Shop OBEX P1 Docs P2 Docs Learn Events
Use of non military/space grade IC's in space — Parallax Forums

Use of non military/space grade IC's in space

skylightskylight Posts: 1,915
edited 2014-07-31 17:21 in General Discussion
I can't remember where I read this but was quite shocked to see that due to increasing costs that there is talk that off the shelf devices could shortly be finding themselves in spacecraft.
Now does this mean that the rigorous testing of old is about to be stopped? I'm not saying that IC's etc are not reliable if specced correctly but that's on earth not in the hostile environment of space and lives can be at stake because of failure.
Just wondered what others views are on the subject, is every human endevour now going to be based on cost over safety?

Found the news item ( http://www.bbc.co.uk/news/science-environment-12253228 ) and I may be being hasty in thinking lives may be at risk but in the future maybe accountants will drive what spacecraft utilise?
«1

Comments

  • Heater.Heater. Posts: 21,230
    edited 2011-01-31 12:35
    If you crack open a Primary Flight Computer box from a Boeing 777, the thing that does the fly by wire, you will find a 486 processor, a 68000 and something processor and a AMD 2900 processor. These were pretty much off the shelf components when it was designed. Perhaps selected from the military grades. The reliability comes from the multiple redundant processors in the boxes and multiple-redundant boxes as much as from the quality of the chips.

    Now, I have no idea about space hardened processors etc, I've just imagined they must be getting harder to find as they get smaller and smaller and less able to withstand cosmic ray strikes. But perhaps they are small enough and power frugal enough that reliability can be obtained from multiple redundant systems. Rather than spending a lot of money building bigger, slower, space hardened chips.
  • Mike GreenMike Green Posts: 23,101
    edited 2011-01-31 12:35
    Do remember that military / space grade ICs become that way in two ways ... by testing and by design. Testing makes sure that the parts function properly over a temperature range that commercial parts are not required to function at. By design refers to several things: 1) The materials used are intended for use over a military temperature range with quick transitions from one extreme to another. 2) The IC processes and physical layout of the chip are more robust to radiation than commercial designs by careful attention to voltage gradients, current levels, insulation thickness, materials choices, etc.

    Early on in the space race, we didn't necessarily understand chip failure mechanisms. Over the years, failure mechanisms have become better understood and much of what we've learned has become common-place in commercial IC development and production. Witness Parallax's progression from commercial grade parts in their Stamp modules to industrial grade parts. The Propeller has already been characterized for military temperature range use.

    Radiation hardening is another issue. That requires special materials, but it also limits designs to older parts, partly because newer parts (mostly memories) are inherently less radiation resistant than older parts because of thinner oxide layers, lower supply voltages and currents, etc. To achieve radiation hardening, systems have to be designed with redundancy and error recovery and that can be done with fewer radiation hardened parts.

    Keep in mind also that increased cost doesn't necessarily mean increased reliability. Radiation hardened parts are not fail-proof. They just hold up on average longer than non-hardened parts. You still have to design the systems for failures.
  • edited 2011-01-31 13:00
    Heater. wrote: »
    If you crack open a Primary Flight Computer box from a Boeing 777, the thing that does the fly by wire, you will find a 486 processor, a 68000 and something processor and a AMD 2900 processor. These were pretty much off the shelf components when it was designed. Perhaps selected from the military grades. The reliability comes from the multiple redundant processors in the boxes and multiple-redundant boxes as much as from the quality of the chips.

    The internet was a military application but we now use it. It has been years from when I heard this but I believe the 68000 was designed for space because they needed chips that could survive radiation and the technology was later sold for civilian use. If you research AMD or Intel along with other chipmakers, some of these companies received a lot of government contracts and which is why the 8088 or 8086 has become the monster that it is today.
  • edited 2011-01-31 13:16
    A lot of things like Velcro were developed for the space program.
  • skylightskylight Posts: 1,915
    edited 2011-02-01 08:23
    Thanks for the informative replies, I suppose thinking about it, now China is commited to a space program they will be most probably be using home grown components and well we have heard on this forum about some of the things that come from that part of the world so perhaps I shouldn't be worrying so much?
  • HollyMinkowskiHollyMinkowski Posts: 1,398
    edited 2011-02-01 14:47
    I can attest to the fact that loads of common items and parts
    are used in aerospace and military applications. Mil-spec and
    rad-hard parts are not always required and when used add a lot
    of cost to projects.

