroids is in fact
my fault. Those poor coins went through two processors and at least
4 subroutines before I twisted them at the last possible moment.

	Anyway, all this is really pretty straightforward on the
standalone audio board. The rightmost switch gets hooked to the rightmost
bit (D0). The rest follow in order (up the byte right to left). These
single bits get translated into two-bit (what else, at least in the U.S.)
fields that get sent to the game board, but again, rightmost is rightmost.
Should we bring up 6502 based boards that connect their coins to the MSB's
MSBs of the inputs? In case you wondered, the rightmost switch connects
to D7 (actually the rule is that rightmost goes to the bit that falls out
first when you shift). Anyway, it's supposed to be the 6502 coin routine's
problem to get things straight before sending them on to the game, and
the S.A. board does it right.

	On a separate-mech game (e.g. Xybots), everything stays consistent
again, the rightmost mech is for the rightmost player. But players are
number left to right (0..3), whereas bits are numbered right to left
(fun, huh). This is still not too confusing, until I re-converted the
Gauntlet-style coin routine, as hacked for S.A. Audio, back to System I
style. 

	The software is clearly wrong, and can be fixed in one of two ways.
I can change which mech I consider "right", leaving right the more valuable
mech. Or I can change the wording of the coin options, acknowledging that
the Left mech gets the higher value. I vote for option one, but will listen
other arguments.

	Can get some mercy by pointing out that if I didn't have to
keep changing this stuff I wouldn't keep breaking it?

				Didn't think so!
				Mike