AI summary by TwinMind
How the Integrated Circuit Took Us to the Moon
Asianometry
19m 52s
NASA Apollo Challenge and Navigation Problems
- President Kennedy challenged America to put man on moon and return safely in 1962
- US had only 15 minutes total human spaceflight experience at time
- Major challenge: navigation system to guide spacecraft 400,000 km through space
- Required precise lunar orbit entry at 110 km above moon surface
- Velocity precision critical: few percentage points error = crash or erratic orbit
- Pioneer 4 example: small timing error caused 27,000 km moon miss
MIT Instrumentation Lab Contract and System Design
- August 1961: NASA awarded first lunar mission contract to MIT Instrumentation Lab
- Lab later renamed Draper Lab (1970), spun off from MIT (1973) due to weapons work protests
- System based on previous Polaris ballistic missile work (submarine-launched)
- Inertial system design: self-contained with basketball-sized sensor, 3 gyroscopes
- General purpose computer calculated position, orientation, velocity
- Astronauts took star readings using sextant, fed observations to computer
- Drift measurement: find known star, center optics, press button - computer handles rest
- Approach controversial: astronauts (former pilots) wanted aircraft-style control vs mission control ground guidance
- MIT ultimately won the debate
Alternative Computer Approaches and Reliability Concerns
- Apollo had multiple computers: guidance computer + Saturn V launch computer
- Saturn V launch computer: less sophisticated, no user interface, provided rocket guidance/navigation
- IBM supplied Saturn computer using traditional transistor technology (not integrated circuits)
- IBM determined integrated circuits not mature enough for volatile, high-temperature rocket environment
- IBM solution: TMR (Triple Modular Redundancy)
- Three identical computer circuits
- Voting circuit selected majority output if one failed
- Method worked perfectly - never failed during operation
- Survived November 1969 lightning strike that wiped out Apollo command module electronics
- AT&T recommended IBM supply guidance computer using TMR approach
- MIT wanted more flexible, generalized approach requiring emerging technology
Electronics Evolution: Tubes to Transistors to Integrated Circuits
- Vacuum tubes: unreliable, power hungry, extremely fragile
- First transistor invented 1947: hailed as vacuum tube replacement for low power applications
- Early germanium transistors had problems:
- Manufacturing difficulties
- Unreliable (broke when shaken or exposed to extreme temperatures)
- Required hand connections
- 1953: IBM 608 first transistorized digital calculator
- 3,000 germanium transistors
- 50% smaller than predecessor
- 90% less electricity usage
- Mid-1950s: US military developed Minuteman ICBM for deterrence/second strike capability
- 1958: North American Aviation's Autonetics chosen as prime contractor
- Transistor suppliers: Texas Instruments, Fairchild Semiconductor (Mountain View, CA)
- Fairchild subjected silicon transistors to centrifuge testing, hot/cold temperatures
- 1960: Military bought more transistors for missiles than space programs ($52M vs $33M)
Integrated Circuit Development and Advantages
- Problem: transistors required manual connections (soldering), causing reliability issues
- Integrated circuit advantage: single silicon crystal with millions of transistors, resistors, capacitors
- Eliminated human manual effort, provided economies of scale
- 1958: Jack Kilby (Texas Instruments) invented integrated circuit (Nobel Prize 2000)
- Kilby's version not commercially viable: silicon elements connected with individual gold wires
- 1959: Fairchild developed planar process
- Silicon dioxide layer previously considered manufacturing nuisance
- Turned into benefit: protected transistors from outside environment
- Combined with photolithography for aluminum wire connections
- Created monolithic integrated circuit on single silicon piece
- Texas Instruments invented, Fairchild made commercially viable
- Companies cross-licensed methods, Fairchild processes dominated industry
Military Adoption and Minuteman II Project
- 1962: Minuteman II project required more powerful guidance system
- Autonetics chose integrated circuits with Texas Instruments, Westinghouse suppliers
- TI developed 24+ IC designs for Minuteman II
- 1965: supplying 15,000+ circuits weekly for missile project
- Minuteman II guidance computer performed dramatically better than Minuteman I
- Word spread through military, eventually reached MIT lab and NASA
Apollo Mission IC Integration Decision
- Apollo engineers knew about ICs (cutting edge electronics development)
- Electronics budget breakdown: 40% boosted rocket, 70% spacecraft, 90% tracking/data
- MIT ordered IC samples from TI (1959-1960) soon after invention
- 1961: Fairchild announced micrologic product line at $120 each (~$1,200 today)
- MIT testing showed promise:
- 2x faster computational operations (multiplication, addition, branch logic)
- 2/3 size and weight reduction
- Cheaper due to economies of scale (34 logic modules reduced to 1)
- Challenge: radical redesign required, single supplier dependency (Fairchild)
- 1962: MIT received permission to rebuild guidance computer with integrated circuits
Block 1 and Reliability Debates
- Block 1: first iteration with 4,100 integrated circuits (all same type for reliability/procurement)
- Major question: can IC-based computer be relied upon?
- AT&T recommended TMR approach for guidance computer
- NASA contracted IBM to study adapting Saturn V computer for Apollo guidance
- IBM claimed feasibility, but MIT computer was 10-20x faster, more flexible
- Position/velocity outputs 5x more accurate than Saturn computers
- Grumman Aircraft (lunar module builder) proposed separate backup system
- Political pressure: Minnesota politician Joseph Carth questioned MIT system reliability
- NASA response: Apollo carried two guidance computers plus Houston command capability
Mercury Atlas 9 Incident and Final Decision
- May 1963: Mercury Atlas 9 electronics failed near flight end
- Pilot Gordon Cooper manually controlled re-entry and safe return
- Cause: acids in zero-g environment entered electronic system
- Incident occurred during Block 2 development (version for all manned Apollo missions)
- Block 2 improvements:
- 5,500+ integrated circuits vs Block 1's 4,100
- Smaller footprint, 17 pounds lighter
- 14 Block 1 prototypes: 38,000+ reliable operating hours by 1965
- MIT decision: seal computer from outside environment rather than allow mid-flight repairs
- NASA built Abort Guidance System backup (TRW supplier, Signetics ICs)
- 1975 NASA reflection: resources better spent reinforcing primary system reliability
Industry Impact and Silicon Revolution Timeline
- Apollo/Minuteman II helped fund integrated circuit era
- 1962: ICs cost $20-30 each, not cheaper than discrete circuits
- NASA/military paid premium prices, helped companies but commercial customers hadn't adopted
- 1961: Fairchild sold <5,000 micrologic pieces
- Texas Instruments forced to cut pricing from $435 to $76 within 90 days
- 1964: Industry takeoff when Robert Noyce/Fairchild sold ICs below cost
- Before price cut: 2.5M ICs total industry sales
- After cut: single Fairchild order for 500,000 ICs (20% of previous total)
- 1965: Gordon Moore formulated Moore's Law, Burroughs bought 20M integrated circuits
- 1966: Entire industry recognized ICs as future
- Conclusion: Apollo didn't kickstart silicon revolution, but tapped and rode it to the moon
