Sequel to Williams High Speed (1986, a different physical machine on an earlier, unrelated
electromechanical/System-6 platform, and out of scope for this record). The Getaway: High Speed II
runs on WPC-Fliptronic hardware (GEN_WPCFLIPTRON, pinned src/wpc/sims/wpc/full/gw.c) with a
128x32 dot-matrix display, one lower flipper pair, and a single upper-right flipper feeding a
motorized "Supercharger" accelerator loop.
A note on the pinned driver's own reliability
gw.c's header comment is explicit: its author, Steve Ellenoff, states he did not have access to the
physical machine and "guessed most of it from a Playfield picture and the rulesheet." This matters for
curation because it means the driver's #define comment labels (swOpto1, sRSling, and so on) are
not automatically trustworthy device-identity ground truth the way they are on most other curated WPC
titles, even though the driver's numeric public addresses (ordinary WPC column-times-ten-plus-row
matrix notation) are still real hardware. Two concrete corrections and one unresolved disagreement
came directly from this gap:
- Solenoids 5 and 6 are reversed in the driver's comments (
sRSling 5, sLSling 6) against two
independently-agreeing manual pages (Solenoid/Flasher Locations and the Solenoid Table wiring page),
both of which print item 5 as "Left Slingshot" and item 6 as "Right Slingshot." The manual wins; see
device.left-slingshot/device.right-slingshot in the definition.
- Switches 84 and 85 are a genuine, unresolved disagreement rather than a clean correction: the
manual's Switch Locations page labels 84 "Opto Made Loop" and 85 "Enter Left Ramp," but the driver's
guessed labels (and the retained known-working script's own runtime grouping/audio design) point the
opposite way. See
conflict.switch-84-85-manual-vs-script-semantics below.
The Supercharger accelerator loop
The machine's signature mechanism is a motorized ball-accelerator loop at the upper-left of the
playfield, built from the A-15297 Accelerator (Entrance) Ramp Assembly, the A-15300 Accelerator
Tray Assembly (the three-wheel drive stack itself), and the A-15293 Accelerator Return Assembly
feeding back down through a diverter to either the left ramp or another lap of the loop. Two dedicated
PC boards instrument and drive it:
- A-15189 Accelerator Board Assembly carries the row/column switch-matrix return wiring for the
three wheel-position optos (switches 81, 82, 83 -- "Opto 1/2/3") and the power/drive wiring for
the three wheel-drive Enable relays (solenoids 25, 26, 28) and the Enable feedback lines routed
back through the Power Driver Board. One board, three opto sensors, three motor-enable relays.
- A-13901-1 Opto Ramp Switch Board Assembly carries only switches 84 and 85 -- no solenoid wiring
at all -- one connector (J2) per address, each wired to its own opto LED/phototransistor pair.
The retained known-working table does not simulate real wheel-drive physics: its SW81_Hit,
SW82_Hit, and SW83_Hit handlers each simply add ball X-velocity in ascending amounts (+11, +15,
+19) as the ball trips each opto in turn, modeling acceleration through scripted velocity boosts
rather than a driven mechanism. The three Enable relay solenoids (25/26/28) have no SolCallback
entry in the script at all and are not visually implemented; this recreation's only visible feedback
for the loop is the ball's own accelerating motion and, separately, the "Supercharger Flasher"
(solenoid 18) and "Free Ride Flasher" (solenoid 20) lighting effects.
At the base of the Accelerator Return Assembly sits a diverter (retained table objects Wall29/
sc_div) controlled by solenoid 10 (Diverter Low, via the script's SuperchargerDiverter handler)
that either sends the ball around the loop again or lets it continue. A second solenoid, 1 (Diverter
High), is printed on the same physical assembly but has no script implementation at all in this
recreation -- the retained table simply does not animate it.
Switch 84/85 identity: unresolved
Both switches sit on the A-13901-1 board and both are printed as opto interrupters, so their
construction and polarity are settled (PinMAME's gwGameData inverted-switch mask covers exactly
81-85, a clean match against the manual's opto part sweep with zero disagreement). What is not
settled is which physical position each address senses. Three pieces of evidence pull in two
directions:
- The manual's own Switch Locations page: 84 = "Opto Made Loop," 85 = "Enter Left Ramp."
- The pinned driver's guessed
#defines: swLRampEnt = 84, swOptoLoopMade = 85 -- the opposite
pairing.
- The retained known-working script's own organization:
SW85_Hit sits in the script's 'supercharger
comment block next to the three wheel-opto handlers and plays a distinctive "sc_loop2" sound (a
sound cue no other switch in the file shares); SW84_Hit sits in a separate 'Ramp Triggers block
next to ordinary lane switches 65/67 and plays only the generic "rollover" sound everything else
uses. This also points toward the driver's guessed pairing, not the manual's.
Two sources agreeing against one would normally resolve this outright under this project's evidence
hierarchy, but the two agreeing sources here (the guessed driver comment, and the script author's own
organizational choices) are not demonstrably independent -- a VPX table author building a recreation
routinely consults the public PinMAME driver source for switch numbers, so the agreement may simply be
inherited rather than confirmed. Recorded as conflict.switch-84-85-manual-vs-script-semantics and
left open pending a LibPinMAME harness trace.
