back to article 'Camera-shy' Raspberry Pi 2 suffers strange 'XENON DEATH FLASH' glitch

An odd little flaw in the new Raspberry Pi 2 has been discovered just days after its geekgasmic release – the PC doesn't like being papped under the bright lights of cameras. It apparently freezes when a Xenon flash is used to photograph the Raspberry Pi 2. The glitch was flagged up in a forum thread entitled "Why is the Pi2 …

Page:

  1. Anonymous Coward
    Facepalm

    E.M.P. anyone ?

    A xenon flash is caused by a massive discharge of a capacitor across the xenon-filled tube, so the resulting electromagnetic pulse might be the real culprit here...

    Has anyone tried a camera with a LED flash yet ?

    (Because I seriously doubt it's the light itself that causes the problem...)

    1. Anonymous Coward
      Anonymous Coward

      Re: E.M.P. anyone ?

      Blutack on the offending package (u16) 'fixes' the issue so photoelectric effect is what's causing the problem.

      1. Mikerr

        Re: E.M.P. anyone ?

        It's not EMP - a 100mW laser pointer pointed at the chip will also do it.

        1. Simon Harris

          Re: E.M.P. anyone ?

          Probably unlikely in an English winter, but I wonder if the problem has been replicated taking a Pi 2 outside and exposing it to sunlight.

          1. Destroy All Monsters Silver badge

            Re: E.M.P. anyone ?

            Morlocks heart the RasPi2

          2. steogede

            Re: E.M.P. anyone ?

            > Probably unlikely in an English winter

            What do you mean? Winter is when we get most sunny days, even with the shorter days there's probably more hours when it is sunny. It's summer when it tends to be grey and overcast.

            1. This post has been deleted by its author

            2. Robert Baker
              Joke

              Re: E.M.P. anyone ?

              Winter is when we get most sunny days, even with the shorter days there's probably more hours when it is sunny. It's summer when it tends to be grey and overcast.

              Surely it's summer when the rain is warm.

        2. cortland

          Re: E.M.P. anyone ?

          100 mW? That's not a pointer, it's laser knife!

      2. Anonymous Coward
        Anonymous Coward

        Re: E.M.P. anyone ?

        "so photoelectric effect is what's causing the problem."

        Someone obviously cut costs a bit too much when it came to setting the thickness of the chip packaging.

    2. This post has been deleted by its author

  2. Anonymous Coward
    Anonymous Coward

    Bizarre, but in the interests of science...

    Yup - just tried this, as I happen to be using my shiny new Pi+ as a desktop. It falls over - no, it's felled as though it had been poleaxed, by the titchy flash on my old Nikon P5000 coolpix. If it's an electromagnetic effect, there goes taking the Pi+ into space unless the ISS has loads of bluetak on board. I'll try duct tape, though, a much more likely option in space.

    1. PaulM 1
      Alien

      Re: Bizarre, but in the interests of science...

      Surrey satellites told me that they sometimes use lead plates in front of and behind commercial grade microprocessors as an alternative to buying space certified micros costing a hundred times as much. Perhaps you could run your Raspberry Pi 2 in a lead box when you take it to the ISS.

      1. Anonymous Coward
        Anonymous Coward

        Re: Bizarre, but in the interests of science...

        > when you take it to the ISS

        Oh please, can I, can I, can I? :-)

        Unfortunately I can confirm that a piece of black duct tape over U16, just inland of the power socket, sorts the problem, so no ISS trip for me. And no smurf poo needed either.

        1. Sykobee

          Re: Bizarre, but in the interests of science...

          You probably should use electrical tape, not duct/duck/etc tape.

          Pretty much a non-issue - most systems will be cased or not exposed to Xenon light sources. Interesting though. Silicon is quite transparent to infrared light apparently, hence the Xenon light (which has massive infrared spikes) penetrates the unshielded component. The red laser pointer requires the light to reflect/diffuse under the package onto the surface of the chip, so merely epoxying the sides works for this.

