Nostalgic

1
Nostalgic?

• 8 - Reliability
• 9 - Reliability, Management
• 10 - Thermal
• 11 - Thermal / Mechanical Design. FEA (Joel
  Pedlikin)
• 12 - Management, Cost Schedule, Digital
• 13 - Design workshop / AeroAstro support (4/25)
• 14 - Presentations (4/30)

• 1 - Introduction
• 2 - Propulsion ?V
• 3 - Attitude Control instruments
• 4 - Orbits Orbit Determination
• 5 - Launch Vehicles
• 6 - Power Mechanisms
• 7 - Radio Comms

2
Due Tonight, April 18

• Draft presentations

3
Due Thursday, April 25
 Therapy for presentations  Update on Projects
4
Burglar Alarm Paradox update
Burglar Alarm Reliability 99.9 False alarm
happens 1365 days (1 per year) Chance of being
robbed 1 10,000 houses (or cars) /yr P(alarm
goes off due to robbery) Assume alarm
sounds P(Robbery) 0.0001 P (False)
0.00275 P(False) / P(Robbery) 0.00275 /
0.0001 27.5 1 -27.5 false alarms for every
real robbery If Alarm lives 10 years and false alarm costs
100 Cost 100 x 1 x 10 (buy and keep
alarm) 1000 (250 10 x 12 x 10)
2450 Cost Expected Value of Alarm 0.0001 x
10 x uninsured deductible (maybe 25k) 25
EV
5
If life is a banquet...

• Mission Definition
• Black tie prime rib for 300 at the Ritz
  vs.
• Beer and hot dogs in the park down the street
• Preliminary Design
• Select entré, drinks, desert, type of music
  1st serious cost estimates
• Detailed Design
• bottles of Schlitz / Perrier Jouet, ft2 of
  cake, place markers, of beef may sign up to
  fixed price
• ICD
• Cash bar? Who supplies the flowers? (Flowers?
  What flowers?). Chairs?
• Management and Standards
• Waiters in tuxedos, sommelier and served hors
  douvres vs. buffet
• Build vs. Buy
• Can you really bake those cookies for less than
  7/lb? (and so what!)
• What wont get done while youre busy at home
  baking?

6
What Management Does

• Planning and Predicting
• What can be done at what budget
• How many people of what types for what duration
  necessary to do a job
• Translate that into contracts, deliverables,
  payment schedules and then constantly reworking
  them as the program evolves
• Creating the environment
• Tools, desks, support staff, purchasing, quality,
  inspection
• Compensation, staffing, benefits, incentives, job
  descriptions and interrelations
• Understanding the clients / applications
  requirements
• Measurement and Intervention
• Program revues and other milestones
• Employee assessment, assignments
• Doing something when it isnt working
• Problem solving
• Supposedly you have those grey hairs for a reason
• Picking significant problems out of the noise of
  day-to-day issues (dont do other peoples jobs
  for them)
• Mediating among teams and between team and
  clients / suppliers

7
What Upper Management Does

• Tech Management CTO
• Technical accuracy, quality (no errors state of
  the art)
• Yellow flags coming disruptions (and
  opportunities), dead-end approaches
• Innovating new solutions make the company more
  technical competitive
• Management of the tech staff - what about me?
• Corporate Management COO
• Legal employees, workplace, contracting /
  auditing, patents
• Finding inefficiencies and stomping on them
• Physical Plant leases (space and equipment)
• Contracting and negotiating
• Finance Management CFO
• Business plans and money raising
• Cash management
• Lease v. buy, investing short / medium term
• VP Biz Dev
• Bid / No-bid, proposal prep
• Marketing, advertising, trade shows corp persona
• Dabble in programs -

• CEO
• Why are we here
• Define our biz niche
• New directions
• Growth (or no-growth)
• True to our roots? corp. memory
• Corporate philosophy
• Look and feel
• Employee relations
• Contracting style and client select
• Who works here
• Strategy
• Relationships
• Person behind the curtain
• Mergers / Acquisitions
• Ambassador (icon)
• Rep. to the board
• Per CEOs strengths

8
(No Transcript)
9
(No Transcript)
10
Program Life Cycle

• Development is a learning process.
• Planning is no substitute for actual experience
  The important thing is the Planning, not the
  Plan
• Everything is negotiable.
• Will you build 1 satellite in your life?