Ramp lift mechanism
The B-12576 Ramp Lifting Mechanism Assembly raises and lowers the right ramp surface between a
lowered position (routing the ball onto the Supercharger return path) and a raised position (routing
it along the ordinary right ramp instead), driven by solenoids 2 (Up Ramp) and 3 (Down Ramp). Its own
exploded-parts page lists a microswitch, part 5647-12001-00 -- the exact part the Switch Locations
page prints for switch 54 (Ramp Down), confirming switch 54 is this mechanism's own built-in position
sensor rather than a separate playfield object. The retained script's SolRampUp/SolRampDown
handlers set Controller.Switch(54) directly in the same code that toggles the moving ramp's
collidability and rotates a diverter flap, with no independent trigger object for the switch -- and
the pinned driver's own gw_handleMech function implements the identical causal pairing in C
(core_setSw(swRampDown, ...) keyed off core_getSol(sRampUp)/core_getSol(sRampDown)), the one
mechanism on this machine where the guessed driver and the manual agree completely.
Three-ball visible lock
Balls can be held on the wire right ramp at Top/Middle/Bottom Lock (switches 74/75/76, all opto
per the same A-13901-1-family construction pattern... no -- these three are not on either opto
board; they use the ordinary 5647-12693-21/A-15103 wire-ramp switch part, plain leaf construction).
Solenoid 4 (Locker / Disappearing Post, A-15127 assembly) raises a post that releases the locked balls
when fired. The retained script models this with the standard cvpmVLock helper class.
Kickback
The retained script's kickback handler (solenoid 8) fires a literal VPX Plunger-type table object
(Plunger1) positioned immediately next to switch 25 (Left Outlane) in the retained geometry -- not
the shooter-lane plunger, which is a separate cvpmImpulseP instance bound to solenoid 12 and switch
78. This confirms the B-11873 Kickback Assembly is a left-outlane mechanism, consistent with its
manual placement.
Gear shifter: not modeled in this recreation
Switches 33 (Gear Shifter Low) and 34 (Gear Shifter High) belong to the 20-9710 Up-Down Shifter, a
cabinet-mounted gear-shift lever with its own internal microswitch and bayonet-base bulb. Its own
manual page carries no item-number balloons or parts table at all -- only the illustration and a
"Cabinet shown for reference" caption -- and no object anywhere in the retained VPX extraction has
"gear" or "shift" in its name. The retained script drives these two switches purely from two unrelated
keyboard keys (LeftMagnaSave for 33, PlungerKey for 34); the physical lever mechanism has no table
geometry in this recreation, and neither switch carries a spatial placement here as a result.
Upper-left flipper: confirmed not fitted
Three independent, game-specific sources agree the machine has no upper-left flipper (only lower-left,
lower-right, and upper-right):
- The Switch Locations parts list lists only two flipper cabinet-button items -- a double switch
(right side, serving both the lower-right and upper-right leaf switches from one paddle) and a
single switch (left side, lower-left only). No
A-15205-L (upper-left leaf switch) part appears
anywhere on the page.
- The Left Flipper Circuit diagram (printed page 3-10) labels its second coil block "UPPER LEFT
(NOT USED)" in so many words, while the companion Right Flipper Circuit diagram labels both of
its coil blocks with no such annotation.
- The Flipper Circuits wire table (printed page 3-21) prints the Upper Left Flipper Button Switch
wire as "Black/Blue(NU)" -- the manual's own explicit Not Used suffix.
gwGameData's FLIP_SW(FLIP_L | FLIP_UR) | FLIP_SOL(FLIP_L | FLIP_UR) independently agrees: no
FLIP_UL bit is set. A later, generic duplicate of the Switch Matrix page near the end of the manual
lists Fliptronic positions F7/F8 without any Not Used annotation of its own, but this is the same
reused Fliptronic-II-controller silkscreen template every WPC-Fliptronic game's manual carries
regardless of that specific game's actual flipper count -- the three game-specific sources above
settle fitment decisively, not this generic block.
Solenoid 31 and the undeclared fast-flip address
The retained known-working script binds SolCallback(31) = "FastFlips.TiltSol", the standard nFozzy
cFastFlips convention for reading PinMAME's synthetic fast-flip flipper-enable signal. But pinned
gw.c never calls wpc_set_fastflip_addr anywhere (confirmed by an exhaustive source grep), and per
src/wpc/wpc.c's core_gameon function, when no fast-flip address is declared PinMAME instead
publishes public solenoids 29-31 as a mirror of bits 5-7 of the WPC_GILAMPS register -- a
general-illumination lamp-state register with no inherent connection to flipper enable. This is a
direct, source-verified contradiction between what the pinned driver says solenoid 31 carries and what
the retained table's script assumes it carries, recorded as
conflict.solenoid-31-fastflip-address-not-declared. Whether this represents a genuine defect in the
recreation's flipper timing, a coincidental correlation in this specific ROM's own logic, or a gap in
this analysis is not resolved here.
General illumination
Five printed illumination strings: strings 1-2 (GI addresses 0-1) are wired exclusively to the
playfield; strings 3-5 (GI addresses 2-4) are wired to the backbox insert panel, with string 5
additionally reaching a cabinet connector. The retained script's UpdateGI routine ignores its own GI
address parameter entirely and drives one shared 25-member playfield light collection for any GI
address activation, so this recreation cannot distinguish which specific playfield bulbs belong to
string 1 versus string 2 -- both addresses are left without an individually validated placement as a
result, an honest gap rather than an invented split.
Physical family
The five-driver clone tree rooted at gw_l5 covers fourteen total drivers in the pinned catalog: the
L-5 production ROM (this definition's primary binding, matching the retained known-working table's own
cGameName), the D-1/D-2/D-3/D-5 "LED Ghost Fix" display-timing revisions, the L-1/L-2/L-3 earlier
production firmware, five prototype ROMs (P-B, P-C, P-D, P-7, P-8),
and gw_l5c, a 2017 community "Competition MOD" patch (rev. L-5 patch bc43) for the identical physical
hardware. Every driver shares the same static gwGameData/init_gw pair; all fourteen are physically
identical.