|
Components and Circuits For discussions about component types, alternatives and availability, circuit configurations and modifications etc. Discussions here should be of a general nature and not about specific sets. |
|
Thread Tools |
18th May 2018, 5:02 pm | #1 |
Dekatron
Join Date: Nov 2006
Location: London, UK.
Posts: 3,496
|
Cosmic rays!
This is from Fuji's care notes for its IGBT modules!
"IGBT modules have random failure mode by primary or secondary cosmic ray in general. This amused me a little. It's unlikely to be an issue for the application I'm working on, unless I hitch a ride outside the earth's atmosphere somehow. Particles man, particles.
__________________
Al |
18th May 2018, 5:17 pm | #2 |
Dekatron
Join Date: Nov 2011
Location: Surrey, UK.
Posts: 4,400
|
Re: Cosmic rays!
That is an interesting caveat! Though perhaps a cynic might venture that this was a new height in disclaimacy- how would you prove that a device failure hadn't been initiated by inadequate screening against extra-terrestrial nasties?
Is this because IGBTs have large areas and are therefore statistically more likely to take a hit, or that something else about construction or their mode of operation makes them vulnerable? I'm sure there will be a few here conversant with semiconductor device hardening against this sort of thing. |
18th May 2018, 6:23 pm | #3 | |
Dekatron
Join Date: Nov 2006
Location: London, UK.
Posts: 3,496
|
Re: Cosmic rays!
Quote:
It's an interesting one. It seems genuine, as they do invite customers to consult them for high-altitude applications. On the other hand, it's no. 17 out of 18 care notes. Maybe a technical writer was feeling a bit bored and wanted to make a joke that was masked by a borderline plausible scenario? On the other hand, I have no idea how these highly energetic particles affect power semiconductors, perhaps in surveillance planes or equipment hurtling through space...
__________________
Al |
|
18th May 2018, 6:28 pm | #4 |
Dekatron
Join Date: Jun 2009
Location: Staffordshire Moorlands, UK.
Posts: 5,276
|
Re: Cosmic rays!
I suspect it's an aeronautical thing, Fuji are one of the major IGBT manufacturers in the world and no doubt the things get everywhere, certainly in airliners and satellites
__________________
Kevin |
18th May 2018, 6:43 pm | #5 |
Dekatron
Join Date: Apr 2011
Location: Cornwall, UK.
Posts: 13,454
|
Re: Cosmic rays!
Passengers.
Lawrence. |
18th May 2018, 7:02 pm | #6 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,903
|
Re: Cosmic rays!
Yes passengers get higher doses of cosmic radiation. The higher they go, the greater. Concorde flights increased dosage rate but decreased exposure time. The overall effect was greater risk.
Electronics packages in high altitude jet have a significantly greater exposure equipment and crew spend appreciable time high and cruising., some balloons go higher and stay up longer, and spacecraft are the extreme case. Modern semiconductors with small feature sizes are more vulnerable than older big and clunky devices. IGBTs and power mosfets are cellular in structure being vast arrays of small devices in parallel. They are vulnerable to an extent. The smaller features in dense CPUs and memory are worse. High altitude stuff often gets a shield, or located somewhere in the craft that is shielded by structures that also have other purposes. Some of my designs go rather high, and it's amazing what you have to do. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
18th May 2018, 8:03 pm | #7 |
Hexode
Join Date: Oct 2014
Location: Wiltshire, UK.
Posts: 382
|
Re: Cosmic rays!
Somewhere in the late70s we had Cray1 serial no 1 supercomputer at the lab where I worked. It had an unheard of 4Mbyte of semiconductor memory which for cost reasons had no error detection circuitry. It's mean time to memory error was around 30 minutes, caused mostly by cosmic ray interactions. Subsequent Cray computers had ECC added but we're still prone to cosmic ray hits.
Roger |
18th May 2018, 9:20 pm | #8 |
Dekatron
Join Date: Jan 2012
Location: Brentwood, Essex, UK.
Posts: 5,349
|
Re: Cosmic rays!
We used to get the American magazine "Electronics" circulated at work and I remember some time in the late 1970's reading a review of I think that year's International Solid State Circuits Conference where the problem of cosmic rays affecting memories was discussed. Memory structures had got so small by then that a stray cosmic ray passing through Silicon could generate several hundred free electrons, sufficient to flip the state of a cell. Interesting to read of Roger's practical experiences with this problem and that it is still ongoing.
