are heat sinks on retro gear a meme?

are heat sinks on retro gear a meme?

  1. 4 weeks ago
    Anonymous

    Nothing pre-6th gen needs heatsinks.

    • 3 weeks ago
      Anonymous

      Almost.
      Apple Pippin has heats and a fan.
      Nintendo 64 had a quite large heatsinks.

    • 3 weeks ago
      Anonymous

      the Intellivision definitely had hot running chips that used triple voltages and they were sinked from the factory

      • 3 weeks ago
        Anonymous

        >that fucking trace routing
        That's actually painful to look at, like looking at the back-side of a PDP-10's boards.

        • 3 weeks ago
          Anonymous

          what's wrong with it? everything has proper thicknesses and it probably has fewer reflections than 45 degree autism, not that it even mattered

      • 3 weeks ago
        Anonymous

        >that fucking trace routing
        That's actually painful to look at, like looking at the back-side of a PDP-10's boards.

        [...]
        >dem wiggly woobly traces

        For what purpose?

        fucking zoomers, it's beautiful

        • 2 weeks ago
          Anonymous

          >board covered in cum
          >beautiful
          okay

          • 2 weeks ago
            Anonymous

            Your world is dark anon. It is an unsightly way to secure a heatsink though

      • 3 weeks ago
        Anonymous

        Those chips are using the old-timey enhancement mode NMOS process like the 8080 used and that needed three power lines. it was already obsolete by the late 70s.

      • 2 weeks ago
        Anonymous

        Actually the RF shield that sat over the chips helped trap heat in. That can be removed, though.

        • 2 weeks ago
          Anonymous

          the Intellivision definitely had hot running chips that used triple voltages and they were sinked from the factory

          from what I read on AtariAge Intellivisions don't seem that unreliable. apparently the PAL version had some overheating issues that NTSC units didn't have.

          • 2 weeks ago
            Anonymous

            I'm pretty sure they're the only console pre-N64 where all the main ICs are sinked.

          • 2 weeks ago
            Anonymous

            the Intellivision definitely had hot running chips that used triple voltages and they were sinked from the factory

            Commodore should have actually done this instead of using the RF shield as an ineffective sink.

            • 2 weeks ago
              Anonymous

              my dad bought a C64 in 1983 and had to return three of them before he got a working unit

              • 2 weeks ago
                Anonymous

                The PLAs had an awful failure rate in the beginning.

        • 2 weeks ago
          Anonymous

          RF shielding requirement on US electronics in that time were really stupid. you open up a ZX Spectrum and there's absolutely no RF shielding in there, and nobody was ever harmed by it that i'm aware of.

      • 2 weeks ago
        Anonymous

        Wow. the SRAM (also handles bus control) in there is only rated for 40C operating temperature. The CPU is rated for up to 70C.

    • 2 weeks ago
      Anonymous

      You've never even touched a Commodore 64

    • 2 weeks ago
      Anonymous

      In C64 the PLA and VIC-II are the biggest offenders for heat generation. It won't harm you to sink other components, certainly, but those are definitely the hottest-running chips.

      • 2 weeks ago
        Anonymous

        >This is a really well-made board, and the chips...National Semiconductor made the POKEY. Much more reliable than MOS and their flaky, flaky chips.

        he had several C64 repair videos and in one comment section guy said "i got tired of replacing dead SIDs in my C64s so I just got a SwinSID and called it a day." the analog filters in those things are so, so fragile and are responsible for probably 80% of SID malfunctions.

  2. 4 weeks ago
    Anonymous

    Their benefits are theoretical. Dissipating heat should help prolong the life of any part, but if those parts were prone to cooking out, it probably would have happened by now.

  3. 4 weeks ago
    Anonymous

    It's a meme. Don't do that shit.

  4. 4 weeks ago
    Anonymous

    The late great Bob Pease of National Semiconductor, who was sadly killed in a 2009 car accident, didn't seem to think so. He said most ICs never get near a temperature high enough to cause internal failure. I read on a forum somewhere a guy who said he'd seen a chip die tested and still work at 660F (!) although the die would likely melt after at most an hour of that kind of extreme heat. The main materials that dies are made of, silicon and aluminum (copper in modern high performance applications) can tolerate temperatures way beyond what would be encountered in normal operating conditions.

    • 3 weeks ago
      Anonymous

      idk on AtariAge they were pretty adamant that a TI VDP chip would melt (literally) if not sinked.

    • 3 weeks ago
      Anonymous

      Chips eventually fail. Temperature plays a role in how long they last. So even if a chip can work at 660F, its lifespan is going to be lower than if it was running at a normal temperature. Also, heating and cooling cycles are also a thing, the hotter something gets, the more stress.

      • 2 weeks ago
        Anonymous

        Doesn't really matter how many he's had. No PC in the last 25 years hasn't come with a heat sink that's screwed or clipped down unless it's made by a turbo tard using cheap shit.

        Ha. How many have you had? Happens all the time. If it's not screwed down it will come off in time.

        Those chips are using the old-timey enhancement mode NMOS process like the 8080 used and that needed three power lines. it was already obsolete by the late 70s.

        lol at these zoomers quoting unsourced Wikipedia pages

        • 2 weeks ago
          Anonymous

          Could be a subreddit he's pasting from. It is still better to run stuff not too hot; a chip's data sheet will state its maximum operating temperature and for the 5 um NMOS chips Commodore used this was about 70C. In typical operating use a VIC-II will probably reach about 40C but a sink will knock several degrees off of that. Keep in mind that 5 um was already an outdated process node by the early 80s and most chips had moved to smaller sizes by then. The C64C has 2 um chips that only get a little warm when running and are significantly more reliable. Only the 8580 SID still gets a bit hot as it retains a separate 9V line to operate its analog filters.

      • 2 weeks ago
        Anonymous

        >Chips eventually fail.
        there is electronics stuff like radios from the 60s which is still working

        • 2 weeks ago
          Anonymous

          1960s electronics are not made using microscopic transistors.

        • 2 weeks ago
          Anonymous

          These apples sure don't taste like oranges.

        • 2 weeks ago
          Anonymous

          Well anyway a minor repair and this Fairchild Channel F from the 70s is in working condition. Chances are it would be a very long time.

          • 2 weeks ago
            Anonymous

            and for what brakes there's usually equivalent modern replacement parts and custom ICs can be reproduced in FPGA

            • 2 weeks ago
              Anonymous

              You could always get a batch of new ones from China or something, for a price though they would be more modern CMOS chips as nobody does NMOS anymore to my knowledge.

          • 2 weeks ago
            Anonymous

            I have never heard of an IC dying simply because it celebrated a certain number of birthdays. It was usually always an external factor like a bad PSU, overheating, or static electricity. Sometimes the chip was made improperly, but those usually die in infancy.

            • 2 weeks ago
              Anonymous

              My tv box had random crashes for years, and last month the CPU just failed catastrophically. It suddenly threw a bsod for cpu cores out of sync. Boot to BIOS and core is running 87 degrees C, bloody hot. Then hang and no more boot, sadly no fireworks or even a little smoke. Just a vaguely hot electronicy smell. Forage replacement i3 from IT store dump -> good as new with no more random hangs.