    Here is an interesting blurb about this from "COTS Journal" .."The Journal of
    Military Electronics & Computing"

    COTS (kots), n. 1. Commercial off-the-shelf. Ter-
    minology popularized in 1994 within U.S. DoD by
    SECDEF Wm. Perry’s “Perry Memo” that changed
    military industry purchasing and design guidelines,
    making Mil-Specs acceptable only by waiver. COTS
    is generally defined for technology, goods and services
    as: a) using commercial business practices and specif-
    cations, b) not developed under government funding,
    c) offered for sale to the general market, d) still must
    meet the program ORD. 2. Commercial business
    practices include the accepted practice of customer-
    paid minor modification to standard COTS products
    to meet the customer’s unique requirements.
    —Ant. When applied to the procurement of
    electronics for the U.S. Military, COTS is a pro-
    curement philosophy and does not imply commer-
    cial, office environment or any other durability grade.
    E.g., rad-hard components designed and offered for sale
    to the general market are COTS if they were developed
    by the company and not under government funding.
  • rod1963rod1963 Posts: 752
    edited 2011-02-01 15:33
    And here's a example of a Rad-hard system for the upcoming Lunar Reconnaissance Orbiter.

    Link:
    http://www.klabs.org/DEI/Processor/PowerPC/rad750/papers/spacewire_con_2007.pdf
  • bee_manbee_man Posts: 109
    edited 2011-02-02 20:39
    NASA used off the shelf componants for both Mars landers and they exceeded their life expectancies. They also had several failures, but the 900 million "old style" Mars Observer was lost just before entering Mars orbit, the 200 million Mars Global Surveyor basically did the same job. It seems component quality is only one factor. Keeping your fingers crossed seems to be another. Unfortunately spending a ton of money on safety does not guarentee safety.
  • rod1963rod1963 Posts: 752
    edited 2011-02-03 11:20
    Well most of the funds often goes for a lot things in a probe or satellite, the rad-hard ic's are just small change in the overall cost. For example a rad-hard 32bit processor like the Mongoose(MIPS R3000) costs around $22,000 a piece.
  • skylightskylight Posts: 1,915
    edited 2011-02-05 11:27
    Again some interesting reading thanks all for the input, If i'm completely honest I suppose i'd rather they cut costs IF it means the space programme continues to expand because of it.
  • RickInTexasRickInTexas Posts: 124
    edited 2013-05-12 09:52
    Having worked for a major Avionics manufacturer, all parts were of the general off-the-shelf variety.
    The difference was in the testing and burn-in performed on the top grade assemblies, e.g. Commercial TACAS
    collision-avoidance transponders were thoroughly and repeatedly tested at -40 to 85°C IIRC, with extended burn-in
    times.

    I would actually feel much safer flying with equipment built/tested in the above manner than something simply built with mil-spec
    parts and shipped, because it is tested as a unit.
  • CuriousOneCuriousOne Posts: 931
    edited 2013-05-12 11:07
    recently ordered samples from TI. One of delivered ICs was military & aerospace degree, while i've ordered normal one
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2013-05-12 11:45
    About the only thing I take serious is getting good Military Specification motor oil.

    The truth is that any mission critical military and aerospace chips are likely never mentioned on the internet, may use completely different semi-conductor technology that is faster, and would land you in jail as a spy just for having in your possession.

    We only get the mundane stuff, like the motor oil. And it is decent quality. The world would be a hellish place without somebody doing quality control and quality assurance.