        2. Anonymous Coward
          Anonymous Coward

          Re: Bizarre, but in the interests of science...

          >> when you take it to the ISS

          >Oh please, can I, can I, can I? :-)

          As someone who even finds some of the rides at Disneyland to be too much can I be the first to say not everyone gets to be an astronaut when they grow up (saw that saying once on a poster with a blown up picture of McDonald's fries). You can bring me a postcard though.

      2. Bobbo

        Re: Bizarre, but in the interests of science...

        Those are mostly to stop cosmic rays causing single event upsets. Although it would also stop the light in this case...

        1. Destroy All Monsters Silver badge
          Paris Hilton

          Re: Bizarre, but in the interests of science...

          On the ISS? I hope that little device has ECC RAM (or at least a memory page hashsum check algorithm)?

      3. Joe Gurman

        Re: Bizarre, but in the interests of science...

        Sorry, but I must cry "Bollocks." Surrey satellites are protecting their chips from energetic particles (e.g. in the Van Allen belts around the earth); the protection from light could be accomplished by something much flimsier. My guess is that the packaging of the chop is to some extent permeable to UV or EUV light, with an energy (h*nu) similar to the work function of the semiconductor.

        And by the way, lead make a terrible energetic particle shielding: it spells more secondary ions than it stops primaries. Indium or tantalum, that's yer stuff.

    2. bill 27
      Alien

      Re: Bizarre, but in the interests of science...

      Oddly enough in the 70's I subscribed to a monthly newsletter from NCAR (I think) which had solar event projections in it. The big mainframe my equipment was attached to crashed pretty much about the same time as the predictions.

      1. Wzrd1 Silver badge

        Re: Bizarre, but in the interests of science...

        "The big mainframe my equipment was attached to crashed pretty much about the same time as the predictions."

        Yeah, been there, done that. Bad enough getting spikes in power lines, ground current differentials and even induced voltages along long conductors raise merry hell with electronics, especially computers and especially with supercomputers and mainframes.

    3. Wzrd1 Silver badge

      Re: Bizarre, but in the interests of science...

      "I'll try duct tape, though, a much more likely option in space."

      Not really, they switched to Kapton tape decades ago. Regrettably, Kapton transmits IR along nicely, which leaves you with the original problem.

      Xenon flash tubes release loads of IR and near IR light. Silicon junctions, in particular PN junctions will act like a photodiode.

      So, personally, I'd go with black epoxy. I'd even go with plumber's epoxy (kneaded two part in a roll) in a pinch. Duck tape, nope. Nothing conductive on a PCB with solid state electronics, thank you.

  3. Anonymous Coward
    Go

    Amazing...

    ...but the thing still comes back up - I am still going to get one (running Slackware of course).

    1. Wzrd1 Silver badge

      Re: Amazing...

      Heh, one enterprising individual plopped an oscilloscope onto the 3.3 volt line and hit his flash on his camera.

      Beautiful waveform showing photvoltaic effect and decay of excitation when disconnected from power.

      Similar waveform, with transition to zero, a bit of negative and spiking near 5 volts ensued.

      Considering the 10cm light source to component, I'd strongly suspect no significant effect in strong sunlight.

      1. Robert Baker
        Joke

        Re: Amazing...

        Considering the 10cm light source to component, I'd strongly suspect no significant effect in strong sunlight.

        Indeed; in the latter case, the distance to the light source is 150*10^9 metres.

    2. asdf

      Re: Amazing...

      What's the over/under on how long Slackware can resist the freedesktop.borg (systemd, etc)?

  4. Flocke Kroes Silver badge

    Advantages of an opaque box

    Tiny parts often have a warning in the data sheet. The packaging is so thin that light can get through turning diodes and transistors become photodiodes and phototransistors. You do not normally see this in action because of EMC shielding or the box that the device comes in.