              • 2 weeks ago
                Anonymous

                TV/router/cable boxes get notoriously hot. I've had a few fail like that.

  5. 4 weeks ago
    Anonymous

    C64Cs are notably more reliable than the breadbin C64 and this is attributed to the more modern HMOS chips that run at lower temperatures.

    • 3 weeks ago
      Anonymous

      C128 is also more reliable because it's mostly HMOS chips (earlier production ones still have 6581 SID though).

      • 3 weeks ago
        Anonymous

        After three years of attempts getting the HMOS process to work because they couldn't figure out how to dope the chip dies properly. the initial batches of the VDU chip in the C128 were coming off the line 70% defective.

    • 3 weeks ago
      Anonymous

      The C64C also had a better ventilated case and it doesn't trap as much heat.

    • 3 weeks ago
      Anonymous

      The TED chip is notoriously unreliable because they used that faulty HMOS process. But at the same time, they seem to hold up better in C16s than the Plus/4 which is believed due to the latter having a terrible thermal environment with almost no case ventilation (also C16 has fewer components to generate heat)

    • 3 weeks ago
      Anonymous

      One thing that was probably bad for breadbins (at least NTSC units) was running all day in hot summer weather. This would have been less of an issue for PAL C64s as European summers aren't as hot.

  6. 3 weeks ago
    Anonymous

    id be more worried about corrosion and failing caps / power supplies +shorts

    the only time i replace thermal shit is when it is damaged by me
    vintage arcade boards wont undergo that much stress realistically (compared to running 15 hours straight 7 days a week in an arcade)

    next

    • 3 weeks ago
      Anonymous

      >id be more worried about corrosion and failing caps
      or the combo of corrosion from leaking caps. I've had two copies of Starfox die on me for that reason, with traces completely rotten all around the faulty cap.

      The old hardware don't need heatsinks. Last night I was playing an old Hong Kong special 115 in 1 NES cart and it wouldn't stop glitching. Took the cart out and it was insanely hot so I opened it up only to find three cold solder joints. Reflowed the joints and not it boots up at the first try and doesn't even get warm. I guess what I'm saying is if it's old it's not supposed to get hot and if it does, installiing a heatsink is only a band-aid fix.

      • 3 weeks ago
        Anonymous

        https://forums.atariage.com/topic/167645-atari-400-shows-random-characters-on-startup/

        He believed his ANTIC shit itself and swapped it for another one, but I wonder if it wasn't some other issue like a cold solder joint and the chip wasn't actually at fault especially because the problem manifested itself after it had been running for a while. The chips in A8s are generally quite reliable and don't get very hot as they're strictly 5V (excluding the RAM in the 400/800 which uses a 12V line).

        • 3 weeks ago
          Anonymous

          it's a PAL unit. didn't Atari use substandard components and reject bin parts in a lot of PAL machines because they didn't really care about the PAL market?

      • 3 weeks ago
        Anonymous

        >The old hardware don't need heatsinks

        the Intellivision definitely had hot running chips that used triple voltages and they were sinked from the factory

        the TI VDP was heat sinked and the data sheet for the chip explicitly required it; the thing got hot.

        Or did it?

        • 2 weeks ago
          Anonymous

          Hardware design was, for the most part, far more competent back then than it is now.
          If something needed cooling, it would have cooling.

          • 2 weeks ago
            Anonymous

            ha ha no. i worked for IBM back when they still had their consumer PC division. if they could save 10 cents from omitting a heat sink then they would omit that heat sink. it's only in mainframe/mini or hardware designed for the military or something where cost is no object.

      • 3 weeks ago
        Anonymous

        It's not really about exceedingly high temperatures so much as the temp cycling causing flex/warp/breakage issues. That said Commodore themselves apparently decided the VIC PLA and SID needed some kind of cooling as the RF shield in most revisions had bent down tabs meant to serve as an admittedly shitty and largely useless heatsink, and those are the three heatsinked in OPs pic

        • 3 weeks ago
          Anonymous

          >so much as the temp cycling causing flex/warp/breakage issues.
          there's an old urban myth that thermal cycling would cause the bonding wires in a chip to fail but like anon said, Bob Pease dismissed that as bullshit. that might have conceivably been an issue in the 70s during the very early days of the IC industry.

        • 3 weeks ago
          Anonymous

          the first VIC-IIs from 1982 had a ceramic shell for thermal reasons but this had to be switched to plastic to reduce costs.

        • 3 weeks ago
          Anonymous

          We know the PLA got hot because they were made with the early faulty version of the HMOS process that Commodore attempted. Those chips on paper should have run cooler than the other components made with the old-style NMOS process.

        • 3 weeks ago
          Anonymous

          they should have put actual heat spreaders on the chips ala

          the Intellivision definitely had hot running chips that used triple voltages and they were sinked from the factory

          • 3 weeks ago
            Anonymous

            Probably, although the anons saying the C64 is much more susceptible to voltage/power issues nuking the various chips are definitely correct. The original PSUs are literal timebombs. When I picked one up out of a thrift shop back in 2018 I ran it just long enough to confirm it could boot and that the 1541 was functional, and even that was a potential risk. If nothing else with these a modern replacement PSU is essential

            • 3 weeks ago
              Anonymous

              In fact a bad PSU will blow up the RAM before it harms any other components in the computer. The RAM chips have small, dense dies and are very easily damaged by overvoltage while the main chips like VIC-II et al have relatively large, coarse dies that can take a fair bit of abuse. So the RAM is basically like a crumple zone in a car frame; it will get toasted first and provide a warning of imminent PSU failure.

              If you happen to find a C64 with bad RAM it's more often than not a PSU fault and an indication that you'd better throw that PSU away at once (occasionally you'll come across a machine with Micron RAM chips which do a fine job of self-destructing with no help needed).

        • 2 weeks ago
          Anonymous

          >temp cycling causing flex/warp/breakage issues
          The Apple III was the most notorious example. That computer had no cooling whatsoever (at Steve Jobs' insistence). Warping from heat was so bad that boards would unseat themselves, and a common fix was to pick the computer up and drop it from a height of about 6 inches to re-seat everything.

      • 3 weeks ago
        Anonymous

        >Last night I was playing an old Hong Kong special 115 in 1 NES cart and it wouldn't stop glitching. Took the cart out and it was insanely hot so I opened it up only to find three cold solder joints
        The cartridge obviously had CMOS chips and the broken lines caused a latch-up condition. If that happens in a CMOS chip it will drive the signal lines high which causes that thermal condition your cartridge experienced. Were they NMOS the signal lines would go low.

        • 3 weeks ago
          Anonymous

          ah, ok

        • 3 weeks ago
          Anonymous

          ah, ok

          is it not possible that the bad solder joints were increasing resistance in the connection which increased excess heat from wasted energy

        • 3 weeks ago
          Anonymous

          it's usually not good to leave lines on a CMOS chip floating

  7. 3 weeks ago
    Anonymous

    >are heat sinks on retro gear a meme?
    fucking lol

  8. 3 weeks ago
    Anonymous

    Heat sinks are NEVER a meme.
    Take care of your hardware.