    What's actually in space.. you will never know. If you find out, they will shoot you. Take the Iridium Satellite System for instance....
  • CircuitBurnerCircuitBurner Posts: 21
    edited 2013-05-12 15:07
    My grand father worked on a nuclear rocket engine at Jackass Flats, designed to propel ships into and out of interplanetary flight, called NERVA.
    He has told me things that would get us both jailed prolly, lol..and I can say with confidence that we have lost much over the years as we gain in other areas.
    I will keep it simple like that.
  • jonesjones Posts: 281
    edited 2013-05-12 17:27
    Space isn't one environment. Low earth orbit is generally protected by the earth's magnetosphere, but lunar orbit or the lunar surface isn't. Even in low earth orbit the specifics make a difference, since a spacecraft that passes through the South Atlantic Anomaly is at higher risk than one that doesn't. Mars doesn't have much of a magnetic field, so no protective magnetosphere to speak of. Jupiter is another problem entirely with intense radiation that would likely kill non-rad hardened parts in short order. Levels of solar activity/solar protons also make a difference.
  • Beau SchwabeBeau Schwabe Posts: 6,566
    edited 2013-05-12 19:28
    Don't confuse the military spec'd 54 series vs. 74 series assuming that the 54 series are radiation hardened.... One of the main things that make the 54 series more robust is that the DRC rules are gauged further out for temperature and metal migration. For example a typical 74 series IC is gauged for 10 years / 125 Deg C where a 54 series is gauged for 25 years / 150 Deg C. ... but this has nothing to do with being hardened for radiation. Metal migration is a very slow almost fluid like motion of metal over time. DRC rules are applied so that this movement is guaranteed as to not produce a "short" or bridge over a specified period of time with two or more adjacent wires. Current running through a wire creating heat speeds this process along, but it also occurs naturally over time without any current passing through the wire at all.
  • CircuitBurnerCircuitBurner Posts: 21
    edited 2013-05-12 19:42
    and dont forget about "tin whiskers",,hehe
    the Rohs compliant solder -without lead- has the nasty habit of growing conductive crystalline protrusions over apparently not a lot of time...
    I bet anything the military would use in high danger / high liability situations IS NOT Rohs compliant, lol
  • ercoerco Posts: 20,256
    edited 2013-05-12 19:50
    NASA's 1997 Pathfinder mission set numerous precedents, among them using an off the shelf radio modem on MARS.

    Great writeup: http://www.electricity-today.com/et/oct97/mars.html
  • CuriousOneCuriousOne Posts: 931
    edited 2013-05-12 21:40
    About two years ago, there were images of dismantled Javelin anti-tank missile floating around the internet. 2N3055 was clearly visible, I doubt it was military grade :D
  • Heater.Heater. Posts: 21,230
    edited 2013-05-13 00:29
    Beau,
    ...For example a typical 74 series IC is gauged for 10 years / 125 Deg C...
    That is very interesting. It's something that I have been aware of but never really thought about much.

    Is there any data on how long this metal migration takes to kill parts verses temperature and/or current. It's unnerving to think that my big stash of 74 chips and assorted old microprocessors and RAMs/ROMs is slowly dying of metal migration even without being used. They are already decades old:(

    Here the biggie though:

    How many years are the Propellers gauged for. At what temperatures/currents?
  • Beau SchwabeBeau Schwabe Posts: 6,566
    edited 2013-05-13 05:40
    Heater.

    "How many years are the Propellers gauged for. At what temperatures/currents?" ... The standard for TSMC's DRC rules are set at 10 years /125C which is what we used. This number though is analogous to a Power rating and does NOT necessarily mean that after 10 years they will stop working. Keep in mind also, that this value is more than likely derated a fair amount also. Usually you notice the metal migration failure in memories, since the DRC rules that apply to memory are generally 'tighter' than other parts of the design....AND that memories will usually experience more "electrical exercise" that other parts of the design.