Order Component
11
The Dilbert WarsManagement vs. Engineering
12
Documentation

• Basic Rule Dont write what no one will read.
• Go for easy documentation
• Email exchanges - Photographs of everything
• Manufacturers data on purchased parts - Test
  failure logs
• Videos of procedures - Well documented code
• Offer automatic documentation
• Fabrication drawings schematic diagrams -
  Block diagrams
• Synthesized documents worth producing
• ICDs - System Requirements Documents
• (HSwr) - Launch environment
• Cabling diagram - Thermal / Structure analysis
  reports
• Users manual - Test plans results
• Contracts, change orders etc.

13
Operations at a minimum

• GS Locations (arranged by cost impact)
• Central GS Their motivation vs. yours Labor
  intensive Capability exceeds needs.
• Field GS Portable, hardened equipment Virtually
  always backed up at office Minimal Autonomy but
  must be idiot and disaster proof.
• Remote GS Similar to Office but rent person
  to power cycle, maintain, trouble shoot max
  investment in environmental protection (radome,
  foundation, heater / AC, backups)
• Office GS Motivates autonomy Employ existing
  staff Already on your network
• No GS per minute charges only
• GS Staffing
• First 30days Engineering staff some (3)
  present, some on call (everyone), frequent
  telecon and in-person briefings dont forget
  your PR staff
• Day 31 to day 90 Engineering (1 or 2) and Ops
  staff (2 or 3) transition anomaly track.
• Ongoing Ops staff One person plus buddy plus
  on-call. Engineering staff on board via email
  and occasional reviews. Probable budget for
  capabilities upgrades. Possible savings by GS
  sharing (multiple antennas or prioritize)
• Software
• Autonomy and anomalies
• Autonomy is not a risk - its a reliability plus
• down time (LANL fire experience)
• Menu selection vs. freehand composition
• Tracking

• The no GS GS
• Geosynchs
• LEO commsat links
• Receive only GS
• Managing the Remote GS
• Site availability, installation test
• On-site maintenance
• Visits for
• Upgrades
• Alignment and maintenance

14
Keeping Ops Cost Down

• Design-in Autonomy
• Satellites go by at the oddest times... - beepers
• Design must tolerate outages gracefully (to lower
  the cost of a GS failure)
• Intuitive, graphic, quick-look, menu driven
  interfaces
• Simple GS
• Rental and staffing costs will exceed spacecraft
  costs
• Office / lab space is never free - for long
• Pick an orbit that passes over your office
• Assume a 6 month mission
• Manage the transition from the development team
  to the ops team
• Dont break things and then have to fix them
• Allow several months overlap - Agree on command
  authorization levels
• Keep the development team plugged in
• i.e. via email for rapid anomaly resolution

15
Populating your program
16
Scheduling Your Program
17
Issues with Space Processors

• Expensive as they are, and even more expensive to
  customize
• Additional hardware required to talk to your
  devices
• May drive design of other subsystems -
  potentially inefficient designs
• e.g. Aux Bus, readout frequency requirements,
• Not produced used in quantity no large
  investment or test
• development environments often buggy, costly, not
  widely supported
• Note 68020 and 80C81 are easy because market
  has invested billions in them
• Large, heavy, high power
• lack of custom, highly integrated components
• most efficient components dont come in space
  qual (e.g. theyre plastic)
• probably developed for larger missions where
  these features are less critical
• Less processing gazorch
• Coding at low level required - adds cost,
  decreases reliability
• Software writers will tend to be specialists not
  familiar with spacecraft
• May require multiple processors (more mass,
  power, risk)
• May encourage solving problems in hardware (e.g.
  attitude solutions)

18
Space Environment Survival

• 0-g
• doesnt matter
• Vacuum
• check electrolytics, on-board battery, plastic
  outgassing
• Thermal
• Copper backplane and/or processor-mounted
  radiator
• Isolated, high-dissipation parts must be
  heat-sinked
• Temperature range adequate?
• Vibration / Shock
• normally not an issue - may need to stake
  connectors(launch loads are trivial compared
  with most consumer apps.)
• Autonomy
• watchdog timers, multiple copies of on-board RAM
  and ROM
• Radiation Tolerance
• SEU / Latchup / Total Dose