    • 3 weeks ago
      Anonymous

      >are heat sinks on retro gear a meme?
      yes

      > machine lasts 30-40 years
      > hasn't failed
      > see photos on facebook posted by numale with low testosterone
      > OMG! I DIDN'T KNOW I WAS AT RISK OF COOKING THE CHIPS!
      the internet has fostered retards and baboons educating fellow retards and baboons. it's one large circle jerk of pure retardation, and nobody knows a fucking thing about how any of this works.

      C64Cs are notably more reliable than the breadbin C64 and this is attributed to the more modern HMOS chips that run at lower temperatures.

      facts.

      • 3 weeks ago
        Anonymous

        Baboons are pretty cool and shill bros.

      • 2 weeks ago
        Anonymous

        I think you should find a new hobby.

  9. 3 weeks ago
    Anonymous

    Even if it only helps prolong their lifespan by 1% I'd say it's cool if you care about your hardware. If anything they're fun to install.

  10. 3 weeks ago
    Anonymous

    the TI VDP was heat sinked and the data sheet for the chip explicitly required it; the thing got hot.

    • 3 weeks ago
      Anonymous

      the VDP would have been ok if the chip package was wider which would allow more thermal dissipation. later revisions used a newer process and didn't need a heat sink.

      • 2 weeks ago
        Anonymous

        I assume the original VDP was 5 um and they did a die shrink later. the versions used in MSX weren't sinked.

    • 3 weeks ago
      Anonymous

      the Intellivision definitely had hot running chips that used triple voltages and they were sinked from the factory

      >dem wiggly woobly traces

      For what purpose?

      • 3 weeks ago
        Anonymous

        Wave soldering.

        • 3 weeks ago
          Anonymous

          ah ok

  11. 3 weeks ago
    Anonymous

    on most hardware I'd say yes, but not these piece of shit Commodores. they are prone to chip failure and with as many bad chips as there are out there, the only modern solutions are FPGAs which cost a fortune compared to what should be a cheap replacement chip

  12. 3 weeks ago
    Anonymous

    there has been some debate over the cause of the high failure rate in early production SNESes (specifically SHVC-CPU-01s) and one plausible theory is that the US model SNES has poor case ventilation compared to the SFC and traps more heat

    • 3 weeks ago
      Anonymous

      I have no idea but the first Famicoms definitely had a thermal problem and would lock up so Nintendo recalled them and the first two PPU revisions were heat sinked.

      • 3 weeks ago
        Anonymous

        it's not clear exactly why that was. either the initial PPUs had faulty dies ala TED or they used a larger process node that got shrunk later.

    • 3 weeks ago
      Anonymous

      The SNES is all CMOS, incidentally, there's no NMOS parts in any version of it so the chips should barely get warm when running. It's a lot more likely that the first run of chips were manufactured improperly than there being a thermal issue.

  13. 3 weeks ago
    Anonymous

    Intellivisions were known back in the day to have overheating problems; on AtariAge there were boomers claiming the damn things would lock up after half an hour of use and you'd have to turn them off and let them cool back down for a while.

  14. 3 weeks ago
    Anonymous

    the chips in C64 are rated for a maximum operating temperature of 70C. while modern chips can take much higher temps than that, it's still way hotter than they'd ever get in normal use.

    • 2 weeks ago
      Anonymous

      Data sheet for a modern SRAM. This can take as much as 255F.

      • 2 weeks ago
        Anonymous

        To be fair it says these are "user guidelines, not tested", and presumably for a chip series rather than a final product. The specs really do go beyond office computer requirements but these chips might also be used in automotive or industrial (and clearly military) applications that could actually approach these temperatures.

  15. 3 weeks ago
    Anonymous

    It was recommended around the time Pentiums came around that a heatsink was recommended. I have a Pentium Overdrive processor that came with one attached with a default fan. Some electronics has protections enabled where it would shut off if the temperatures reach past a certain threshold to protect the chip itself.

    I am still trying to find ways to improve the cooling and ways to attach more robust heatsink/fans to a Voodoo 5500 and the safe way to remove the legacy adhesive.

    • 3 weeks ago
      Anonymous

      I believe 486s were the first x86 CPUs to need a sink. Yet the 68040 ran very cool and never needed it. I guess it was a more efficient design.

      • 3 weeks ago
        Anonymous

        Things that run at a slower clock, have less transistors and don't have to manage more complex instructions do tend to not produce as much heat yeah.

        • 3 weeks ago
          Anonymous

          the 486 and 68040 both had about the same transistor count IIRC

          • 3 weeks ago
            Anonymous

            I just read up on this and was totally mistaken. They are near identical chips. I mixed up the two other texas instruments cpu's in my head. My bad.

        • 3 weeks ago
          Anonymous

          the 486 and 68040 both had about the same transistor count IIRC

          The 68040 wasn't weighted down with 16-bit legacy modes.

      • 3 weeks ago
        Anonymous

        I was not sure which came first at the time. There were some advanced 486 processors, the DX4-100 I had was definitely after Pentiums were available. I just know for sure Pentiums highly suggested the use of heatsinks.

        >Voodoo 5500
        muh financial difficulties please understand

        Yeah, I happened to have kept mine from back then. That's what was so special about 3Dfx: they were affordable solutions to 3D acceleration available to the public. The 5500 was falling behind the alternatives, it was still a good all-in-one solution without daisy chaining graphics cards. I'd like to maintain this childhood tech since it is a key part in a JWin98SE time machine. I read that people add heatsinks to the back of the card, but I also wonder how to improve the heatsinks and fans on the front of the card too.

      • 3 weeks ago
        Anonymous

        Dx40s were usually fan-less and even some dx2-66s. I wouldn't do it but these things can take some punishment(especially the ones rated for higher temperatures).

    • 3 weeks ago
      Anonymous

      >Voodoo 5500
      muh financial difficulties please understand

  16. 3 weeks ago
    Anonymous

    heatsinks mayber
    RF shielding isn't

  17. 3 weeks ago
    Anonymous

    Intel has made some inefficient chips. The P4 especially; Dell used to underclock those in their PCs to keep operating temperatures at a safe level.

  18. 3 weeks ago
    Anonymous

    If the chip feels hot to the touch it probably should.

  19. 3 weeks ago
    Anonymous

    I lost two PCs to overheating years ago. One (a Compaq Pentium 4) I remember it would either power on to garbage characters or sometimes the BIOS boot screen but then stop there. It had a number of blown electrolytic caps near the back of the motherboard. I couldn't tell you what the specific point of failure was though or if replacing those caps would have fixed anything.

    • 3 weeks ago
      Anonymous

      Probably just needed reapplication of thermal paste
      Could’ve also been the PSU

  20. 3 weeks ago
    Anonymous

    The original revision TMS9918 was an NMOS chip and the later ones that were used in MSX and SG-1000 used the more modern HMOS process that ran cooler so you didn't need a heat sink anymore.

