The quartz is used to make a very precise frequency in an oscillator, in this case it was for a clock. This is the episode where that particular quartz fails, just about here: th-cam.com/video/CeXYZKg5jdo/w-d-xo.html . And here is the episode where I explain how quartz crystals oscillators work, also timed at about the right place: th-cam.com/video/ad8azNovsF8/w-d-xo.html
Great video! Reminds me of when I used to teach digital techniques class when I was in the USAF at Keesler AFB. I would bring in some slides of chips I had opened and photographed through a microscope. It was one of the most fun classes I taught. USAF 77-83 ECM. After I got out of the USAF, intel hired me. Some litho corrections on giant Nikons were in microns. We also had to adjust laser interferometers, focus and alignment (overlay), One layer lines up on the layer below it. In the early days, operators would align the chip layers manually, looking through scopes. Later, computers would correct everything, (mag, overlay, run out, ) on the fly.. as the wafer stage moved and the reticle image was enlarged through a slit. (Perkin Elmers). Nikons used rotating x and y leadscrews were soon replaced by mag levs. Wafer stage positions were calculated by bouncing modulated laser beams off of long mirrors attached to each stage. One for X, One for Y. Some litho tools made exposures through a liquid, not air. (No bubbles allowed!) retired after 28.5 years. I still have nightmares!!
The crystal was dropped or otherwise subjected to high g forces. Believe it or not, they can survive the initial hit but shatter on the impact after the rebound. The pin sheared first then the edge was chipped on impact with the casing. The best way to open them is to saw them just above the seam or weld line of the base, then the top lifts off cleanly and you don’t risk debris contamination. I liked the ballet anyway. MIL Spec crystals are tested on a drop jig in different orientations and bounced off a designated cushion or anvil and caught on the rebound. This imparts a predicable haversine acceleration for qualifying the device class. This is done by 100% sampling. Civilian crystals are prepared the same way but not necessarily tested 100%. Had the electrical connection survived the crystal would have worked but those chips would have impaired performance, shifted frequency, tendency for spurious response along the curve in your vector traces and different “activity”, the term for motional resistance.
I guess that maybe the entire board was dropped. As the quartz crystal is mounted on springs it might hit the casing, bounce back and hit another wall of the casing. So you may be able to deduce the approx. direction of the fall. Big shatter and impact markings might have hit first. Have a look at the pcb for strike marks. But I'm a programmer, not an engineer.
Previous job we started having failures. Traced to crystal. SCEM examination revealed tin whiskers. The supplier had switched to RoHS compliant without changing the part number.
Back in 1973, I spent 6 months on an industrial placement, developing an environmental radiation monitor to run off a battery and sit in various locations dotted around the CERN site. This used a number of 4000 series cmos chips for their low power needs. However, the circuitry required an op amp. We used a 4069 inverter as an amplifier with a feedback resistor from output to input so the output quiescent level was 0.5 x Vcc. So both transistors were on at the same time.
I love this channel, I'm a retired navy electronic tech, I was very good at troubleshooting and repairing but I was always fascinated by the engineering aspect of electronics. I learn about so many things from watching your videos, keep it up :)
*I had a communications shop long ago. With crystal controlled CB's and Marine Radios we would remove the little pink rock from its looped whisker-wire holder and burnish it with tooth paste on a flat surface thus moving working freqs in between channels. Most popular.*
Motorola uses the 14xxx series for CMOS because they already had a 4xxx series for one of their old TTL lines (MTTLsomething). That old line fell to the standard 74xx series long ago, but two numbers survived for many years - MC4024 and MC4044.
It really made me feel old to see the "vintage components" were a crystal and CMOS! I still have TTL and CMOS in my component collection, along with transistors, silicon and germanium diodes and a few valves (vacuum tubes). I worked in electronics during the 1970s and 1980s.
I used to work at AMD in the early 1980's. I have many unmounted chips that I collected. I can't say what the chips we're used for or what their ID number is, but would you want them? I have no reason to keep them any longer. Some chips are uncut wafers and some are cut in their holding container. Let me know and I'll send them to you if wanted.
Please contact ne through the link in the video description (my channel about page). The persons in the team interested in such dies are Antoine and Ken.