    1. Anonymous Coward
      Thumb Up

      Re: Advantages of an opaque box

      Ah, right. So if you get a solid black/red/blue case for your Pi rather than a transparent one (which I have on two of mine), that will suffice?

      1. Wzrd1 Silver badge

        Re: Advantages of an opaque box

        "So if you get a solid black/red/blue case for your Pi rather than a transparent one (which I have on two of mine), that will suffice?"

        Why bother replacing before testing? Many things that are clear optically to our eyes are opaque to IR and some even opaque to near IR.

        So, get a flash and snap shots of your packaged device at a distance of 10cm or so. If it crashes, why replace the case when a spot of black epoxy would cure the problem?

  5. Anonymous Coward
    Anonymous Coward

    Does anybody remember eproms?

    It was necessary to stick little metallic stickers over the windows if you didn't want bright lights to cause strange glitches. Basically, any transistor or diode which is exposed to light is a phototransistor or photodiode. In the germanium days, Mullard used to sell the OC71 transistor and the OCP71 phototransistor which was much more expensive. The OCP71 had an opaque blob of gel over the transistor element so you couldn't simply scrape the coating off an OC71. A lab centrifuge made a good OC71->OCP71 converter.

    The EMP from a modern Xenon flash is quite small, and it needs to be because it is very close to a whole lot of electronics that control the flash - not to mention the camera.

    1. Anonymous Coward
      Anonymous Coward

      Re: Does anybody remember eproms?

      Yep, you could buy an eprom write flasher (didn't it use ultraviolet light?) for the ZX Spectrum years ago to flash your programs on so it booted up with it. I could never afford one in them days, so never ever tried one out :(

      Edit: A quick Google:

      http://www.worldofspectrum.org/infoseekid.cgi?id=0008150

      1. Anonymous Coward
        Anonymous Coward

        Re: Does anybody remember eproms?

        We people working on industrial computers had great big high speed eprom erasers. I almost regretted the coming of flash, convenient as it was, as the process of erasing a buggy piece of firmware was so visible and conclusive.

        1. werdsmith Silver badge

          Re: Does anybody remember eproms?

          Eproms indeed. We had a UV Eprom eraser and a Stag eprom programmer with a batch of 10 ZIF sockets, plus a master socket. Drop the master Eprom in, and a load of empty ones in the 10 sockets and it would duplicate for you. There was a lad employed to do the job, and one of his tasks was to clean the adhesive gunk from the sticky labels off the windows using isopropyl alcohol, a nice method of recycling. Brings back memories of RS roller leg straighteners for DIL package ICs and tweaking IC legs straight with bent nose pliers.

          1. Kevin Johnston

            Re: Does anybody remember eproms?

            One of my main tasks as an apprentice was to burn and mark the Proms/Eproms for a system. I learnt the hard way that when you varnish over the white ink used to mark them you should use normal varnish and a fine brush.....Spray varnish tends to coat the legs and form a perfect insulating layer......Oooops

            Was going to Anon but the damage to my credibility was a long time ago so it's too late to be precious.

          2. Anonymous Coward
            Anonymous Coward

            Re: Does anybody remember eproms?

            We never had eprom erasers, we would just let the chips cooking on high-noon sun for 12 hours.

            Slow but steady.

            1. john.w

              Re: Does anybody remember eproms?

              The first digital audio console, a Neve DSP1, was killed by one of the company's directors when he took a flash picture of it during its press launch, very embarrassing! An engineer was blamed for not covering the EPROM windows.

      2. Suricou Raven

        Re: Does anybody remember eproms?

        PROM: Writes once. Often just an EPROM without a window.

        EPROM: The one with the window in, erases on exposure to UV light.

        EEPROM: No window, erases electrically.