    • 3 weeks ago
      Anonymous

      just fwiw I recently rebuilt a ZX-81 (the Spectrum's predecessor) by replacing the NMOS Z80 with a new CMOS one, the 1k SRAMs with a single new 16k chip, and the ULA with an FPGA replacement. it was quite impressive how much I cut down on the heat and power consumption. some measuring confirmed that the new components draw about 30% less power than original vintage chips did.

      • 3 weeks ago
        Anonymous

        Not sure exactly, but the first generation Z80s from the late 70s (like what the TRS-80 Model I used) were NMOS and could get quite toasty. I imagine by the time the ZX Spectrum came out in '82 they'd improved them a lot.

    • 3 weeks ago
      Anonymous

      it was probably an early HMOS type. true NMOS would use multiple power rails like the VIC-II or the chips in the Intellivision. i imagine the later sink-less VDP used HMOS II.

      C64Cs are notably more reliable than the breadbin C64 and this is attributed to the more modern HMOS chips that run at lower temperatures.

      MOS had been using the 5 um HMOS I for the 6502 since 1975 with VIC-II and SID using the original multi-voltage NMOS process (SID was 6 um while VIC-II used 5 um). when they tried to introduce the 3 um HMOS II process was where they got into trouble and kept fucking it up.

      • 3 weeks ago
        Anonymous

        in fact the very earliest 6502s from 1975-76 were 8 um. over time MOS managed to shrink this some. the VIC-II as anon said was 5 um but also had substantially more transistors than a 6502. by the time C64 was in development, they had two main production lines running. the first was the NMOS line that couldn't do anything smaller than 5 um. there was another line that did 3 um HMOS. the large process nodes that VIC-II and SID used were outdated by the time Reagan was president but they were there and there was no point in not using them as the production line was there and the VIC-II and SID wouldn't benefit from die shrink as much as a CPU or RAM. also MOS had years of experience with the older large node NMOS process and the initial production run of VIC-II/SID in 1982 got something like 80% yields. by the time they'd moved to the plastic-shelled R2 VIC-II in early '83 it would have been better than 95% yields.

        during this time Commodore was trying to go down to 2 um for CPUs and other such components. they had trouble getting it to work which resulted in the notoriously unreliable PLA and TED. a lot of these were faulty; the PLA had such low yields that they intentionally shipped C64s with defective ones to meet shipping quotas with the idea that they would be returned and swapped out for known good chips.

        yes those large node processes were outdated, but since Commodore only made chips for in-house use and didn't have to be concerned with consumer demand as Intel and Motorola did, they didn't necessarily need to be aggressive about adopting the latest tech.

        • 3 weeks ago
          Anonymous

          VIC II had 13,000 transistors. TED had something like 20,000 because it has to do all graphics functions plus RAM refresh plus I/O plus sound.

        • 3 weeks ago
          Anonymous

          >during this time Commodore was trying to go down to 2 um for CPUs and other such components. they had trouble getting it to work which resulted in the notoriously unreliable PLA and TED. a lot of these were faulty; the PLA had such low yields that they intentionally shipped C64s with defective ones to meet shipping quotas with the idea that they would be returned and swapped out for known good chips.
          Bil Herd claimed they had difficulty doping the wafers properly and applied too much dopant for the die size--this resulted in excessive conductivity and was visible on the die as "creeping crud" (his words). For contrast here's a section of an 8086 die. The transistors are dark purple from the dopant coating. What presumably happened with those TED chips was excess dopant oozing out onto the substrate.

          • 3 weeks ago
            Anonymous

            The 8086 initially used a 3.5 um node. This was later shrunk down to 2 and then 1 um. Without knowing for sure, I'd assume they already had the 2 um versions by the time the IBM PC was out.

        • 3 weeks ago
          Anonymous

          p. sure the actual die size of those first 6502s was 6 micrometers. in one of their early press releases MOS were promising a die shrink soon. this was accomplished in late 1976 when they went to 5 micrometers. the initial 6502s also had a non-functional ROR instruction which was fixed in the 5 micrometer version.

  21. 3 weeks ago
    Anonymous

    the NES/Famicom was NMOS, no?

    • 3 weeks ago
      Anonymous

      the CPU and PPU are NMOS (what process specifically I'm not sure but probably HMOS II). the RAM chips in the console are CMOS.

      • 3 weeks ago
        Anonymous

        Also NES cartridge components are usually CMOS but sometimes you find NMOS ROMs which you can easily identify as the cartridge will get noticeably warm after running for a while.

        • 3 weeks ago
          Anonymous

          I have a SMB3 cart like that. It gets definitely warm when running for a while. There were apparently several different brands of ROMs used in those, some of which could be NMOS ones. The Ricoh ones might have been NMOS which make sense as they would be able to make them on the same production line as the main NES chips.

      • 2 weeks ago
        Anonymous

        HMOS is simply the name Intel came up with for the 3.5 um depletion mode process they developed for the 8086 and licensed to other manufacturers (initially tested on a DRAM chip prior to the 8086's introduction). It's really just NMOS with a smaller die size, nothing special about it. Using Intel's terminology, HMOS I was the original 3.5 um process, HMOS II was 2 um, and HMOS III was 1 um.

  22. 3 weeks ago
    Anonymous

    Actually even the Model 1 Mega Drive used some NMOS parts (specifically the 68000, Z80, and VDP). the Model 2 has entirely CMOS parts that don't get above room temperature.

    • 3 weeks ago
      Anonymous

      CMOS was always better but in the early days ran really slowly. It took a while to figure out that you could speed it up by thinning the dopant layer (at the expense of making the chip more vulnerable to overvoltage).

      • 3 weeks ago
        Anonymous

        not the dopant layer, the metallization layer

  23. 3 weeks ago
    Anonymous

    I think that's accurate. MOS used 6xxx P/Ns for chips using the original 5-6 um process, 7xxx for the faulty HMOS I process, and 8xxx for the "fixed" HMOS II chips.

  24. 3 weeks ago
    Anonymous

    As an EE, I say this has to be case-by-case. Throwing heat sinks around without reason can do more harm than good e.g. if they fall off and short the PCB.
    Quality of IC manufacture, PCB/circuit design and tolerance to other part variation/failure are a few factors affecting IC life. True it is a waste of time for most instances, but a stoopid looking heat sink in the right place could help if a chip is poorly made, or could be subject to abnormal operating conditions. Any actual benefit, if any, will depend on the context

    • 3 weeks ago
      Anonymous

      In 25 years I've never had a PC where the CPU sink fell off.

      • 3 weeks ago
        Anonymous

        What's your point? PC hsfs are required by design, (supposedly) engineered to a minimum standard and securely anchored.

        This is about retrofitting hs. There have been plenty of instances of cheap aftermarket stick on hs falling off because of poor quality adhesive or crummy installation method. Potential to deface the i/c may also be a concern to a perfectionist.

      • 3 weeks ago
        Anonymous

        Ha. How many have you had? Happens all the time. If it's not screwed down it will come off in time.

        • 3 weeks ago
          Anonymous

          Doesn't really matter how many he's had. No PC in the last 25 years hasn't come with a heat sink that's screwed or clipped down unless it's made by a turbo tard using cheap shit.