This is genuinely one of the most fascinating videos I've seen in a very long time. I studied computer engineering and I've seen explanations here and there of wafer level geometry, but it was never explained with the same amount of detail and context. Thank you
Takes me back to Geology at school... "Quartz instead of having neat cleavage planes like most other minerals, exhibits conchoidal fracture"... and you've got a perfect example on that thar failed crystal.
Dooblydoo is generally attributed to WheezyWaiter and popularized by the vlogbrothers and PBS idea channel. AvE is certainly not a follower though, he's brought lots of original flare to the english language. Also I don't think dooblydoo needs to be capitalized but now I'm just being pedantic.
Marc is channelling AvE today! Nicely done... I use to repair broadcast FM Lav microphone transmitters, that were crystal-controlled PLL. The operators took great pride in somehow smashing the crystals in those transmitters, as we had to replace them constantly... and they aren't cheap, and take months of lead time to grow and manufacture! We use to have to order the crystals in bulk, because of the six-month manufacturing lead time...
May I make a suggestion about decapping ceramic chips? Hold the chisel with the angled face down. When the chisel starts to move after the cap breaks free, it will rise up and avoid the chip and bond wires.
why would anyone have this expertise? Everyone I know throws faulty ones away without a second thought, gets a new one. I'm not criticizing you at all, I'm genuinely curious.
Had a ULA design done back in the early 80s. Initial production was shipped in ceramic packages with the tops misaligned just enough that when loaded into standard IC tubes that allowed a bit of movement many would arrive with the tops sheared off.
If you think the chip failed due to ionic contamination simply bake the chip overnight at ~250C and see if there is any recovery. The ionic contamination is usually Na+ which has high mobility in SiO2 at elevated temperatures. If the failure is due to ESD there will be no recovery and most often ESD will take out one input rather than the whole chip failing. Latchup induced failure will usually cause high current flow so you will often see visual evidence even at the magnifications you were looking at the chip with. Some curve tracer work will also sort out these failures. Enjoy your videos, good stuff, keep them coming! de AI6XG
Well what am I going to do now? I’ve watched all your videos. Absolutely love watching your stuff. Vintage tech always amazes me and while I could never do what you do I wish I had the time, money and knowledge to. Living vicariously though your videos.
7:06 - I'd add that there is no shoot-through current, ever. The gate threshold voltages of the transistors are chosen for the open-circuit regions to overlap, so the only currents that flow are those supplying the gate charge and overcoming the Miller's effect.
You can open ceramic components easily if you start by making a slight cut or scratch with a cutter right under the lid. This way the glass frit sealing will break at the scratch and you can save the wires unbroken. This will allow you to test each inverter separately and guess wich part of the die is failing.
Recently had two MC14516 counters fail in HP3580A spectrum analyzers, both chips had date codes from the mid 1970's. I suspect these had similar type of failure as the functional failures and the manufacturers were different, also in these cases ESD is an unlikely cause.
Wow, this brings back memories ! I used to be employed in a classified position in the British government trying to protect communications in the early seventies. When I first joined we were using low power T.T.L. then people got rather exited to see that stores on site had a few CMOS chips available. Those were soon snapped up and people soon saw the possibilities these devices would offer. I got my hands on a few and made a couple of shift registers to test bit encryption algorithms. Taking things a little to far, I pushed the clock speed too far and was very surprised to see the current rise rapidly and the chips radiate like little angry bees. It took me a little while to realize my mistake. The clock speed had turned the P-Channel and the N-Channel transistors almost directly across the positive and ground, creating a short. I was young, it wasn’t my last electronic folly !
All chips from that era have their metal layer (aluminum) with a bubbly texture to it. In the 1990s they started polishing the wafers after each step to enable the addition of a large number of metal layers. In addition they switched from aluminum to copper. The visible (top) metal layer looks very smooth on those chips. Completely unrelated but visually similar, solder masks for PCBs from the same time frame also have a bubbly texture if they went through a wave soldering machine.
undervolt it and cool it well, don't OC it, that will increase the lifetime dramatically. even at stock voltages, CPUs last very VERY long, usually outlasting their motherboards. Even ~20 year old Pentium III / IV CPUs still work great today, even if used heavily. it's only the really old CPUs (30y+) like the motorola 68K series that start to fail often these days. Sad for collectors, but a CPU lasts FAR longer than its usable lifespan (which might be 10-15 years today - I would say core 2 duo are the oldest CPUs that are still somewhat usable for normal desktop tasks).
