Hi! This is the new Raspberry Pi - just released very recently and a user by the name of Peter Onion discovered something very interesting with this board. Let's take a photo of this lovely little board with a camera with a Zen on photo flash on it. Here we go. Oops, Look what happened.

We have just reset. not only reset our board, but we've actually locked it up. It is no longer working at all. To get it working again, we have to repower the thing.

What's going on? Well, it's actually pretty damn obvious now in this thing was first reported. Of course it's spread like wildfire on all the forums and blogs and everywhere else, including the Eevblog forum. Everyone's going. oh, what's going on this mysterious effect? But hey, anyone who's been in the electronics industry a long time would instantly have seen this like I did and just went.

Ah yeah, that's the photoelectric effect. Something on there is photosensitive - the Xenon flash, no worries, happens all the time. and when I say happens all the time. Well, it's actually not that common, but it's been a very well-known effect for a long time.

That light, as I'll explain in a minute, can affect semiconductors. And normally it's not an issue because semiconductors like the main Broadcom chipset here, for example, and the main Ethernet chipset down in here and all the other little black blobs you can see on there. They're plastic encapsulated or ceramic encapsulated or whatever. And of course, the photons of light can't get through to affect any of the semiconductors inside there, so it obviously wasn't that.

And people were, you know, just did trial and error, looked around and they finally found the culprit. Wow Right down here. And bingo, there's our culprit. They're that tiny little chip.

They're used 16. You can see it in comparison to an Oh 402 capacitor there. It's absolutely tiny. And anyone who with any electronics packaging experience knows that he's a chip scale package.

which I'll talk about in a min. And huh? of course that sucker is going to be photosensitive because it's effectively a beard I flipped on its front side. So what we've got here is what's called a wafer level chip scale Package CSP And what it is is basically a bare semiconductor die with the balls directly on the bottom and you can see the balls under there like this. and this is different to other balls on a regular BGA device.

For example, they are plastic encapsulated or ceramic encapsulated chips. This is not. This is merely a bare die on there like that with just little metal balls on the bottom and flipped over. So essentially the wafer of this chip here is actually exposed on the bottom.

If we actually flip that over, you'd be able to see the circuitry on the back. and this is how the light is able to get in. It's able to sneak in under those balls in there and actually affect the semiconductor junctions in there the PN junctions and hence cause the thing to latch up, do something silly given impulse in the wrong part of the circuit. Whatever is actually functionality-wise is causing this switch mode power supply chip to lock up.

So what's happening here is basic quantum physics. You should have learnt this in Physics 101. Yeah, no doubt. familiar with the Planck relationship relationship.

Energy equals Planck's constant times the frequency now Albert Einstein In 1905, wrote a paper that explains some results of this and how that previously. Of course we thought that light was a wave, but he proposed that light was actually packetized bunch of photons. So when the photons hit a metallic material like this, it actually emits electrons like that because we're based on the Planck formulas. so that that's knowing now as the Planck Einstein formula Planck Einstein relationship.

And not only does it work with just basic metal surfaces like this, you've got energy in these photons they hit and they release any MIT electrons. Well, it's not just like a sheet of metal just sitting there. it's also gonna happen with a semiconductor PN junctions ie. your transistors and your diodes and everything else that modern electronics and everything on this Raspberry Pi board is made with.

It's all going to somehow emit some amount of electrons if any sort of photon if any amount of photon hits it based on that relationship and it's based on the frequency which I'll talk about in a minute and that you can demonstrate this in a vacuum tube. For example, if you've got the cathode down here and the anode up here, photons come in, then electrons leave the plate and you can actually get a current flow around there and the same thing happens in your PN Junction here. if you expose it to light as what we're doing up here with this switch mode converter, then electrons can actually flow around the circuit and that can completely screw up the chip. depending on how much photon energy is actually coming in here and at what part of the circuit, and how the circuit works and all that sort of stuff.

But it can, certainly, and in this case we've demonstrated, it does affect it. Now here's something that's often not well known: Albert Einstein actually won his Nobel Prize in Physics for the photoelectric effect. Exactly what's happening here. He did not win it for the more famous Theory of Relativity because that was still a bit debated by the time it came around.

So he actually won it for this photoelectric effect, which of course was the start of both him and Planck. This was effectively the start of Quantum Fit the theory of quantum physics as we know it today. Now, it's actually not just a weird side effect of affecting chips that are exposed like this chip scale package we've got here. This is principle is fundamental to all the sensors that we have these days.

The camera that you're using the camera I'm shooting this with now is not possible. If it wasn't for the photoelectric effect. Photons actually strike the semiconductor sensor inside the camera and generate a current and that can be measured and turned in to the image that you see. Now, the infrared receiver you got on your TV the remote control.

all that sort of stuff. solar cells, for example. They weren't based on the photoelectric effect and a whole bunch of other stuff, so it really a fundamental physical phenomenon. Fundamental Principle of Physics and Electronics 101.