          • 2 weeks ago
            Anonymous

            >cheap shit
            Yes? That's the vast majority of computers sold. You can't say shit like it's never happened to me in 25 years if you don't also say on how many computers you've worked on. I own a little over a hundred computers and work on a few thousand a year and I see it all the time. The plastic clips get brittle with age and the cpu cooler comes off.

            • 2 weeks ago
              Anonymous

              >That's the vast majority of computers sold
              No. That's only stupid little children buying garbage online and trying to save a few bucks for fortnite items.
              >You can't say shit like it's never happened to me in 25 years
              You can. Especially when the gay you're replying to hasn't even been alive for 25 years
              >I own a little over a hundred computers and work on a few thousand a year
              No you don't. You tried to put together a machine on the cheap once and broke it due to stupidity.

              • 2 weeks ago
                Anonymous

                If you think I'm bragging I'll have you know I make minimum wage in a computer repair shop and the computers I got from the trash at work so it's not really impressive or anything which is the point.

              • 2 weeks ago
                Anonymous

                But of course you did sweaty

      • 2 weeks ago
        Anonymous

        >In 25 years I've never had a PC where the CPU sink fell off.
        CPU heatsinks are fastened on in various ways.

        I used to put custom heatsinks on graphics cards, the heatsinks you put on memory chips fell down all the time if you didn't use some proper thermal glue. Even the ones with glue pre-applied ended dried and did not hold on. Eventually I just used an excess of thermal paste where the viscosity of the paste itself was enough to keep the heatsinks on, when pushed on right. Only problem is that if it was applied right and dried, you needed to pry it off with a screwdriver with so much force it could damage the board.

    • 3 weeks ago
      Anonymous

      >As an EE

      • 2 weeks ago
        Anonymous

        >t. Schizophrenic
        Scepticism is healthy. Judge the advice. Notice no contradiction to it? Using this information is a process known as "judgement". It may come in handy if you learned to use it. Semiconductor manufacturing quality and component tolerances varied a lot more in the 70s-80s. Parts degrade at varying rates e.g electrolytic caps causing unintended effects on one example of old hardware vs another. The frustrating result is that there is no one answer to OPs question

        • 2 weeks ago
          Anonymous

          >Semiconductor manufacturing quality and component tolerances varied a lot more in the 70s-80s

          i don't dispute that point. it does seem Japan generally had the best chip fabs back then while American or Korean chips could be questionable. some people mistakes hardware failures for the component being 40 years old when it may be likely it just wasn't made well to begin with or up to modern standards of reliability. modern cars are also infinitely better and more reliable than 1950s cars.

          it's like how they used to have parity checking and power-on RAM tests on PCs but that was eventually dropped by the time RAM reliability got good enough that you no longer had to worry.

          • 2 weeks ago
            Anonymous

            Chips are not the only factor, but the design of a PCB and PSU can matter as well. IBM had an advantage here because they knew how to build computers better than upstarts like Apple or Radio Shack and vintage PCs are actually pretty reliable compared to some other retro gear. an IBM PC felt like you could trust using it in an office and storing your valuable data on it while some early computers came off as a collection of electronic test equipment.

          • 2 weeks ago
            Anonymous

            Nah parity RAM still exists its just that ecc ram is much preferred for most cases where it matters. Your pc will likely support ecc if you really wanted to use it.
            Sure RAM is much more reliable now but it still won't last forever.

            • 2 weeks ago
              Anonymous

              >t. Schizophrenic
              Scepticism is healthy. Judge the advice. Notice no contradiction to it? Using this information is a process known as "judgement". It may come in handy if you learned to use it. Semiconductor manufacturing quality and component tolerances varied a lot more in the 70s-80s. Parts degrade at varying rates e.g electrolytic caps causing unintended effects on one example of old hardware vs another. The frustrating result is that there is no one answer to OPs question

              As an EE, I say this has to be case-by-case. Throwing heat sinks around without reason can do more harm than good e.g. if they fall off and short the PCB.
              Quality of IC manufacture, PCB/circuit design and tolerance to other part variation/failure are a few factors affecting IC life. True it is a waste of time for most instances, but a stoopid looking heat sink in the right place could help if a chip is poorly made, or could be subject to abnormal operating conditions. Any actual benefit, if any, will depend on the context

              lol at the assembly language LARPer here. according to a boomer acquaintance of mine, in 40 odd years he's seen only 3-4 or so RAM faults where the chip itself was to blame. in all other cases it was ESD damage or corroded pins/traces. properly made RAM will probably outlast you.

              • 2 weeks ago
                Anonymous

                nobody knows how long an IC on average can last. they've only been around 50 odd years and the average human lifespan is something like 76.

              • 2 weeks ago
                Anonymous

                Probably. My dad has a cabinet-sized tube radio from the 1930s that someone gave him once. It works, although it's useless and has shit sound quality--he just uses it to set a potted plant on.

              • 2 weeks ago
                Anonymous

                >knows jack shit, so relies on boomer who knows jack shit
                You are a living contradiction, aren't you?

    • 1 week ago
      Anonymous

      Now q&a time...

      1)
      Yes I am this guy. Hope you appreciate I was doing my best to give an answer anons could all understand (possibly the ones dumber than you). I still stand by this answer 110%.

      2) yes I just read this shit too, except from science papers and reliable text books. I can send you the right way if you want them.

      3) I typically develop 2 projects or more a year. Don't care if you believe that, but realise me and every other EE actually need to rely on the info in said papers being correct.

      4)

      [...]
      [...]
      lol at the assembly language LARPer here. according to a boomer acquaintance of mine, in 40 odd years he's seen only 3-4 or so RAM faults where the chip itself was to blame. in all other cases it was ESD damage or corroded pins/traces. properly made RAM will probably outlast you.

      Yes I can code in assembly, it's easier than most languages. But you probably can't code in Scratch yet so have no reference point. Off to bed kiddo and don't forget to give your boomer buddy his dementia pills.

      [...]
      If heat is caused by node size why aren't these chips hot all the time due to node size? Why do chips with very small node size run cool when running at a low voltage and frequency? Could it be that there's more to this than misunderstanding a wikipeda page?

      Heat output is a balance of more factors than this. Here's a few starters:

      1) Fab processes vary between companies, even for the same size. This means they vary in efficiency too. Heat is a waste product usually from electrical resistance/impedance and more efficient processes have less waste heat.

      2) there is more to a cpu than the process it is made on. E.g Gold interconnects are expensive but far less resistive than copper which itself outclasses aluminum. What do the CPUs you are comparing use?

      3) your cpu does not have 100% of its circuits being used at any time, but generally the more being used, and the more often in a certain space of time, the greater the current draw and resultant waste heat. So, yes frequency as well as architecture are specifically responsible for generating heat too.

      How many circuits can be used at once is an architectural matter, nothing to do with the process node itself. I believe operating frequency is better determined by working backwards from this (how fast can I run this chip while maintaining electrical and thermal stability) due to variances. This is why overclocking is a thing. A manufacturer should have a good idea of what this should be in the design phase though.