@@KingJellyfishII depends on the other components, especially motherboard and PSU, temperature and how good you take care of it. But yes, 20years can easily be achieved.
My ancient mid-70's Signetics 2650 8-bit CPU still works fine in the project that I put it in, in 1976. So do all the other chips, such as the 2102 RAMs and all the logic. The Motorola MC3410 DAC still works as well, and so does the 2708 EEPROM that holds the program. That's 44 years now.
How did the chip fail electrically? There are many ways to isolate an electrical failure. PEM (Photon Emission Microscopy) or OBIRCH (Optical Beam induced Resistance CHange) would work well on this chip, which likely has some type of curve trace anomaly.
The slightly poor layer alignment maybe was why the IC failed and the others did not, it's no secret that better made IC's last long, but back then it could also have been impurities, commerical level production will always not be as good as the absolute best we can make.
Those early CMOS ICs are a nightmare in vintage synthesizers, especially analog ones. Replaced a lot of them in my synths. Together with op-amps and capacitors, they're a common failure point.
5:50 So that means, any story in which some form of a Computer or Computer A.I. is still around after hundreds of years is not realistic. Which would mean that if we truly succeed in making Human-like Robots/Androids there lifespan would be limited by the life of their chips, be it CPU, RAM or Memory ones. After which amount of time do that electro migration set in? I am just curious how long a CPU can last.
You can actually trade that off as an engineering parameter. It gets worse at smaller geometries and higher current densities, and I believe it is therefore become a relatively recent dominant limiting factor (like leakage). I think it affects all high density / high power / small geometry modern chips. I don't remember what the number was (this is probably a confidential number anyhow) but there is a minimum accepted industry target for consumer chips, something like more than 25 years under extreme temperature and voltage condition - once again I am not a reliability specialist and I don't know exactly what it is, so I might be quite off. It's also a statistical thing, they won't all die once they reach this age, it's usually spec'ed as max number failures after a certain (very large) number of hours (it's known as FIT), which is usually kept as a relatively small number. Regrettably all electronic components fail with age, their FIT is never zero. You just get it down to an acceptable range for your application. Military and industrial chips are made to much higher standards and lower FIT, and aerospace chips to even greater standards. Which is part of the reason why they use such antiquated processes and large geometries (along with the better radiation resistance of the larger geometries).
@@CuriousMarc Yeah, so as example like in that 2002 movie "The Time Machine" where the hero travels from the 19th century to the 21st century where he finds some Hologram AI that he talks with, only to travel like 800,000 years forward in time... where he later meets the exact same Hologram AI which even still remembers him. Something like that would be impossible, even if we had a means to always have electricity. Some Computer chips that are permanently under current, extreme voltage condition after 800,000 still working is impossible. I would say even working for 100 years is at the edge then. The sad part for Space Travel is, because of the radiation and temperature conditions, making chips with a too small production size makes it harder. Like the 7nm process most AMD CPUs are made under. Somewhere here on youtube I seen someone talk about chips that are soon on atomic level created. I wonder for how long those would work. But yeah, most stuff made today, would break and be nearly gone in 100 years without any human around.
@@Momoka7 That’s basically the story of this channel. Most of what we deal with is 40 to 50 year old, and it’s rare that anything still works. And then after we repair it, and it fails again a few years later ;-)
Is it possible that the chip failed due to CMOS latchup? My understanding is that this can be precipitated by cosmic rays and the like in susceptible CMOS devices. Would such a failure be visible on the die?
Can we decap and read the contents of current day flash memories without relying on the controller or any electrical input, using the technologies we have today? I know it was done with the NES bios ROM, but I guess that was a much simpler memory chip ...
Since many of today's flash memory devices, especially the higher capacity ones, use Multi-Level Cells, i think it would be very difficult to determine the contents (or rather, state) of each level below the topmost one. The NES ROM you mention is probably a mask-programmed ROM, which, as you say, is a much simpler technology.