19
Selecting Your Processor

• Not a harmonious process
• Strong individual opinions (religion - like
  Macs vs. PCs, - compromise impossible)
• Huge of candidates to choose among
• Program-wide impact - everyone gets in the act
• Everyone thinks they know something about it
• Disinformation was invented for the processor
  world
• vapor hardware, vapor software, capability
  overstatement
• never admit a bug exists - even after numerous
  customer complaints
•  Some Suggestions
• Scrutinize hardware availability and support
  tools
• Believe NO predicted delivery or availability
  dates for new products
• Create, in advance, a mutually agreed evaluation
  matrix including
• speed (determine whats required) - electric
  power
• radiation (whats really required) - other
  required features
• development environments and platforms
• compatibility with existing software / hardware
• extra features - these are a liability, not a plus

Adapted from thoughts of Jan King,
past-president, AMSAT-NA
20
Centralized Processing and Kings Funnel

• Pros
• Single µP simple architecture
• Central multitasking well understood
• Single state machine easier testing

• Cons
• Kings Funnel (below)
• Fast enough for multitask or no interrupt?
• Interrupt may be overwhelmed
• µP is single point of failure

• Kings Funnel Ingredients
• Software not complete at time of integration
• System bugs start affecting program - mostly
  requiring software resolution
• Hardware and firmware / controller developers
  cant help much
• Lone Ranger has long ago intimidated or
  demotivated all others is only one left who can
  operate spacecraft
• Program will wait for Lone Ranger to emerge
  from Funnel

21
Distributed Processing

• Cons
• Latest and Greatest thing
• poor heritage
• selected for sex appeal?
• pay for teams education
• Multiple state machines
• hard to test / verify state
• crash prone
• May still suffer
• intercept overload
• Kings Funnel
• single point failure(since many multi-processor
  systems are actually hub spoke)
• Lots more electric power

• Pros
• Eliminate Kings Funnel (questionable)
• Eliminate single point failures (also
  questionable)
• Lots of horsepower

22
Features You Will Need

• Reliability
• Watchdog timer
• Multiple systems with toggle / voting
• On-board EDAC
• Self-booting
• Hard O/S copy
• Multiple copies of O/S and applications
• State saving (e.g. flash RAM)
• Compatibility
• Mechanical strength and robustness
• Thermal margin / heat sinking
• Adequate I/O interfaces
• Instrumentable
• Programmatic
• Afford multiple copies (normal for us has been 15
  to 20)
• good, widely used development and support systems
• Sufficient gazorch to enable high level coding
  and easy debugging

23
ALEXIS Block Diagram

• Single 80C86 (commercial, plastic)
• Clocked at 4 MHz
• A and B sides
• Tested after 9 years on orbit
• Modeled on PC backplane Components treated as
  plug in cards
• ACS
• Power controller
• Housekeeping sensors
• Memory
• Tx / Rx
• Payload interface via dual port RAM
• Easy to simulate
• Excellent isolation
• Quasi multi-processor (payload has its own to
  deliver bits to DPR)

24
A whole page on code uplinking

• Pros
• Allow late (post-launch) implementation of
  upgrades
• may save program schedule!
• Can run off uplinked code if native controller
  has significant fault
• Allows work-arounds for in-flight failures and
  aging effects
• Increase autonomy as system is learned and
  confidence gained

• Cons
• Encourages continuous creation of upgrades
• may destroy program schedule!
• Demotivates testing to find and squash bugs in
  controller
• Additional complication in overall system design

25
HETE / TERRIERS Processor
26
HETE Processor Highlights

• Three distinct processor sections Transputer,
  DSP0, DSP1.
• Transputer networked to other Transputers
  board-level redundancyRuns spacecraft
  housekeeping power management, attitude
  control, etc.
• DSP1 runs science code fast at low power.
• DSP0 manages spacecraft bus
• 2 W, no space qual parts, all plastic, commercial
  data bus (Nubus) interface, 30 MIPS

27
Course Goals

• The Design Process Augenblick of a higher level
  of complexity
• Aerospace Engineering An application of
  engineering sciences
• Using Analysis and Design Tools
• Working on something too big to even think about
  doing yourself - Teams
• Systems Engineering Optimize around solutions
• Design and build something, and present it
• See yourself 5, 10, 15, 20 years from now

28
Last slide

• Why you should / shouldnt go further with
  aerospace / space engineering
• Does your field matter?
• Does engineering matter?