EEVblog #901 - Raspberry Pi 3 Photoflash Problem

[EEVblog]

The Raspberry Pi 3 has the same Xenon photoflash reset problem as the RPi2 !
There is also a problem with the Broadcom BCM43438 WiFi & Bluetooth chipset!
The internet connection will lockup when exposed to sufficient Xenon flash light.
Datasheet:
http://www.cypress.com/file/298076/download



MAKE SURE YOU WATCH PART 2:

[mux]

Meh, I wouldn't qualify the flash problem as 'really bad' before, or really at all noteworthy now. It's a fun physics lesson for sure, but there is very little chance of a user actually getting into trouble because of it.

[EEVblog]

there is very little chance of a user actually getting into trouble because of it.

One guy did, the one who originally found it.

[EEVblog]

Meh, I wouldn't qualify the flash problem as 'really bad' before, or really at all noteworthy now.

It's just as noteworthy now as it was back then.

[MK14]

Meh, I wouldn't qualify the flash problem as 'really bad' before, or really at all noteworthy now.

It's just as noteworthy now as it was back then.

When Intel had the Pentium Divide bug, a very long time ago. Initially Intel wanted to just ignore it and carry on. Eventually they were somewhat forced to address the issue, so there was an expensive recall and it got fixed.

Since the Rasberry PI 3, is an extremely good value for money, educational/simple computer system, they are aiming for the $29.99 market. Many computers are $299.99 or even $999.

They may have decided that the issue is not important enough to fix and/or it will be fixed when the new generation eventually comes out and/or they will eventually fix it in the current model.

I bet if you properly/thoroughly investigated the system, you could find at least 10 detectable problems/bugs/issues.

Looking at the typical return fail modes (on this forum, via the chap selling lots of untested ones), the SD card system seemed to be a common problem area. If I remember correctly the WiFi doesn't work, or is problematic and/or the driver is not ready yet.

tl;dr
It has apparently been designed down to a low/fixed (apparently) price point, and they make little/no profit on them (maybe, I'm not sure). So I'm not at all surprised.

Maybe it is on a todo/fixme list, and is too low priority to get fixed yet ?

For the price, I'm very impressed with the quality of them. They seem really nice and usable. I've not had any problems with them myself. They seem very well built.

[MK14]

I hadn't seen the video, with my previous post. Now I've seen a lot of it.

So you have to continuously use the WiFi, have the unit uncased (some people case them), turn it upside down, obtain a rare expensive professional (I'd imagine) powerful flashed camera, then patiently flash it about ten times, holding the flash an unrealistic 1 cm away from the chip. Then you act like this problem is a complete disaster, on a fun $29.99 computer ?
tl;dr
It seems considerably better than the earlier board (3 vs 2), and seems like a very minor problem, which needs fairly complicated and involved techniques, just to reproduce it.

Anyway they are getting there, and seem to be fixing things, which is good.

The top chip being susceptible is still at least a minor issue I guess. But you have to go so close with the flash now, it does not seem too bad.

As regards the WiFi chip, is someone in real life going to uncase it, turn it upside down, and flash it ten times, from 1 cm away, while aiming for a particular component on the PCB and running WiFi heavy software ?

I'm not convinced this is a serious problem now.

[EEVblog]

Since the Rasberry PI 3, is an extremely good value for money, educational/simple computer system, they are aiming for the $29.99 market. Many computers are $299.99 or even $999.
They may have decided that the issue is not important enough to fix and/or it will be fixed when the new generation eventually comes out and/or they will eventually fix it in the current model.

Sure, but that doesn't mean I shouldn't do a video showing that it's still a possibility?

I bet if you properly/thoroughly investigated the system, you could find at least 10 detectable problems/bugs/issues.

No doubt.

[EEVblog]

So you have to continuously use the WiFi

Nope. I  recon it will lock up regardless if you are "using" it.

obtain a rare expensive professional (I'd imagine) powerful flashed camera

Nope, any photoflash camera will do. Modern digital still have them you know.

Then you act like this problem is a complete disaster, on a fun $29.99 computer ?

What's the problem with me making a video showing that it's still possible?
I don't understand all the people complaining about this (see youtube comments)
I'm not making out it's a big issue, in fact I said the opposite that it's a remote possibility.