Interesting if the quartz crystal failed due to high g shock or just mechanical degradation? Or maybe some factory defects was primary reason? Also in some old books from the tube era I know that it was very easy to overdrive the crystal so it can crack. But of course low power cmos ic can't do that.
I think the CMOS IC chip is most likely damaged by an Overvoltage rather than by electromigration, because 1) the size of the trace is way larger than in modern CPU, so how could it be less resilient than a modern CPU to electromigration, which lasted a decade. 2) Overvoltage damage is when a transparent oxide layer is shorted, so there's no way to see it with optical microscope, like in this instance, so I think it is Overvoltage damage.
I am no specialist, but my mundane understanding of it is that it happens when ions, metals or other materials used to make the transistors move or migrate from their original position due to the action of the electric field in the circuit while it works.
So what exactly does the quartz do? I missed the first video plus I don't know much about this stuff.
The quartz is used to make a very precise frequency in an oscillator, in this case it was for a clock. This is the episode where that particular quartz fails, just about here: th-cam.com/video/CeXYZKg5jdo/w-d-xo.html . And here is the episode where I explain how quartz crystals oscillators work, also timed at about the right place: th-cam.com/video/ad8azNovsF8/w-d-xo.html
@@CuriousMarc thanks!
Great video! Reminds me of when I used to teach digital techniques class when I was in the USAF at Keesler AFB. I would bring in some slides of chips I had opened and photographed through a microscope. It was one of the most fun classes I taught. USAF 77-83 ECM. After I got out of the USAF, intel hired me. Some litho corrections on giant Nikons were in microns. We also had to adjust laser interferometers, focus and alignment (overlay), One layer lines up on the layer below it. In the early days, operators would align the chip layers manually, looking through scopes. Later, computers would correct everything, (mag, overlay, run out, ) on the fly.. as the wafer stage moved and the reticle image was enlarged through a slit. (Perkin Elmers). Nikons used rotating x and y leadscrews were soon replaced by mag levs. Wafer stage positions were calculated by bouncing modulated laser beams off of long mirrors attached to each stage. One for X, One for Y. Some litho tools made exposures through a liquid, not air. (No bubbles allowed!) retired after 28.5 years. I still have nightmares!!
The crystal was dropped or otherwise subjected to high g forces. Believe it or not, they can survive the initial hit but shatter on the impact after the rebound. The pin sheared first then the edge was chipped on impact with the casing. The best way to open them is to saw them just above the seam or weld line of the base, then the top lifts off cleanly and you don’t risk debris contamination. I liked the ballet anyway. MIL Spec crystals are tested on a drop jig in different orientations and bounced off a designated cushion or anvil and caught on the rebound. This imparts a predicable haversine acceleration for qualifying the device class. This is done by 100% sampling. Civilian crystals are prepared the same way but not necessarily tested 100%. Had the electrical connection survived the crystal would have worked but those chips would have impaired performance, shifted frequency, tendency for spurious response along the curve in your vector traces and different “activity”, the term for motional resistance.
Excellent! Musta taken a wrong turn at Albuquerque
I guess that maybe the entire board was dropped. As the quartz crystal is mounted on springs it might hit the casing, bounce back and hit another wall of the casing. So you may be able to deduce the approx. direction of the fall. Big shatter and impact markings might have hit first. Have a look at the pcb for strike marks.
But I'm a programmer, not an engineer.
Thanks
Previous job we started having failures. Traced to crystal. SCEM examination revealed tin whiskers. The supplier had switched to RoHS compliant without changing the part number.
Or made in China..
Back in 1973, I spent 6 months on an industrial placement, developing an environmental radiation monitor to run off a battery and sit in various locations dotted around the CERN site. This used a number of 4000 series cmos chips for their low power needs. However, the circuitry required an op amp. We used a 4069 inverter as an amplifier with a feedback resistor from output to input so the output quiescent level was 0.5 x Vcc. So both transistors were on at the same time.
You can see this method was also used for the quartz oscillator in our Russian clock.
Nice!