        You don't see them so often these days because of the rise of flash really-cheap flash, but EPROM does still have a niche in hobby-electronics because it's a lot easier to interface to than a flash chip. It's a very small niche now though - almost any hobby-application which would once have been tackled by the classic arrangement of EPROM+processor* is now easier to tackle with an arduino or one of the many self-contained microcontrollers. The only place you're likely to find an EPROM these days is when taking apart obsolete hardware - a lot of 386-era mainboards used one for the BIOS.

        You can read the data out of an EPROM with nothing more than a battery, some wires and a voltmeter - and a great deal of patience, at about one byte per minute by hand.

        *RAM optional. Sometimes you only need registers.

    2. This post has been deleted by its author

  6. gregthecanuck
    Angel

    Look at the plus side....

    ... it reboots in a flash!!

    1. Anonymous Coward
      Anonymous Coward

      Re: Look at the plus side....

      The Pi's do. Running mine with Slackware (yea, I keep mentioning it), boots in about 15 seconds.

      Bloody wonderful things!

      1. VinceH
        Coat

        Re: Look at the plus side....

        "Running mine with Slackware (yea, I keep mentioning it), boots in about 15 seconds."

        How long? I'll stick with RISC OS! :p

        1. Graham Dawson Silver badge

          Re: Look at the plus side....

          Custom linux with buildroot, approximately 3 seconds to desktop. Well... xorg. Loading any DE on top of that would double the time at least.

  7. Christian Berger

    Makes sense

    Those cases are likely to be a bit transparent, and semiconductors are known to be light sensitive. The cases are likely a bit transparent as they are meant to be as thin as possible to take up as little space as possible. Blocking light was no consideration as they are meant to be operated in an opaque environment.

    1. Anonymous Coward
      FAIL

      Re: Makes sense

      "The cases are likely a bit transparent as they are meant to be as thin as possible to take up as little space as possible."

      That's nonsense. I have 3 Pi's - one with a pink case and two transparent. The pink case is thicker on the OUTSIDE, but the transparent ones are thinner.

      Remember, this is the TARDIS reversed.... it's bigger on the outside if you have a coloured case.

      1. Anonymous Coward
        Anonymous Coward

        Re: Makes sense - @linick

        I assumed he meant the cases of the semiconductors. Surface mount is indeed made very thin compared to the old DIL cases, because (a) soldering is easier - earlier thick surface mounts had a nasty tendency to get swept up by the solder wave and the mass complicates oven reflow - and (b) the overall circuits need to be thin for modern gadgets. A DIL board wouldn't even fit into a modern 6mm thick phone.

      2. Dr. Mouse

        Re: Makes sense

        As mentioned elsewhere, it is the components which are made like that. Sounds to me like it's bad component choice for a board which is designed for hobbyists, or bad design (if those components are to be used, they need to add some covering to them).

        1. Dave 126 Silver badge

          Re: Makes sense

          > Sounds to me like it's bad component choice for a board which is designed for hobbyists

          Methinks the opposite - what hobbyist doesn't have some Blu-tack or duck tape to hand?

    2. Havin_it

      Re: Makes sense

      >Blocking light was no consideration as they are meant to be operated in an opaque environment.

      How d'you see what the LEDs are doing then?

      IMHO I doubt they "meant" anything either way in this case, it probably just didn't occur to anyone and didn't come to light in testing. I don't remember any guidelines from the RPF regarding requirement for opaque cases; were that the case, the "official" perspex case I got from RS for my B would be extremely wrong-headed, if not legally actionable.

      (Believe it or not, the repeated uses of "case" and "light" were unintentional.)

  8. PeterO
    Happy

    That'll be me then....

    "A user with the handle PeterO said:"

  9. Conundrum1885

    Re. Xenon DEATH FLASH!!!!!

    I've also noticed the same effect on Aurora calculators, ie Poundland el cheapo ones.

    The culprit is the transparent case and chip mounted on the back of the board as is common with such things.

    It does occur to me that as many 'phones use the same construction it could be an interesting way to insert intentional glitches into the running system for the purposes of finding vulnerable sections in the firmware.

Page:

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like