So that's why something like this. that semiconductor junctions are affected by light is nothing new to electronics people. It's a very well-known phenomenon and we can actually demonstrate exactly what's happening here. I've got a standard five millimeter LED here I think it's a yellow one doesn't matter about the color and I've got it hooked up to my triplet analog multimeter here in current mode so it's 60 micro amps full scale deflection here and I'll get my camera this flash Here we go.

Let's hook it up. let's flush this and see that. see the needle job there that is the photoelectric effect in action. It's converting the photons from this xenon arc flash here into the PN Junction of this LED and it actually generates a current.

Not a huge amount, but you can see that it actually does have an effect. and in a modern IC like we have on this Raspberry Pi it might have lots of high impedance aren't nodes on the inside? Well, that amount of current can be a real big deal. and that's what's causing that chip to latch up. And here's a metal can transistor, which you no doubt used - it's a 2 n double - double - I and I've cut the top off that and you can probably see the dye inside there with the two bonding wires jumping over to it.

that tiny little blob in there. that is the little transistor, die and the wires go over there. I've hooked it up to a breadboard now I've exposed this thing to light. Let's see what happens.

Okay, here we go see that jumped up. it's doing exactly the same thing on that top. PN Junction inside the transistor. and of course all your modern eye sees in here.

they're all just physical PN junctions or transistors. exactly what's inside there. Oh, that's that's nice and sharp. Cutting that thing open.

Oh, and no wonder this thing is locking up when you expose the PN Junction in there to all this photon energy coming from that xenon flash. It's really gonna ruin your day now. I Mentioned before, it's all determined upon the actual frequency here, not necessarily the amplitude. That's why some people on the Forums have been investigating this.

They've been shining like I think like eighteen hundred lumens under the chip and you can't make it. Do it because it's not necessarily about the amplitude, its the frequency. And here is a typical Xenon Arc spectrum in nanometers, the wavelength, and the visible spectrum. Of course, around about 392.

While 780 nanometers or there. Abouts that's the visible spectra. so it's generating. You know, the bulk of the energy within that visible wavelengths.

But look at these massive spikes up here in the near-infrared So it's possibly the near-infrared stuff here that's really giving a kick into the chip and whatever is causing that there. So you know it's not necessarily about just shining light into. it has to be a specific frequency. No electrons will be emitted from this metal surface unless it hits a specific frequency.

And that spectrum from the Xenon arc lamp is why it's able to work in this particular case. And that's why people have had no luck with their phone flashes like this. Look, see, no problem whatsoever because it's just regular ambient light. All of the energy is contained down here.

and I suspect what's happening is up in the near infrared or something like that. That's what we get in the Huge. Not only does it have the frequency, but it has the energy up in that range as well. So I suspect that's what's happening.

And the actual chip used in here. the switch mode power supply converter. It's actually an On semiconductor NCP 63 43 and unfortunately, there's not a public datasheet for it. But yeah, like even if you had the datasheet, you're probably only just guessing what aspect of the circuit is actually latching up and doing that sort of stuff.

But you know it's your regular switch mode a buck power supply converter, so it's a step-down converter. Typical topology, but it does have you know a fair amount of circuitry in there to enable that, so it could be any aspect of that you'd have to go back to on semi Can to themselves and they'd have to do extensive experiments to figure out what's going on here. So if anyone thinks they know, it's just a guess so you might think okay, what happens if we should say a laser? onto here. Well, I've got their little you know, half a million what laser pointer or whatever it is.

Yeah, it's no. Well, lesson 1 milliwatt. So it's just your typical laser pointer here. and well.

I can't get that to do anything regardless of what angle I shooter that or anything like that I Can't get a damned sausage. look at that. So that's rock-solid But it's certainly a possibility because we don't know the exact wavelength that's actually causing this thing. so it will be a specific frequency.

Of course, Amplitude of course plays a role too. You can't, just you know, hit up with bugger-all energy. and just because it's the right frequency, it's going to upset it. No, it has to be a specific energy and a specific frequency in the spectrum in order to generate what's called photocurrent.

And when you actually get a current flow through a PN Junction or circuit or solar cell or whatever it is that's actually called a photon occurrence, it's are converting the photon into a current in your circuit. So how can they fix the problem of this little chip? Well, it's very simple. There's two ways to do it. Either you use another chip that is not a chip scale package ie.

and exposed die. so you use like a BGA part or something like that that'll be just fine. Just like this. BGA part is just fine.

no lights going to sneak under there and get onto that Broadcom um, processor die in there. It's just not going to happen because it is fully encapsulated. It's only on that tiny little beast down in there, the chip scale package. So you can either change the package which probably means that you know a totally different chip, you have to change the layout, the pin out, all that sort of thing perhaps.