      • 1 week ago
        Anonymous

        > How many circuits can be used at once is an architectural matter, nothing to do with the process node itself.
        In writing this I realised this isn't strictly true, if you are comparing distinctly different processes, but it is good enough for government work!

  25. 3 weeks ago
    Anonymous

    yeah i've also seen NES carts that get noticeably warm in use while others stay cold

  26. 3 weeks ago
    Anonymous

    I saw a video from, I believe it was an old Commodore engineer, on YouTube a while ago about this. He thinks heatsinks won't do much to increse the life of these chips. He also speculated, it's not the constant heat that kills these chips but the thermal shock they have to endure when you power the machine on.
    He further speculated, that a chip pre-heater that increases the chip temperature more gently, would increase their life capacity more than heatsinks.

    • 2 weeks ago
      Anonymous

      >it's not the constant heat that kills these chips but the thermal shock they have to endure when you power the machine on.
      Then shouldn't the extra mass of a heat sink lessen the shock?

  27. 2 weeks ago
    Anonymous

    well i never killed my PS2 despite hours of playing in summer weather when i was a kid

    • 2 weeks ago
      Anonymous

      It has more modern chips that are rated for a higher temperature. pretty sure the PS2's chips can take over 200F. modern high performance chips like a PC CPU also use a copper instead of aluminum insulating layer for even more temperature hardiness.

    • 2 weeks ago
      Anonymous

      the ps2 also has several extremely easy to short ribbon cables if they aren't properly seated, only worth taking off the plastic shell and cleaning out the filters imo.

  28. 2 weeks ago
    Anonymous

    The most reliable retro console of all is probably the M2 Genesis and it's entirely CMOS chips that don't get above room temperature.

  29. 2 weeks ago
    Anonymous

    I put heatsinks on my C64. It's cheap. Easy. And obviously will expand the life on your hardware.

  30. 2 weeks ago
    Anonymous

    you could also replace the DRAM in C64 with modern low power CMOS SRAM if you build an adapter and do some wiring work.

  31. 2 weeks ago
    Anonymous

    [...]

    i think the first Famicoms had heat sinks on PPU IIRC

    • 2 weeks ago
      Anonymous

      They did. the CPU and PPU in a NES are NMOS so they do get somewhat warm when running but nothing too bad and the other chips in there (really just some RAM and a few TTLs) are CMOS chips that essentially don't get warm at all. a C64 has mucho more chips and they're all NMOS.

      • 2 weeks ago
        Anonymous

        >a C64 has mucho more chips and they're all NMOS.
        Still, a C64C or C128 has the newer lower power 2 um chips that don't turn into a pizza oven when running.

  32. 2 weeks ago
    Anonymous

    That's a sad story. I've been banned for not even doing anything though.

  33. 2 weeks ago
    Anonymous

    the Model 1 Mega Drive had some NMOS components mainly the 68000, Z80, and VDP and the former two can be swapped for CMOS variants if you like.

    • 2 weeks ago
      Anonymous

      Are they known for getting hot? I have a 1988 Mega Drive I'd like to keep functional.

      • 2 weeks ago
        Anonymous

        Not really, they just get a little warm. The 68000 in M1 Mega Drives will be either a Motorola or Synertek chip and the latter runs a bit warmer. You can swap the two CPUs for CMOS versions if you like for a little less power consumption but it's not any major necessity.

  34. 2 weeks ago
    Anonymous

    Also I should mention that socketed chips run warmer than soldered ones as the socket traps heat underneath the chip. On C64, note that the VIC-II and SID are always socketed.

    • 2 weeks ago
      Anonymous

      the PCB quality matters too. a thicker PCB with wider traces will dissipate heat away better.

    • 2 weeks ago
      Anonymous

      the PCB quality matters too. a thicker PCB with wider traces will dissipate heat away better.

      I've never considered this. That's some good points.

      in my experience SID seems to always die from ESD damage rather than overheating. the VIC-II definitely has thermal issues though and needs proper cooling.

      All the SIDs I've seen dead were either from Power Supply problems or Joystick Port problems.

      • 2 weeks ago
        Anonymous

        >All the SIDs I've seen dead were either from Power Supply problems or Joystick Port problems.
        nah that's the CIAs that would get zapped from joystick port ESD. SIDs tend to get ESD damage from hot plugging cables or plugging them into a stereo amp although in most cases the analog filters get zapped rather than the core circuitry.

        WDC sells new CIAs but they're a modernized CMOS version that doesn't behave exactly like the original NMOS 6522. also it lacks the RTC feature although that's no major loss.

        • 2 weeks ago
          Anonymous

          >WDC sells new CIAs but they're a modernized CMOS version that doesn't behave exactly like the original NMOS 6522. also it lacks the RTC feature although that's no major loss.
          The W65C22N is billed as 100% compatible with the original 6522 which is one of the more common failure points in C64 due to getting easily zapped from ESD. Beware of CMOS 6502s however as they don't have illegal opcodes and those are surprisingly more common in C64 software than you'd think.

      • 2 weeks ago
        Anonymous

        >All the SIDs I've seen dead were either from Power Supply problems or
        I believe that must be due to the 9V line that operates the analog filters in the chip. Those are very fragile and damaged easily.

  35. 2 weeks ago
    Anonymous

    in my experience SID seems to always die from ESD damage rather than overheating. the VIC-II definitely has thermal issues though and needs proper cooling.

    • 2 weeks ago
      Anonymous

      >the VIC-II definitely has thermal issues though and needs proper cooling
      It has a quite high transistor count for 5 um (13k transistors) and runs internally at 8Mhz.

  36. 2 weeks ago
    Anonymous

    [...]

    Is a 1950s car emulation because you're using modern tires on it and not 70 year old tires?

    • 2 weeks ago
      Anonymous

      you're probably not using the original paint to repaint the thing, and you're also using modern gasoline, oil, etc which aren't quite like 50s gasoline and oil.

  37. 2 weeks ago
    Anonymous

    a couple power cycles killed his ROM chip. really, they didn't make this stuff up to modern reliability standards. at least there's no custom chips in an Apple II.

  38. 2 weeks ago
    Anonymous

    It was in the late 80s or so when electronics started getting to near modern reliability levels.

  39. 2 weeks ago
    Anonymous

    man I'm typing this on a PC from 2011. in the Apple II's time you didn't expect something to run for like 11 years without some repairs. today stuff just works for the most part.

  40. 2 weeks ago
    Anonymous

    Here's an important reason why ICs have gotten more reliable with time that is often overlooked: Modern chips normally have clamp diodes to protect them from ESD damage while 70s-80s chips generally did not.

  41. 2 weeks ago
    Anonymous

    Oldfag bro here. Back in the day me and friends of mine ran our C64s all day long, must have power cycled them thousands of times, ran BBSes off of them and I don't remember anyone ever blowing a chip as a result. Honestly you guys exaggerate how unreliable this stuff was.

    • 2 weeks ago
      Anonymous

      here's something to remember: Storage conditions. when a lot of people would retire a computer as their daily driver it'd end up in an attic/shed etc where it was subjected to moisture and huge temperature swings. That is responsible for a lot of h/w failures.