The top chip being susceptible is still at least a minor issue I guess. But you have to go so close with the flash now, it does not seem too bad.

Correct, and I said precisely that in the video.

I'm not convinced this is a serious problem now.

It's not, and I never said it was.

[Jeroen3]

Next time if you want to check the status of a network link continuously you could use

Code: [Select]

ping 8.8.8.8 And it will ping continuously on *unix. On Windows you need to add -t.
To stop press ctrl+c.

[MK14]

Long post, see above for original

Sorry if I've appeared overly critical!

The "real problem", I think is a psychological/human nature one, which I've got (I admit and think it), and maybe the Youtube commentators have the same issues in some cases.
The thing is, the Rasberry PI is "our baby", our/my favorite "toy" (conceptually speaking, of course, it's not anywhere near being my favorite in practice).

Analogy:
How would you react, if someone made a Youtube video, which seemed to "attacked" Fluke multimeters ?
How would Apple Iphone users, react to a Youtube video, criticizing them ?

So I think I have reacted to a combination of the Youtube video title, and criticism of "our baby".

I don't know if what I am saying makes any kind of sense ?

I mostly agree with your comments in principle, and I'm sorry if I've appeared overly critical of "Your baby", that particular video 

tl;dr
I over-reacted to just the title, and the fact that my very much liked series of very nicely priced (UK) computers were being apparently criticized, possibly unfairly, causing me to produce the pair of annoying to you (maybe), posts.

tl;dr2
Yes, I think I've been somewhat non-impartial, due to the Raspberry PI's success in the UK, and world wide. With me being very prowd that it was designed and made in the UK.

Disclaimer:
I'm trying to interpret my psychological feelings and motivations. Such analysis can be error prone.

My apologies again. I should have taken much more time and care, before criticizing the thread and video.

[chris_leyson]

I think the real engineering issue is whether the RPi resets because of the light getting into photosensitive areas of some chips or the magnetic pulse from a Xenon flash. Try covering the board with something opaque, that will eliminate the light issue. My guess is that it's a magnetic pulse issue.

[EEVblog]

Next time if you want to check the status of a network link continuously you could use

Code: [Select]

ping 8.8.8.8 And it will ping continuously on *unix. On Windows you need to add -t.
To stop press ctrl+c.

Didn't know that, thanks.

[EEVblog]

Analogy:
How would you react, if someone made a Youtube video, which seemed to "attacked" Fluke multimeters ?

I've done one:


And many others for brands and products I'm big fanboy of.

[EEVblog]

I don't know if what I am saying makes any kind of sense ?

Yes it does, it's the fanboy effect, and Youtubers like me are all too aware of it.
I did not expect it in this case though, hence my surprise, as IIRC I got none of this criticism for my RPi2 video on the exact same thing.
I thought it was obvious from the video that it's essentially a "non-issue" for almost everyone, and was mainly of academic issue. And somewhat humorous if they tried to fix it and potentially didn't fix it entirely.

[Monkeh]

I think the real engineering issue is whether the RPi resets because of the light getting into photosensitive areas of some chips or the magnetic pulse from a Xenon flash. Try covering the board with something opaque, that will eliminate the light issue. My guess is that it's a magnetic pulse issue.

With the RPi2 it was most definitely light. It's very likely the same issue now, as although they've changed the package, the silicon should still be exposed underneath.

[chris_leyson]

With the RPi2 it was most definitely light

Undoubtedly it was light on the RPi2 but if you cover the light sensetive area and then try a Xenon flash at the same distance, if it still resets then it's a magnetic coupling problem. di/dt is quite high in a Xenon flash and and it probably deosn't take much flux linkage to induce enough volts into a high impedance track. How is the ground plane arranged, if there is a slot in the ground plane then there will be an induced voltage across the slot for example.

[Kleinstein]

If chips are light sensitive, it's more the near IR that causes problems. UV light is easily absorbed and does not come very far in silicon or most plastics. So it could be light to the still exposed sides.  A magnetic coupling is also possible - the flash is quite a powerful source of EMI.

I don't think this is a real issue - you can cover the board if needed, just be aware of the possible trouble.

[EEVblog]

I think the real engineering issue is whether the RPi resets because of the light getting into photosensitive areas of some chips or the magnetic pulse from a Xenon flash. Try covering the board with something opaque, that will eliminate the light issue. My guess is that it's a magnetic pulse issue.

