Rank · Frame Advance

On this page

Related
Games
References
Sources

Rank makes a game harder as the player does well. Contra counts loops cleared, up to 3, and reads the strongest gun either player holds; both shorten the wait between random soldiers¹. Super C keeps only a flag for a beaten game and counts the players holding the spread gun². Asteroids reads the score where Contra reads loops and guns^3,1. The saucer’s size goes by count and score: the first three saucers large; after that small at random, or always from 30,000 points³. Its aim tightens with the score: the small saucer aims within 16 directions (of 256), then within 8 from 35,000 points⁴.

Games

Game	Year	How	Note
Asteroids	1979	Score	after the third saucer, small saucers come more often and always from 30,000 points; they aim tighter from 35,000
Contra	1987	Loops and guns	random soldiers come faster with each loop cleared, up to 3, and with the strongest gun held
Super C	1990	Loops and guns	enemies come faster once the game is beaten and for each player holding the spread gun

Every claim about this mechanic, by game (4)

Asteroids 1979	Each player's saucer reload value starts at 146 when a game starts and drops by 6 each time a saucer appears, as long as the result stays 32 or more. While it is 128 or more, every saucer is the large one. Below 128, a player with 30,000 points or more always gets the small saucer; under 30,000 a random byte decides, giving the small saucer when the byte is more than half the reload value: about 3 in 4 at the first chance (122) and about 15 in 16 at the floor of 32 (computed, for evenly spread bytes).	Code
Asteroids 1979	Between shots, a live saucer counts down 10 runs of its routine, which runs every fourth frame (40 frames, computed), and the count pauses while the player's ship is waiting to return. The large saucer fires in a random direction, any of 256. The small saucer aims: it takes the ship's position minus its own on each axis, subtracts half the saucer's velocity, turns the pair into a direction with an arctangent routine, and adds a random error of −16 to +15 directions while the player has under 35,000 points, and −7 to +8 from 35,000.	Code
Contra 1987	Each completed loop of the US NES Contra and the strongest gun either player holds both shorten the timer between random soldiers: from the level's start value (144 on level 1), 40 comes off per loop, counting up to 3, and 5 per point of weapon strength (S 3, M and L 2, F 1, the default 0).	Code
Super C 1990	The US NES Super C takes 16 off the delay between random enemies once the game has been beaten and 4 for each living player holding the spread gun. The delay's start is read from an eight-entry per-level table with an index that isn't the level; the disassembly reads it as always `$80`, a value the table doesn't hold, and the atlas hasn't confirmed it. Where Contra counts loops up to 3, Super C keeps a beaten-once flag that stays at 1; where Contra reads every gun's strength, Super C counts only the spread gun.	Code

References

^ a b cEach completed loop of the US NES Contra and the strongest gun either player holds both shorten the timer between random soldiers: from the level's start value (144 on level 1), 40 comes off per loop, counting up to 3, and 5 per point of weapon strength (S 3, M and L 2, F 1, the default 0).Codevermiceli/nes-contra-us

How we know

1736adjust_generation_timer:
1737    jsr get_soldier_gen_timer_for_lvl
1738    sta SOLDIER_GENERATION_TIMER      ; soldier generation timer
1739    ldy GAME_COMPLETION_COUNT         ; load the number of times the game has been completed
1740    beq @adjust_timer_for_weapon      ; branch if haven't yet beat the game any times
1741    tya
1742    cmp #$04                          ; compare GAME_COMPLETION_COUNT to #$04
1743    bcc @continue                     ; branch if GAME_COMPLETION_COUNT < #$04
1744    ldy #$03                          ; maximum multiplier of increase chance for generating soldier is #$03
⋮
1747    lda #$28          ; lower soldier generation timer
⋮
1752    ldy PLAYER_WEAPON_STRENGTH ; load player weapon strength (damage strength)
1753    beq @exit                  ; don't generate if player's weapon strength is #$00
⋮

src/bank2.asm lines 1736–1779 at 687d651 · upstream ↗

3270    inc GAME_COMPLETION_COUNT       ; increment game completion count, used mainly to increase enemy difficulty every play-through

src/bank7.asm line 3270 at 687d651 · upstream ↗

3920set_frame_scroll_weapon_strength:
⋮
3926    sta PLAYER_WEAPON_STRENGTH ; clear player weapon strength