Last edited by emeritus; 18th May 2018 at 9:50 pm. |
18th May 2018, 9:41 pm | #9 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,903
|
Re: Cosmic rays!
Astronauts report occasionally seeing light when a cosmic ray photon passes through an eye. Makes me feel squeamish!
David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
18th May 2018, 10:01 pm | #10 |
Nonode
Join Date: Dec 2012
Location: Kirk Michael, Isle of Man
Posts: 2,350
|
Re: Cosmic rays!
Over 50 years ago I read about cosmic rays striking the retina and how you could see the flashes. We had a "wardrobe" at work, originally a big, built in cabinet, and with the door closed it was pitch black. I took a chair, went in, closed door and waited for my eyes to settle down. After about 2 mins, a definite flash, with regular (random) flashes following. I forget exactly, but probably more than one per minute.
When I was told years later about how micoms could be switched by cosmic rays, it was easy to believe. My Grundig Satelit 600 has twice "screwed itself" in the 30 plus years I have had it. Remove back up battery and back to normal. This has (rightly or wrongly) been attributed to cosmic ray assault. Les. |
18th May 2018, 11:01 pm | #11 |
Dekatron
Join Date: Jun 2016
Location: London, UK.
Posts: 3,687
|
Re: Cosmic rays!
This is a failure mode we have to deal with in the IT sector. One reason we have ECC memory. Cosmic rays genuinely do flip bits. As DRAM cells get smaller, this becomes a big problem.
Also when you look at the size of machines. Typically there is one cosmic-ray-induced error per 256 megabytes of RAM per month (according to IBM). Now our mid-range servers have 128-256Gb of RAM and we have 32 of them. I’ll let you do the mathematics there but it’s a massive risk for us. If a program crashes because a pointer is corrupted, meh, but if a bit changes half way through a math op, big trouble. As we carry financial data, this could be ruinous. Fortunately in practice the actual problems are rare and the software surface so large that the data has a low probability of actually being corrupted. I have only seen one inexplicable corruption event and that was on a non ECC machine at an insurer. It flipped an FP bit and raised someone’s risk profile considerably. Because the input and outputs were logged we could reproduce it by fuzzing memory bits and could attribute it to one bit changing. Took three days and six people to find that problem. Also look at SpaceX who use multiple redundant bits of consumer grade hardware in their space craft. They have 2-3 systems comparing results. This turns out to be considerably cheaper than one silicon on sapphire radiation hardened CPU. Last edited by MrBungle; 18th May 2018 at 11:09 pm. |
18th May 2018, 11:35 pm | #12 |
Dekatron
Join Date: Nov 2010
Location: Oxford, UK.
Posts: 4,999
|
Re: Cosmic rays!
Yup. Redundancy is important in space missions, for that sort of reason. An instrument I was project manager for a little while back is due for launch this October to Mercury. And for the science of looking at the X-ray spectrum of the elements that make up the regolith, it needs an active sun (ie lots of sunspots). And that means that it produces not only the X-rays we needed for the instrument measurements, but just about every high energy particle going.
That meant that every darned electronic component has to be qualified for radiation hardness, and that bumped up the price by several orders of magnitude. As an example, in your mobile phone there is an FPGA, which costs pennies. The space qualified, radiation hard gull-wing package FPGA's we used were £30k each. Surface mount resistors were £50 each. I forget what capacitors cost - but it was equally silly money. Now sending astronauts up into space is a big risk, particularly if a coronal mass ejection is heading your way. Apart from all the other risks, going to Mars and back will involve at best a damaging radiation dose, and possibly a lethal one if they get hit with a CME. |
19th May 2018, 2:04 am | #13 |
Heptode
Join Date: Jan 2004
Location: Stockport, Greater Manchester, UK.
Posts: 827
|
Re: Cosmic rays!
When I worked for a CCTV camera manufacturer over a decade ago, we found the statement by the Japanese manufacturer of the CCD sensors about them being more prone to damage by cosmic rays when sent by air rather than by ship to be all too true.