Or as is very common with that, chip on board. Co B Technology In the industry which uses bear dies directly on the board. like generally facing up, they'll actually put them facing up and then a little bond wires going over. They have a machine that actually just bonds the wires directly from the chip onto the pad.

and that, you know, really low-cost super low-cost greeting cards, for example, might work. although sort of. you know, throwaway products that cost you know, a cent or something like that for the circuitry. That's how they get them if using chip on board technology and they may encapsulate it rather than the bare die to protect it all and also to shield it from light as well.

They dunk it with an epoch see: a big black epoxy. So you've ever seen a big black blob on a board? That's too pond board technology and they could come along. It's probably their factory that assembles this. probably has that.

They just have a machine which comes over. You know a human usually does that. They just bring it over and goat gunk. You know, like a big syringe type thing just comes over and just Gunks it all in big black goo like that and it sets.

And Bob's your uncle. All right. So let's see if we can probe and capture something here when we actually do this. So I've got my scope probe connected across the 1.2 volt output of U 16.

That's the switch mode power supply under question. Here and there we go. There's our nice 1.2 volts, 500 millivolts per division. Everything's hunky-dory Let's it's all working and trust me, it's on the screen there.

so let's hook it up and flush it. Bingo. captured. Look at that.

We've got some sort of transients actually trigger when it's going back up so it's obviously dropped here. This something happened right back over here. let's have a look. You can see a tiny little impulse there.

That's it's not going up by much and that stuff is zoom right into that. We've got the caption memory to do that. There's really nothing doing there because we've got our big antenna earthly connected up to this. We're not actually probing it properly.

You've got a this is where you've got to be careful. Could be a trophy on players when you try to measure this sort of thing. When you've got this ground lis connected like that, you've got a nice big turn there, which can pick up any electromagnetic pulse generated by that xenon arc flash in there. All the currents flowing can easily couple into that, and you know, cause that sort of spike.

So I don't think that's actually what's causing the thing, what's genuinely there. I Think that's actually being picked up by the probe. You can see that your regulator is obviously dropping out and then it's coming back into regulation like that. But it's so it still works.

And switch. My power supply is recovering. That chip is recovering just fine, but it's the Raspberry Pi processor or whatever else well, is only a processor in there really that is locked up and causing it to do it? So it's not the switch mode power supply controller itself. I Think that's recovering just fine and dandy.

Let's see if we can I see a couple that okay, a Raspberry Pi is running. we're AC coupled this. we're now down at 20 millivolts per division. Let's flash it.

So now I can see that big impulse in that No look at that. No, it's still fine so you can see it's just fine. There it is. Even though a screen is blank, our processors locked up everything else that Switch Mode is.

it's just fine. So it's not entirely the folder, the Switch Mode controller chip that's causing that. but it's certainly something to do with the arm, a process of the Broadcom processor that's not allowing it to gracefully restart. After that, the the big dip we saw in the power arm.

So just to make sure this is actually genuinely the output and not some sort of current induced in the scope probe itself, which I believe that high frequency content there is so scaled this up, we're now on 200 millivolts per division. We can see the dip a lot better there, but this is by just by the shape of it and the recovery. like that, it looks like it is the Switch mode our controller actually doing that, recovering and then ramping up and leveling back out there. But hey, let's just prove that by putting some blue tack over the chip ie.

masking out the light, doing the flush again, see if we can get any trigger. Okay, so I've restarted the thing I put a big blob of blue tack over that that should keep out the light from the sides and around the chip and let's flush it trigger again. Oh no, we still get it. Look at that.

and no. I didn't get the light out enough. Jeez, it's sensitive. Well, it seems like my blue tacks not up to snuff.

I had to put a hell of a lot more on there before I could get it so it's not sensitive. So now if we single-shot trigger, of course, there we go. it's we did get that impulse there. that's interesting, actually.

check it out. So we've got that impulse as I suspected, that is due to the electromagnetic pickup by the coil. That's why it's all high frequency stuff, but we do get a little little bit blip like that positive going up. once again, that's them at a much much lower frequency.

so that's that's rather interesting so there is still a bit of a hiccup in that supply. There's one thing I do want to check and that's it's a reverse side dependency note. So we've got that same thing happening there. So yeah, I think that's yeah, that's no problem whatsoever.

Anyway, we've proved that the huge dip that we saw there was actually the dropping out of the switch mode regulator and then recovering and restarting. so it's pretty well proven. and I saw that somebody actually referred to this as the Mogwai effect ahead? Well, all you youngsters out there who've never seen gremlins you won't know what we're talking about, but it's kind of funny, but not really accurate because it's not sunlight that does is it's a specific high-frequency xenon arc flash that doesn't. Oh well.

Nice term though. And yes, that is awful blue tack I've got I don't know how long I've had that sitting around, but man I'll never get all that off. So there you go I hope you found that interesting. There's nothing unusual happening here at all, but if you haven't seen or heard about the photoelectric effect or even if you had, but you didn't know it applied to our basic electronics and PN junctions and everything else in today's modern electronics, then well, you've learned something new.

Catch you next time you.