src/bank7.asm lines 3920–3926 at 687d651 · upstream ↗

4096set_player_weapon_strength:
⋮
4102    lda weapon_strength,y         ; load how strong the weapon is
4103    cmp PLAYER_WEAPON_STRENGTH    ; compare against current weapon strength
4104    bcc @exit                     ; exit and do not lower player's current weapon strength
4105    sta PLAYER_WEAPON_STRENGTH    ; store current weapon strength code (#$00-#$03)
⋮
4111weapon_strength:
4112    .byte $00 ; Regular = Weak
4113    .byte $02 ; M = Strong
4114    .byte $01 ; F = Medium
4115    .byte $03 ; S = Very Strong
4116    .byte $02 ; L = Strong

src/bank7.asm lines 4096–4116 at 687d651 · upstream ↗

Checked against the evidence by its author, an AI model, 2026-10-21.

Checked by a second AI model, 2026-10-21.

In the claim ledger

^ a bThe US NES Super C takes 16 off the delay between random enemies once the game has been beaten and 4 for each living player holding the spread gun. The delay's start is read from an eight-entry per-level table with an index that isn't the level; the disassembly reads it as always $80, a value the table doesn't hold, and the atlas hasn't confirmed it. Where Contra counts loops up to 3, Super C keeps a beaten-once flag that stays at 1; where Contra reads every gun's strength, Super C counts only the spread gun.Codevermiceli/nes-super-c · vermiceli/nes-contra-usrevised

How we know

3755enemy_gen_routine_00:
3756    lda CURRENT_LEVEL               ; !(BUG) a is immediately overwritten
3757                                    ; this should probably have been ldy CURRENT_LEVEL
3758    lda level_enemy_gen_delay_tbl,y ; y is actually PRG bank number, e.g. #$33 and not current level
3759                                    ; this causes the delay to always be #$80
⋮
3762    ldy GAME_COMPLETED              ; load whether or not the game has been completed
3763    beq @adj_for_difficulty         ; branch if haven't beaten the game yet
⋮
3770    lda #$10
3771    jsr @set_enemy_gen_delay ; subtract #$10 * GAME_COMPLETED from ENEMY_GEN_DELAY
⋮
3775    ldy ENEMY_DIFFICULTY     ; load current difficulty
3776    beq @adv_routine         ; branch to not adjust initial enemy generation delay
3777    lda #$04                 ; difficulty is non-zero, subtract #$04 * ENEMY_DIFFICULTY from ENEMY_GEN_DELAY

src/bank-0.asm lines 3755–3779 at be6ccce · upstream ↗

4286level_enemy_gen_delay_tbl:
4287    .byte $7b,$53,$53,$53,$53,$53,$53,$53

src/bank-0.asm lines 4286–4287 at be6ccce · upstream ↗

165; calculate ENEMY_DIFFICULTY based on number of active players with S weapon
⋮
172    lda PLAYER_CURRENT_WEAPON,x ; 0 = Regular, 1 = M, 2 = S, 3 = L, 4 = F
173    and #$0f                    ; strip rapid fire flag
174    cmp #$02                    ; compare to S weapon
175    bne @next_player            ; move to next player if current player doesn't have S weapon
176    iny                         ; increment number of players in normal state that have the S weapon
⋮
181    sty ENEMY_DIFFICULTY          ; set enemy difficulty

src/bank-0.asm lines 165–181 at be6ccce · upstream ↗

2421    lda GAME_COMPLETED          ; load whether or not the game has been completed
2422    adc #$00                    ; increment game completion count (carry is set)
2423    cmp #$02                    ; see if game already completed
2424    bcc @continue               ; branch if first time completing game to set value to #$01
2425    lda #$01                    ; game already beaten at least once, keep value of #$01 (don't set to #$02)

src/bank-f.asm lines 2421–2431 at be6ccce · upstream ↗

1736adjust_generation_timer:
⋮
1742    cmp #$04                          ; compare GAME_COMPLETION_COUNT to #$04
1743    bcc @continue                     ; branch if GAME_COMPLETION_COUNT < #$04
1744    ldy #$03                          ; maximum multiplier of increase chance for generating soldier is #$03
⋮
1752    ldy PLAYER_WEAPON_STRENGTH ; load player weapon strength (damage strength)

src/bank2.asm lines 1736–1757 at 687d651 · upstream ↗

4111weapon_strength:
4112    .byte $00 ; Regular = Weak
4113    .byte $02 ; M = Strong
4114    .byte $01 ; F = Medium
4115    .byte $03 ; S = Very Strong
4116    .byte $02 ; L = Strong

src/bank7.asm lines 4111–4116 at 687d651 · upstream ↗

Checked against the evidence by its author, an AI model, 2026-10-21.