This also applied when supplying finished products to an American customer. who always required quick delivery, which was unfortunate to put it mildly... The most severe damage to CCD sensors we ever saw was to some earlier Philips monochrome ones in our cameras used by a nucleonics (or similar) lab. and subjected to X-rays. The other problem, only affecting colour CCDs, was burning of the colour mosaic, giving burnt-in pink patches where pointed at bright lights for a long time. Not helped by installers spending good money for auto-iris lenses but setting the camera to "electronic iris", instead of "DD" (direct drive to the lens iris) and setting it up properly. |
19th May 2018, 5:19 am | #14 |
Nonode
Join Date: Jan 2010
Location: North Somerset, UK.
Posts: 2,130
|
Re: Cosmic rays!
I have heard, but can not substantiate, that cameras fitted to robots in high radiation environments have a very short life.
At the scene of a nuclear reactor accident, ISTR that the pictures relayed from a robot could be viewed in real time and degraded noticeably in a short time. Around the world there are a number of high voltage DC power transmission schemes. These make use of high voltage high power semi-conductors in the converters. I wonder if anyone has collected data on failure rates in different environments, such as high and low altitude operation, or high versus low natural background radiation places. The quality of the components, the degree of cooling, and how hard they are pushed are probably of greater significance. But perhaps there are schemes where these factors are identical, but where one end of the link is at sea level and one end up a mountain, for example. |
19th May 2018, 9:35 am | #15 |
Octode
Join Date: Feb 2006
Location: Bristol, UK.
Posts: 1,042
|
Re: Cosmic rays!
I used to work with clinical linacs which produce 6Mv x-rays for cancer treatment. The bunkers were monitored with ccd dome cameras. Usually after a few years the image had a lot of 'sparklies' caused by pixel damage to the ccd by the x rays. These dead pixels could be black white or any colour and when the beam was on they twinkled like stars. Vert pretty but not much use. Luckily the cameras were used for monitoring the patients and not high resolution images so the carried on using them.
Malcolm |
19th May 2018, 10:22 am | #16 |
Rest in Peace
Join Date: Sep 2008
Location: Solihull, West Midlands, UK.
Posts: 4,872
|
Re: Cosmic rays!
We have a cheap 'trail' camera which we sometimes leave in the back garden overnight to see what the local foxes are up to. A few images have streaks across them which I have blamed on cosmic rays.
|
19th May 2018, 10:57 am | #17 | |
Dekatron
Join Date: Sep 2007
Location: Oxfordshire, UK.
Posts: 4,311
|
Re: Cosmic rays!
Quote:
Cheers, GJ
__________________
http://www.ampregen.com |
|
19th May 2018, 11:31 am | #18 |
Dekatron
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,088
|
Re: Cosmic rays!
There's also Cerenkov radiation from a very fast particle - charged or not. If it is moving at near the speed of light and it passes through your eye, then in the fluids in your eyeball (where refractive index is <1) then it will emit radiation as it slows down.
As for cosmic ray damage to IGBT's - these are a small cellular structure and are inherently NPNP configuration, but specified to be operating below latching current. If one of these gets turned 'on', then the thing will go bang very quickly! |
19th May 2018, 11:48 am | #19 |
Dekatron
Join Date: Apr 2012
Location: Wiltshire, UK.
Posts: 14,010
|
Re: Cosmic rays!
I remember many years ago being given a colleague's design for a relatively simple bit of electronics. This was late-1970s - the very early days of 'microprocessors' and this particular design had progressed through several technical iterations before anyone bothered with anything as seemingly trivial as how much the thing was going to cost to manufacture.
Then they asked me to look at the costings, which resulted in a minor panic. The 'chip' the designer had specified cost two orders-of-magnitude more than the entire components-and-manufacture cost for the rest of the thing! Seems the designer had specified the "Space/Defense-grade" SoS radiation-hardened available-to-special-order-only version of the RCA COSMAC (1802) chip by mistake. Oops! A quick call to the 'states got us a 'normal consumer chip' price that was much more in-line with expectations. |
19th May 2018, 12:05 pm | #20 | |
Dekatron
Join Date: Sep 2007
Location: Oxfordshire, UK.
Posts: 4,311
|
Re: Cosmic rays!
Quote:
Cheers, GJ
__________________
http://www.ampregen.com |
|