I have only ever tried that trick in SPICE :)
I love this channel, I'm a retired navy electronic tech, I was very good at troubleshooting and repairing but I was always fascinated by the engineering aspect of electronics. I learn about so many things from watching your videos, keep it up :)
*I had a communications shop long ago. With crystal controlled CB's and Marine Radios we would remove the little pink rock from its looped whisker-wire holder and burnish it with tooth paste on a flat surface thus moving working freqs in between channels. Most popular.*
This was without a doubt the best opening demonstration of an IC I've seen so far. Thank you very much.
Work with lot of planning and precision. Very good.
Motorola uses the 14xxx series for CMOS because they already had a 4xxx series for one of their old TTL lines (MTTLsomething). That old line fell to the standard 74xx series long ago, but two numbers survived for many years - MC4024 and MC4044.
The CMOS IC microscope view is so cool!
It turns what seemed like magic into something clearly visible.
It really made me feel old to see the "vintage components" were a crystal and CMOS! I still have TTL and CMOS in my component collection, along with transistors, silicon and germanium diodes and a few valves (vacuum tubes). I worked in electronics during the 1970s and 1980s.
His blog is insane. Casually reads silicon like a book and own the IR library
I used to work at AMD in the early 1980's. I have many unmounted chips that I collected. I can't say what the chips we're used for or what their ID number is, but would you want them? I have no reason to keep them any longer. Some chips are uncut wafers and some are cut in their holding container. Let me know and I'll send them to you if wanted.
I’d love to see some of those wafers!
Send them to zeptobars!
Please contact ne through the link in the video description (my channel about page). The persons in the team interested in such dies are Antoine and Ken.
No don't send them. Ken is just looking for raw material like that to build his robot army. Haven't we had enough problem in 2020 already?
@@BobWiersema 🤣🤣🤣🤣🤣
This is genuinely one of the most fascinating videos I've seen in a very long time. I studied computer engineering and I've seen explanations here and there of wafer level geometry, but it was never explained with the same amount of detail and context. Thank you
Takes me back to Geology at school... "Quartz instead of having neat cleavage planes like most other minerals, exhibits conchoidal fracture"... and you've got a perfect example on that thar failed crystal.
Very illustrative.
Yes, as a flintknapper I saw that, too...
Mentioning the Dooblydoo, and setting classical music to milling action. You are a true AvE fan :)
I was looking for a comment like this. My thoughts too. :-D
SAME!!
I thought the same thing too haha
Dooblydoo is generally attributed to WheezyWaiter and popularized by the vlogbrothers and PBS idea channel. AvE is certainly not a follower though, he's brought lots of original flare to the english language. Also I don't think dooblydoo needs to be capitalized but now I'm just being pedantic.
Not just ANY Classical music but The Blue Danube Walz... a perennial AvE favourite.
Marc is channelling AvE today! Nicely done...
I use to repair broadcast FM Lav microphone transmitters, that were crystal-controlled PLL. The operators took great pride in somehow smashing the crystals in those transmitters, as we had to replace them constantly... and they aren't cheap, and take months of lead time to grow and manufacture!
We use to have to order the crystals in bulk, because of the six-month manufacturing lead time...
May I make a suggestion about decapping ceramic chips? Hold the chisel with the angled face down. When the chisel starts to move after the cap breaks free, it will rise up and avoid the chip and bond wires.
why would anyone have this expertise? Everyone I know throws faulty ones away without a second thought, gets a new one. I'm not criticizing you at all, I'm genuinely curious.
Amazing, that die reversing was simple enough I almost understood it completely!
Excellent! No paperclips were harmed in the making of this video.
Ha, some 8bit guy reference? That was so cringing to watch!
Nonononono don't do iiit aaaaand spack
A failed crystal is a rarity, and about the last thing you expect. In 45 years of engineering, I’ve found 2 or 3, maybe.
That was my first one! Clearly this one was dropped pretty hard.
Had a ULA design done back in the early 80s. Initial production was shipped in ceramic packages with the tops misaligned just enough that when loaded into standard IC tubes that allowed a bit of movement many would arrive with the tops sheared off.
It is my first time to see someone revers engineer a silicon chip and it was fascinating! Thank you for this video!