Revised after a check by a second AI model, 2026-10-21.

In the claim ledger

^ a b cEach player's saucer reload value starts at 146 when a game starts and drops by 6 each time a saucer appears, as long as the result stays 32 or more. While it is 128 or more, every saucer is the large one. Below 128, a player with 30,000 points or more always gets the small saucer; under 30,000 a random byte decides, giving the small saucer when the byte is more than half the reload value: about 3 in 4 at the first chance (122) and about 15 in 16 at the floor of 32 (computed, for evenly spread bytes).CodeComputer Archeology, Asteroidsrevised
How we know
LDA #$92 STA $02F8 ; {ram.saucerTimeReload} … STA $02F7 ; {ram.saucerTimer} Countdown Timer For When Saucer Appears
content/Arcade/Asteroids/Code.md#L136-L140 · upstream ↗
LDA $02F8 ; {ram.saucerTimeReload} SEC SBC #$06 ; Shorten Time Between Saucer Appearence CMP #$20 ; At Lowest Value? BCC $6BDD ; {} Yes, Branch STA $02F8 ; {ram.saucerTimeReload} Make The Update
content/Arcade/Asteroids/Code.md#L483-L488 · upstream ↗
LDX #$02 ; Start With Large Saucer LDA $02F8 ; {ram.saucerTimeReload} Start Checking Score When @ #$7F And Below BMI $6C30 ; {} Not There Yet, Branch Around Score Check … CMP #$30 ; 30,000 Points Or More? BCS $6C2F ; {} Yes, Branch JSR $77B5 ; {} STA $08 ; {ram.$08} LDA $02F8 ; {ram.saucerTimeReload} LSR A CMP $08 ; {ram.$08} BCS $6C30 ; {} Going To Be A large Saucer, Branch …
content/Arcade/Asteroids/Code.md#L512-L527 · upstream ↗
Checked against the evidence by its author, an AI model, 2026-10-22.
Revised after a check by a second AI model, 2026-10-22.
In the claim ledger
^ a bBetween shots, a live saucer counts down 10 runs of its routine, which runs every fourth frame (40 frames, computed), and the count pauses while the player's ship is waiting to return. The large saucer fires in a random direction, any of 256. The small saucer aims: it takes the ship's position minus its own on each axis, subtracts half the saucer's velocity, turns the pair into a direction with an arctangent routine, and adds a random error of −16 to +15 directions while the player has under 35,000 points, and −7 to +8 from 35,000.CodeComputer Archeology, Asteroids
How we know
LDA $5C ; {ram.fastTimer} AND #$03 BEQ $6B9A ; {} RTS
content/Arcade/Asteroids/Code.md#L457-L460 · upstream ↗
LDA $1C ; {ram.numPlayers} Number Of Players In Current Game BEQ $6C4E ; {} None, Branch LDA $02FA ; {ram.shipSpawnTimer} BNE $6C53 ; {} DEC $02F7 ; {ram.saucerTimer} Time for saucer to shoot? BEQ $6C54 ; {} Yes, branch? RTS LDA #$0A ; Time Between Saucer Shots STA $02F7 ; {ram.saucerTimer} Reset for next shot LDA $021C ; {ram.statusSaucer} Saucer Flag LSR A ; Check Saucer Size BEQ $6C65 ; {} Branch If Small One …
content/Arcade/Asteroids/Code.md#L536-L604 · upstream ↗
TYA BPL $76FC ; {} … EOR #$80 EOR #$FF … LDA #$20 RTS JSR $776C ; {} LDA $772F,X ; {}
content/Arcade/Asteroids/Code.md#L1867-L1898 · upstream ↗
Checked against the evidence by its author, an AI model, 2026-10-22.
Checked by a second AI model, 2026-10-22.
In the claim ledger

Sources

Code

(vermiceli), Michael Miceli. 2026. “Contra (US), annotated disassembly (vermiceli).” Commit 687d651. vermiceli/nes-contra-us @ 687d651 · licence, all uses
(vermiceli), Michael Miceli. 2026. “Super C (US), annotated disassembly (vermiceli).” Commit be6ccce. vermiceli/nes-super-c @ be6ccce · licence, all uses
Howell, Lonnie, Mark McDougall, and Christopher Cantrell. 2026. “Computer Archeology: Asteroids.” Commit 1adc2a7. topherCantrell/computerarcheology @ 1adc2a7 · licence, all uses

Related

See also

Games

References

Sources

Code