catnmous: fix sprite banking

src/mame/audio/laserbat.cpp

@@ -282,7 +282,7 @@ WRITE8_MEMBER(laserbat_state::csound2_w)
     |   6 | SOUND 5  | PIA CA1     |
     |   7 |          |             |
     |   8 |          |             |
-    |   9 |          |             |
+    |   9 |          | 14L A11     |
     |  10 |          |             |
     |  11 |          |             |
     |  12 |          |             |
@@ -292,6 +292,10 @@ WRITE8_MEMBER(laserbat_state::csound2_w)
     |  16 | RESET    | Unknown     |
     +-----+----------+-------------+
 
+    Bit 9 is used to select the sprite ROM bank.  There's a wire visible
+    on the component side of the PCB connecting it to the high address
+    bit (A11) of the sprite ROM at 14L.
+
     There could well be other connections on the sound board - these are
     just what can be deduced by tracing the sound program.
 
@@ -308,8 +312,12 @@ WRITE8_MEMBER(catnmous_state::csound1_w)
 
 WRITE8_MEMBER(catnmous_state::csound2_w)
 {
+	// the bottom bit is used for sprite banking, of all things
+	m_gfx2 = memregion("gfx2")->base() + ((data & 0x01) ? 0x0800 : 0x0000);
+
 	// the top bit is called RESET on the wiring diagram - assume it resets the sound CPU
 	m_audiocpu->set_input_line(INPUT_LINE_RESET, (data & 0x80) ? ASSERT_LINE : CLEAR_LINE);
+
 	m_csound2 = data;
 }
 

src/mame/drivers/laserbat.cpp

@@ -62,6 +62,9 @@
     * Service coin 1 input grants two credits the first time it's
       pushed, but remembers this and won't grant credits again unless
       unless you trigger the tilt input
+    * The sprite ROM is twice the size as Laser Battle with the bank
+      selected using bit 9 of the 16-bit sound interface (there's a wire
+      making this connection visible on the component side of the PCB)
     * Judging by the PLA program, the colour weight resistors are likely
       different to what Laser Battle/Lazarian uses - we need a detailed
       colour photo of the game board or a schematic to confirm values
@@ -71,7 +74,6 @@
 
     TODO:
     - work out where all the magic layer offsets come from
-    - catnmous sprite ROM appears to be underdumped
     - need to confirm colour weight resistors on catnmous (detailed photo required):
       R58, R59, R60, R61, R62, R65, R66, R67, R68, R69, R72, R73, R74, R75
       (network connected between 11M, 12M, Q5, Q7, Q8)