With the RPi2 it was most definitely light. It's very likely the same issue now, as although they've changed the package, the silicon should still be exposed underneath.

Yes, almost certainly the same issue now that I didn't even think to actually test it. Layout looks identical to the RPi2
I should probably test the WiFi chip better though, but i couldn't get it to fail when the board was flipped over, no matter how close.

[EEVblog]

You've put "AGAIN!" in capitals. My first impression was that you are criticising them for not ensuring the issue was resolved once and for all.

Fair point, that was not my intention, it was meant to be read as as though it's a surprise.

[EEVblog]

If anything this seems like a video that didn't need to be made.

I think you have a distinct lack of imagination.
The original RPi2 problem was very well publiciced, and it would be very easy for someone to see that and ask the question "Does that happen on the RPi3 as well or have they fixed it?"
I know, because I asked that exact same question myself, so I did a quick search and couldn't find the answer, so I did a video on it.

[MK14]

If anything this seems like a video that didn't need to be made.

I think you have a distinct lack of imagination.
The original RPi2 problem was very well publiciced, and it would be very easy for someone to see that and ask the question "Does that happen on the RPi3 as well or have they fixed it?"
I know, because I asked that exact same question myself, so I did a quick search and couldn't find the answer, so I did a video on it.

The video is/was fine, and the concept of revisiting this issue, and examining it is fine.

But I agree (with others), that the title was causing provocation.

Rather than
"EEVblog #901 - Raspberry Pi 3 Photoflash Problem AGAIN"

(I think you already edited the title, partly).

"EEVblog #901 - Raspberry PI 3 Is the Photoflash problem gone now ?"

Or

"EEVblog #901 - Raspberry PI 3 Is the Photoflash problem gone ?"

Then it tempts people to watch it, and does not imply that it hasn't been fixed and/or greatly improved as regards photoflash.

[Maalobs]

Next time if you want to check the status of a network link continuously you could use

Code: [Select]

ping 8.8.8.8 And it will ping continuously on *unix. On Windows you need to add -t.
To stop press ctrl+c.

Didn't know that, thanks.

I had the same thought when I watched the video, why use a buffering video-stream in an attempt to show interruptions in the network?
You don't even need to ping the internet, just ping your office router, and use this commandline:

Code: [Select]

sudo ping -s 8 -A <router ip-address>This will blast your router with pings as fast as possible, and you'll instantly see if there are any interruptions.

As for the butthurt expressed in this thread, the people I know who use the RPi and myself included, do it with the bare boards.
I think it's a very valid question to investigate if the board is susceptible to any form of radiation.
If you're aware of it you can take proper precautions, so of course it should be investigated.
Maybe in Raspberrry Pi 4 someone will make sure that no new bare flipchips are added to the board, thanks to the awareness raised about the issue.

[SeanB]

They look to have reduced the effect on the PSU chip by using a back opaque epoxy coat, likely applied during manufacture to the whole die before scribing and then handled like a regular die, just with the thin back coat. Probably should do the same to all the flip chips there, but as they likely only buy in small amounts doing it for the larger chips would not be cost effective, doing a few wafers of a small chip at a cost of a few hundred dollars for the extra step ( and buying this special lot of tested dies all up front) was likely cheap enough, only adding a cent or so to the final BOM cost, and reducing the likely occurrence again by a large amount.

Not really a worry in most cases, you really should have bare dice in a package to protect from moisture in any case, so a simple solution for the few who want a bare board only is to simply get a COB vendor to use the blob of resin dispenser to coat all the chips on the boards, and sell them as developer versions, or a more moisture resistant version after a selective transparent conformal coat of the rest of the board.

This light effect is a nice edge case though, often not though of in design, and a good reminder that to most electronics the world is a hostile place.

Wonder just how much radiation the board will survive while remaining operational, anybody ( Mike, Fraser) care to place one in an Xray scanner for a while while powered and see if they fail as you ramp up the dose. Do not think anybody will want to go into a working or failed nuclear reactor with one as a control, unless there is a lurker working at a Cyclotron with experiment time to spare.

[G7PSK]

What happens if it is exposed to a professional photo flash set up, there is many times more energy from those. Will it kill the chip permanently and from what distance etc. 

[6581]

I think it was interesting follow up. How about laser pointer?