    Here is a quote from one of the Design sheets I have:
    "These rules for aluminum alloy guarantee 10 years use life, with less than 0.01% cumulative failure."
  • david314david314 Posts: 6
    edited 2013-07-08 07:41
    We had the opportunity to abuse some props by bombarding them with 2GeV protons using a synchrotron. The beam had a particle density of ~10^8/s*cm² and while the PICs stood their ground for ~8..15min without apparent malfuntion the propeller chip did IMMEDIATELY latchup every time we turned on the beam, I~600mA seems to be a latchup anyway. The propeller could reboot after the event, so the only apparent use in a radiation environment seems to be as a canary... Any ideas why it would do that?
  • skylightskylight Posts: 1,915
    edited 2013-07-08 11:31
    david314 wrote: »
    We had the opportunity to abuse some props by bombarding them with 2GeV protons using a synchrotron......... Any ideas why it would do that?
    It was probably the Electromagnetic Radiation Causing Overheating.
  • PublisonPublison Posts: 12,366
    edited 2013-07-08 11:55
    skylight wrote: »
    It was probably the Electromagnetic Radiation Causing Overheating.

    You mean erco is causing this! :)
  • skylightskylight Posts: 1,915
    edited 2013-07-08 12:28
    It's those flamethrowers again!!
  • ercoerco Posts: 20,256
    edited 2013-07-08 13:57
    I am undone! :)
  • Beau SchwabeBeau Schwabe Posts: 6,566
    edited 2013-07-08 14:02
    david314,

    I am not sure why it would be so sensitive. My guess is the sensitive / finicky PLL circuitry on the Propeller that is the culprit in this case that may otherwise not exist on a PIC. I would think shielding would improve your results.
  • david314david314 Posts: 6
    edited 2013-07-08 16:55
    Dear Beau,

    these were not 2MeV protons that could be shielded quite easily but 2GeV particles. Shielding would be ... cumbersome. Anyway these are the cosmic ray particles one would expect to be most effective on bringing a circuit down as shielding is not really possible and they are still quite common in space (of course not with this flux rate). The circuit board is still active from the treatment (~2µSv/h) and we will have to wait for it to cool down before looking at the components. The result is - to say the least - quite disappointing, we had hoped that the prop would have some resilience of at least a few minutes before going into whatever event. The charge injection into the whole chip is of the order of 10^8 Protons or ~16pC before failure after ~a second, so I suspect that something else is happening here. Maybe an inner photoeffect of secondary particles/gamma rays in the substrate that ignite some parasitic thyristors. Unfortunately it is not easy to pinpoint the location where exactly the event is taking place.
  • evanhevanh Posts: 15,923
    edited 2013-07-09 05:48
    jones wrote: »
    Space isn't one environment. Low earth orbit is generally protected by the earth's magnetosphere, but lunar orbit or the lunar surface isn't.

    There is some protection from solar winds, particularly on the near side where the tail of the magnetosphere offers protection for a few days each month. And also when it's dark, obviously. But yeah, side on is going to get the full brunt twice a month.

    I'm guessing geosynchronous orbit has limited continuous protection by the magnetosphere.

    PS: I found this interesting titbit about the moon crossing earth's tail: "During the crossing, the moon comes in contact with a gigantic “plasma sheet” of hot charged particles trapped in the tail. The lightest and most mobile of these particles, electrons, pepper the moon’s surface and give the moon a negative charge.

    On the moon’s dayside this effect is counteracted to a degree by sunlight: UV photons knock electrons back off the surface, keeping the build-up of charge at relatively low levels. But on the nightside, in the cold lunar dark, electrons accumulate and surface voltages can climb to hundreds or thousands of volts.

    Imagine what it feels like to be a sock pulled crackling from a dryer. Astronauts on the moon during a magnetotail crossing might be able to tell you. Walking across the dusty charged-up lunar terrain, the astronauts themselves would gather a load of excess charge. Touching another astronaut, a doorknob, a piece of sensitive electronics -- any of these simple actions could produce an unwelcome discharge. “Proper grounding is strongly recommended,” says Stubbs."
  • evanhevanh Posts: 15,923
    edited 2013-07-09 06:03
    david314 wrote: »
    Unfortunately it is not easy to pinpoint the location where exactly the event is taking place.
    Turn off the PLL. Use direct clocking at 5 MHz.
Sign In or Register to comment.