If you think the chip failed due to ionic contamination simply bake the chip overnight at ~250C and see if there is any recovery. The ionic contamination is usually Na+ which has high mobility in SiO2 at elevated temperatures. If the failure is due to ESD there will be no recovery and most often ESD will take out one input rather than the whole chip failing. Latchup induced failure will usually cause high current flow so you will often see visual evidence even at the magnifications you were looking at the chip with. Some curve tracer work will also sort out these failures. Enjoy your videos, good stuff, keep them coming! de AI6XG
The Hex Inverter.......a simple electronic method to counteract curses.
The magical form of "I am rubber, you are glue."
They’re from the machinists screaming “more RPM and why are you climb cutting?!!"
@@beefchicken more speed yes but cut thick to thin for the win.
Oh boy, more adventures in electronic archeology!
Man, Ken is extremely skilled at reversing ICs
Ken's dry sense of humour gets me every time!
Well what am I going to do now? I’ve watched all your videos. Absolutely love watching your stuff. Vintage tech always amazes me and while I could never do what you do I wish I had the time, money and knowledge to. Living vicariously though your videos.
I can hear AvE crying tears of joy at your precise, delicate milling, and your choice of music. It seems Chickadee isn't his only protégé 🙂🙂
The Blue Danube makes everything instantly ten times more elegant and relaxing! ◡̈
It does!
For a minute I thought I was watching an AvE video.
I was thinking the PanAm spaceliner approaching the space station in 2001, A Space Odyssey
@@neilshep50 Yup
I'm looking at ken's blog, it's amazing!
Wonderful video, you explain everything elegantly
Makes me wish I had taken computer class more seriously in high school. Love the tear down. Very interesting how these chips and crystals work.
7:06 - I'd add that there is no shoot-through current, ever. The gate threshold voltages of the transistors are chosen for the open-circuit regions to overlap, so the only currents that flow are those supplying the gate charge and overcoming the Miller's effect.
You can open ceramic components easily if you start by making a slight cut or scratch with a cutter right under the lid. This way the glass frit sealing will break at the scratch and you can save the wires unbroken. This will allow you to test each inverter separately and guess wich part of the die is failing.
Recently had two MC14516 counters fail in HP3580A spectrum analyzers, both chips had date codes from the mid 1970's. I suspect these had similar type of failure as the functional failures and the manufacturers were different, also in these cases ESD is an unlikely cause.
Space Odyssey theme while using that end mill on the crystal. Priceless!
The classical music overdub of the endmilling. You like AvE I see.
Richard Smith what is ave?
@@bigbaddms A skookum channel.
No bukkake machine. I'm disappointed.
@@BlackEpyon XD Bukkake
@@ManofCulture Why that coolant is WHITE, I can not fathom.
Wow, this brings back memories ! I used to be employed in a classified position in the British government trying to protect communications in the early seventies. When I first joined we were using low power T.T.L. then people got rather exited to see that stores on site had a few CMOS chips available. Those were soon snapped up and people soon saw the possibilities these devices would offer.
I got my hands on a few and made a couple of shift registers to test bit encryption algorithms.
Taking things a little to far, I pushed the clock speed too far and was very surprised to see the current rise rapidly and the chips radiate like little angry bees.
It took me a little while to realize my mistake. The clock speed had turned the P-Channel and the N-Channel transistors almost directly across the positive and ground, creating a short.
I was young, it wasn’t my last electronic folly !
Another amazing video Marc.
Nerdy! Thoroughly enjoyed it. I'm definitely milling my decap bin from now on!
When the end comes around and they ask "what did you do with your life," I'm gonna be in trouble.
Marc, thank you for the video. It made my day. Such a great repair stories spiced with good good information.
"link in the Doodley Doo" 🤣🤣🤣
I cant believe i get too see this stuff on youtube for free :D
That was a fantastic explanation by you and Ken. I am now completely convinced that electronic circuits are High Magic Indeed. :D
Can you please make a video explaining how chips are designed? All I see here are lines!
You should go to Ken’s blog at righto.com . He has many articles where he explains how chips are designed.
Amazing on how small even for a 1968 chippy. Bravo
Man after just watching Applied Sciences video on ultrasonic soldering, I'd love to see if it would work on repairing the broken quartz.
Or reconnecting the 4069 die :)
Why hasn't TH-cam's algorithms recommended you to me before? Brilliant work!
because yt's algorithms are a joke!
Fantastic video as always.
The hex inverter chip looks like it has measles. Do all chips look like that or is it a kind if corrosion?