    • 2 weeks ago
      Anonymous

      I don't see that as the consensus here. Anecdotes of failure aren't equivalent to a claim of ICs being generally unreliable, just that they can and do occasionally go. Again, without justifying their choice of solution, some folks here may have paid big coin for /vr gear. Isn't it fair that they want to err on the side of caution and do what they can to preserve it?

  42. 2 weeks ago
    Anonymous

    Supposedly an 8080 would consume 0.1 watts of power while a modern multicore CPU is more like 100 watts.

    • 2 weeks ago
      Anonymous

      ha no it's not like that. an 8080 actually did get pretty warm as the things used triple rail power. at least the earlier production ones used a ceramic shell for thermal reasons.

      • 2 weeks ago
        Anonymous

        The 8080 had a 6 micrometer die. Larger die sizes get hotter as more power is needed, and the power requirement goes up the more transistors are in there. So while a 6502 and its 3,527 transistors were not a big problem at 5 micrometers, the VIC-II had roughly 13,000 on the same die size.

        Shrinking the die but using the same transistor count will reduce power use and heat output.

        Nothing compared to a modern Intel CPU where if you didn't have the heat sink+fan stack it will literally blow its top like Mt. Krakatoa and catch on fire.

  43. 2 weeks ago
    Anonymous

    you might wonder why phones get by with no heat sink? that's because phone chips use a gallium arsenide rather than a silicon die which is more heat-resistant.

    • 2 weeks ago
      Anonymous

      >phones get by with no heat sink
      That hasn't been true for almost a decade. Started out with the Snapdragon 808.
      >gallium arsenide rather than a silicon die
      Gallium Arsenide isn't a thing with logic chips at all like CPUs and etc. used in SOCs. It's used for any light emission or microwave circuits like LEDs, solar cells, etc. because yields suck and device density is a magnitude less than with silicon.

  44. 2 weeks ago
    Anonymous

    The 8080 had a 6 micrometer die. Larger die sizes get hotter as more power is needed, and the power requirement goes up the more transistors are in there. So while a 6502 and its 3,527 transistors were not a big problem at 5 micrometers, the VIC-II had roughly 13,000 on the same die size.

    Shrinking the die but using the same transistor count will reduce power use and heat output.

  45. 2 weeks ago
    Anonymous

    I don't think a C64 is intrinsically unreliable especially after the first year of production when the early teething problems were overcome but yes, it is helpful to heat sink the main ICs and never ever use the original PSUs.

  46. 2 weeks ago
    Anonymous

    Satan's power supply. Bil Herd said of these "I'm pretty sure the potted power supplies were there to protect Commodore rather than the customer. After all, it's hard to start a fire inside a pound of epoxy."

    • 2 weeks ago
      Anonymous

      I hear the Euro "cheese wedge" was worse than the black US brick.

  47. 2 weeks ago
    Anonymous

    I have seriously never heard of this before.

    • 2 weeks ago
      Anonymous

      like this? the early Famicoms did have a sinked PPU but it disappeared on the Revision D chipset. the CPU is running at less than 2Mhz but the PPU is something like 5Mhz so they may have had thermal problems in the beginning.

      • 2 weeks ago
        Anonymous

        That's a revision E and still sinked. the only toaster NES to use this chipset was the first test run in late '85 while revision G accounts for at least 85% of toaster NESes.

        • 2 weeks ago
          Anonymous

          like this? the early Famicoms did have a sinked PPU but it disappeared on the Revision D chipset. the CPU is running at less than 2Mhz but the PPU is something like 5Mhz so they may have had thermal problems in the beginning.

          they might have done a die shrink later on. there were never any sinks on toaster NESes although the PCB is higher quality than the Famicom's and there's more spacing between components which could have improved heat dissipation a little. the only real sources of heat are the CPU and PPU (since the other ICs are CMOS chips) and they're not as close to each other as they are on the Famicom PCBs.

          • 2 weeks ago
            Anonymous

            that's one of those silly hipster repro PCBs. this is a real original one. the NES PCB is mounted upside down which is a little strange.

            • 2 weeks ago
              Anonymous

              Would crud getting on top of the pcb over time be a possible reason? The silly vcr design might encourage dirt getting inside

              • 2 weeks ago
                Anonymous

                It was probably harder to dissipate heat if the PCB was facing up since it would be obstructed by the cartridge slot and tray. The setup that was used allows heat to simply drift downward and out the vent holes in the bottom of the case.

              • 2 weeks ago
                Anonymous

                ah, ok

          • 2 weeks ago
            Anonymous

            >although the PCB is higher quality than the Famicom's and
            Also true of cartridges. Famicom ones often have rather low quality PCBs and sometimes quit working due to the traces pulling up or breaking. The quality of the shells is also all over the map with some using good quality plastic and others flimsy brittle garbage.

      • 2 weeks ago
        Anonymous

        The initial Famicoms from 1983 had a rather severe thermal problem and would lock up when running which led to them being recalled--those first PPUs had a sink underneath the chip with holes cut out for the chip legs to go through. The Revision B chips had a spreader on top of the PPU instead of under it, and it seems at least some units as late as Rev E were still sinked.

        • 2 weeks ago
          Anonymous

          i wonder if they didn't just have a batch of bad dies and the sinks were a last minute fix

          • 2 weeks ago
            Anonymous

            Wouldn't be the only time that had happened.

            ah, ok

            The vents were actually a problem with cold getting thru the vent and up into the chip on cold mornings causing the ppu to just up and freeze until the engineers designed a little blankie to go over it. That's my story and I'm sticking with it.

          • 2 weeks ago
            Anonymous

            That's like the Commodore PLA where they had such a low yield in the beginning that they were forced to deliberately ship C64s with dud PLAs to meet shipping quotas. The plan was that people would return non-working machines and they would be sent back to the factory and the PLA swapped for a known good chip.

            • 2 weeks ago
              Anonymous

              what happened with those was that they applied too much boron when doping the wafers, which corroded and ate away the aluminum interconnects in the chip

  48. 2 weeks ago
    Anonymous

    Hardware will degrade and fail over time at all temperatures. Higher temperature just grades your MTBF on a curve.
    Keeping hardware cool will improve its lifespan. The only question is just whether or not said thermal factors will lead to it failing at a place and time that actually impacts you.

  49. 2 weeks ago
    Anonymous

    >a meme

    Is OP retarded? Most likely?

  50. 2 weeks ago
    Anonymous

    criminy this thread is 10 days old and still up

    • 2 weeks ago
      Anonymous

      Thread is running hot? Quick, whack a hs on before it fails

      • 2 weeks ago
        Anonymous

        criminy this thread is 10 days old and still up

        was it necessary to samefag upboat your own lame joke?

        • 2 weeks ago
          Anonymous

          Different anon buddy.

  51. 2 weeks ago
    Anonymous

    You can Google questions like "how long can an IC last" and the truth is, there's no definite conclusion or answer to it. Most EEs will tell you it's hard to say, and may depend on operating conditions and sometimes how well-made the IC was to begin with (it's become obvious for example that the first run SNES chipset was not made properly).