All chips from that era have their metal layer (aluminum) with a bubbly texture to it. In the 1990s they started polishing the wafers after each step to enable the addition of a large number of metal layers. In addition they switched from aluminum to copper. The visible (top) metal layer looks very smooth on those chips.
Completely unrelated but visually similar, solder masks for PCBs from the same time frame also have a bubbly texture if they went through a wave soldering machine.
@@jecelassumpcaojr890 Thank you for your explanation.
The decapping musical sequence is so oddly relaxing
Nice- now I am anxious about my CPU dying of electromigration...
Same bro
undervolt it and cool it well, don't OC it, that will increase the lifetime dramatically.
even at stock voltages, CPUs last very VERY long, usually outlasting their motherboards.
Even ~20 year old Pentium III / IV CPUs still work great today, even if used heavily.
it's only the really old CPUs (30y+) like the motorola 68K series that start to fail often these days.
Sad for collectors, but a CPU lasts FAR longer than its usable lifespan (which might be 10-15 years today - I would say core 2 duo are the oldest CPUs that are still somewhat usable for normal desktop tasks).
@@TheRailroad99 ok so my computer will theoretically last for a max of 20 years ish?
@@KingJellyfishII depends on the other components, especially motherboard and PSU, temperature and how good you take care of it.
But yes, 20years can easily be achieved.
My ancient mid-70's Signetics 2650 8-bit CPU still works fine in the project that I put it in, in 1976. So do all the other chips, such as the 2102 RAMs and all the logic. The Motorola MC3410 DAC still works as well, and so does the 2708 EEPROM that holds the program. That's 44 years now.
How did the chip fail electrically? There are many ways to isolate an electrical failure. PEM (Photon Emission Microscopy) or OBIRCH (Optical Beam induced Resistance CHange) would work well on this chip, which likely has some type of curve trace anomaly.
I'm looking forward to watching curiousmarc repair the remaining failed clock modules
Very Interesting Content! Can you tell me what kind of digital(?) microscope you use? Those images look really good.
2:50 in my childhood we used a sandpaper for such delicate operations :-)
The slightly poor layer alignment maybe was why the IC failed and the others did not, it's no secret that better made IC's last long, but back then it could also have been impurities, commerical level production will always not be as good as the absolute best we can make.
Those early CMOS ICs are a nightmare in vintage synthesizers, especially analog ones. Replaced a lot of them in my synths. Together with op-amps and capacitors, they're a common failure point.
"Even your Intel CPU will one day die to electro-migration"
P A N I K
*Remembers I have a Ryzen CPU*
K A L M
Still the same
That was the joke.
@@michaelallison2836 Maybe electro-optical CPUs could solve the issue of electromigration in the future.
You have brought art into technical learning…🎶🎵
Thanks for doing this, really interesting!
5:50 So that means, any story in which some form of a Computer or Computer A.I. is still around after hundreds of years is not realistic. Which would mean that if we truly succeed in making Human-like Robots/Androids there lifespan would be limited by the life of their chips, be it CPU, RAM or Memory ones. After which amount of time do that electro migration set in? I am just curious how long a CPU can last.
You can actually trade that off as an engineering parameter. It gets worse at smaller geometries and higher current densities, and I believe it is therefore become a relatively recent dominant limiting factor (like leakage). I think it affects all high density / high power / small geometry modern chips. I don't remember what the number was (this is probably a confidential number anyhow) but there is a minimum accepted industry target for consumer chips, something like more than 25 years under extreme temperature and voltage condition - once again I am not a reliability specialist and I don't know exactly what it is, so I might be quite off. It's also a statistical thing, they won't all die once they reach this
age, it's usually spec'ed as max number failures after a certain (very large) number of hours (it's known as FIT), which is usually kept as a relatively small number. Regrettably all electronic components fail with age, their FIT is never zero. You just get it down to an acceptable range for your application. Military and industrial chips are made to much higher standards and lower FIT, and aerospace chips to even greater standards. Which is part of the reason why they use such antiquated processes and large geometries (along with the better radiation resistance of the larger geometries).
@@CuriousMarc Yeah, so as example like in that 2002 movie "The Time Machine" where the hero travels from the 19th century to the 21st century where he finds some Hologram AI that he talks with, only to travel like 800,000 years forward in time... where he later meets the exact same Hologram AI which even still remembers him. Something like that would be impossible, even if we had a means to always have electricity.
Some Computer chips that are permanently under current, extreme voltage condition after 800,000 still working is impossible. I would say even working for 100 years is at the edge then.
The sad part for Space Travel is, because of the radiation and temperature conditions, making chips with a too small production size makes it harder. Like the 7nm process most AMD CPUs are made under. Somewhere here on youtube I seen someone talk about chips that are soon on atomic level created. I wonder for how long those would work.
But yeah, most stuff made today, would break and be nearly gone in 100 years without any human around.
@@Momoka7 That’s basically the story of this channel. Most of what we deal with is 40 to 50 year old, and it’s rare that anything still works. And then after we repair it, and it fails again a few years later ;-)
Is it possible that the chip failed due to CMOS latchup? My understanding is that this can be precipitated by cosmic rays and the like in susceptible CMOS devices. Would such a failure be visible on the die?
CuriousMarc, what make/model of video microscope do you use? thanks.
Very interesting indeed. We forget sometimes these things have a tough life before they get to us.
Very interested in the tools you used to open such delicate components
What software do you use to create those schematics?
My old Atari 800XL suffered the same inverter gate chip failure which led to issues in the display.
It seems like electrostatic discharge break. 5:53, bottom right corner, white mark between traces.
hello, could You tell me what a model of microscope Your're using?
Was the card made or designed in Australia ? The schematic shows the map.
which kind of microscope are you using?
Nice choice of music for the milling operation. Was that the Deutsche Grammophon recording of the Berlin Philharmonic?
Love your music choices ☺️
Awesome video 😁
Can we decap and read the contents of current day flash memories without relying on the controller or any electrical input, using the technologies we have today?
I know it was done with the NES bios ROM, but I guess that was a much simpler memory chip ...
Since many of today's flash memory devices, especially the higher capacity ones, use Multi-Level Cells, i think it would be very difficult to determine the contents (or rather, state) of each level below the topmost one.
The NES ROM you mention is probably a mask-programmed ROM, which, as you say, is a much simpler technology.
Marc, Did all 6 sections die or just one?
I did not check. It’s a bit late to do it now...
@@CuriousMarc perhaps on another one sometime - thanks
Wow! Amazing video! I mean the drilling part together with music!
2:08 the drill is dancing to the music !!
Waltzing
Fantastic, nice job.
Great work!
Interesting, Marc!
Blue Danube plays in the background while sloooooowly removing cap
I was kinda curious to look inside it!!
thought I was watching an AvE video for a minute there with the doobly do and classical music during milling procdeures!
Very very interesting stuff.
Very informative video, thank you !
Do you have an IBM 5100?
Sadly no.
So I'm curious what could cause that mechanical failure on the crystal oscillator.. did someone just drop it?
It was likely donated by Linus from Linus tech tips...
Do you watch the Green Brothers?
I always got this Question How long does Intel CPU life mine is 11 years old and Still working fine
Very very nice video!
Which kind of microscope are you using?
thank you for the knowledge
where did the fragments go?
Never found them. I suppose they were sucked out by the vacuum cleaner maybe?
@@CuriousMarc I dont know, id have thought they would still be in the little case? strange!
Interesting if the quartz crystal failed due to high g shock or just mechanical degradation? Or maybe some factory defects was primary reason? Also in some old books from the tube era I know that it was very easy to overdrive the crystal so it can crack. But of course low power cmos ic can't do that.
I think the CMOS IC chip is most likely damaged by an Overvoltage rather than by electromigration, because 1) the size of the trace is way larger than in modern CPU, so how could it be less resilient than a modern CPU to electromigration, which lasted a decade. 2) Overvoltage damage is when a transparent oxide layer is shorted, so there's no way to see it with optical microscope, like in this instance, so I think it is Overvoltage damage.
Not electro migration on these early ICs. Ionic contamination or migration.
What is electro migration?
I am no specialist, but my mundane understanding of it is that it happens when ions, metals or other materials used to make the transistors move or migrate from their original position due to the action of the electric field in the circuit while it works.