    • 2 weeks ago
      Anonymous

      Heat is the main issue. A hot running IC will die early due to electron migration of the aluminum interconnects causing an open circuit. Most chips have a stated maximum temperature on their data sheet and the closer you get to the threshold the more likely it will experience an internal failure from electron migration.

      It's usually one of those two things--overheating or overvoltage that kills an IC.

      • 2 weeks ago
        Anonymous

        so you're saying it really is temperature uber alles?

        • 2 weeks ago
          Anonymous

          Definitely. A cool-running IC will last a long, long time. It's one reason stuff has gotten more reliable nowadays; since modern chips are all cool-running CMOS instead of NMOS chips that you can roast weenies on.

          • 2 weeks ago
            Anonymous

            don't modern CPUs and GPUs need a giant fan stack?

            • 2 weeks ago
              Anonymous

              due to the extreme die density.

              • 2 weeks ago
                Anonymous

                CMOS chips at lower die densities have an extremely tiny power draw when idling and most of them generate little or no heat in operation. It wasn't until 486+ die densities when thermal issues became a problem due to the nanometer gate sizes causing electron leakage.

              • 2 weeks ago
                Anonymous

                Still, new CPUs with orders-of-magnitude smaller lithography are, so far, looking quite reliable with sufficient cooling. Designers are (sometimes) more thermal displacement conscious too. Will these chips last as long as their heat sink free ancestors, is the question

              • 2 weeks ago
                Anonymous

                today they have temperature sensors to cut power off if it gets too hot, and modern high performance chips use copper metallization instead of aluminum for increased thermal capacity. most modern chips have a much higher maximum rated operating temperature than 70s-80s chips despite much smaller gate sizes.

  52. 2 weeks ago
    Anonymous
    • 2 weeks ago
      Anonymous

      I had heard the issue with the first run SNES chips was a packaging one, and may well have been the delamination he describes.

  53. 2 weeks ago
    Anonymous

    as anon mentioned earlier, IC fabrication is a much better-understood and far more standardized process nowadays while in the 70s-80s there could be quite a bit of variance between manufacturers.

  54. 2 weeks ago
    Anonymous

    Military and aerospace-spec ICs would use cerdip packaging as they're subjected to extreme heat, moisture, and vibration.

    • 2 weeks ago
      Anonymous

      Point being?

  55. 2 weeks ago
    Anonymous

    Well fucking shit, this is new to me. I knew the SHVC SNES was a lemon but I never heard that the issue was the chips cooking themselves to death.

    • 2 weeks ago
      Anonymous

      i remember some anon saying the SFC was more reliable as the case has better ventilation than the US SNES.

      • 2 weeks ago
        Anonymous

        conceivable

    • 2 weeks ago
      Anonymous

      someone should use a thermal camera or temperature probe a 3 chip SNES and see how hot they get.

    • 2 weeks ago
      Anonymous

      like this? the early Famicoms did have a sinked PPU but it disappeared on the Revision D chipset. the CPU is running at less than 2Mhz but the PPU is something like 5Mhz so they may have had thermal problems in the beginning.

      I see Nintendo/Ricoh didn't learn their lesson from the first time.

    • 2 weeks ago
      Anonymous

      The SNES chipset is CMOS so I'm surprised that would happen. I would expect to not get much of any heat output.

    • 2 weeks ago
      Anonymous

      [...]
      I see Nintendo/Ricoh didn't learn their lesson from the first time.

      smells like yet another issue with shitty dies. at least with the Famicom they did put a heat spreader on the PPU as a band-aid.

      • 2 weeks ago
        Anonymous

        Well fucking shit, this is new to me. I knew the SHVC SNES was a lemon but I never heard that the issue was the chips cooking themselves to death.

        I was working for a decent IT company when this struck. We sold thousands of PCs with Fujitsu hard disks. This was largely responsible for the business going broke. https://datacent.com/datarecovery/hdd/fujitsu tl,dr it was the epoxy used in the chip packaging. A local rep for Fujitsu said moisture was shorting the silicon but it often took more than a year before a perfectly fine hdd suddenly spectacularly failed.

  56. 2 weeks ago
    Anonymous

    the issue in this case would seem to be similar to the Commodore TED chips where the dies were improperly doped. The guy reporting that the SHVC SNES chips get hot is a giveaway. these are CMOS chips; normally they shouldn't generate any heat at all.

    • 2 weeks ago
      Anonymous

      This. Everyone knows CMOS chips are room temperature superconductors.

    • 2 weeks ago
      Anonymous

      >these are CMOS chips; normally they shouldn't generate any heat at all
      you have to get to nanometer-sized process nodes before that's an issue as the gates experience electron leakage when they're that small. The SNES chipset is probably 1.5 micrometers as Ricoh produced the contemporary RF5C164 for the Sega CD which used that process node, so they probably manufactured it on the same production line as the SNES CPU/PPU.

      • 2 weeks ago
        Anonymous

        That's more of a technicality thing. By the time 1 um(micrometer) was a thing, things were already verging on needing at least a heatsink and the next shrink down to 800 nm, Intel's 486 needed at least a heatsink and by the end of that node, the Pentium P5 made on the same 800 nm node needed active cooling. Those nodes were not easy to transition to and to get right without active cooling so I wouldn't be surprised if the late 80s to early 90s would've had chips with heat problems.

        • 2 weeks ago
          Anonymous

          the 486 also had over a million transistors. the stupid high transistor count matters as much as the process node.

          • 2 weeks ago
            Anonymous

            I ain't saying that it was the sole reason, but just that it was that point where you couldn't ignore heat generation any longer if you were aiming for high performance. Motorola, AMD, MIPs and Sun all needed heatsinks which were in most retro computers in that era other than really low end stuff which could get away with it but which barely anyone remembers or wants to use nowadays.
            5th generation consoles largely escaped that because they didn't clock as high but N64 didn't even though it consumes like 19W. PSX narrowly avoided it but it did cause plastic warping which in turn caused CD laser read issues which were eventual RMAs down the line. Saturn avoided this by using dual weaker CPUs but it was a nightmare to program in return. But no one did everything passive once 6th generation rolled in, not even the Dreamcast which was the weakest but lowest consumption console managed that.

      • 2 weeks ago
        Anonymous

        That's more of a technicality thing. By the time 1 um(micrometer) was a thing, things were already verging on needing at least a heatsink and the next shrink down to 800 nm, Intel's 486 needed at least a heatsink and by the end of that node, the Pentium P5 made on the same 800 nm node needed active cooling. Those nodes were not easy to transition to and to get right without active cooling so I wouldn't be surprised if the late 80s to early 90s would've had chips with heat problems.

        If heat is caused by node size why aren't these chips hot all the time due to node size? Why do chips with very small node size run cool when running at a low voltage and frequency? Could it be that there's more to this than misunderstanding a wikipeda page?

  57. 2 weeks ago
    Anonymous

    If you want to know specifically why heat is bad for ICs, two reasons.

    1. It can cause the dopant molecules on the die to shift around. If they drift out of place, the transistor ceases to be a transistor. This doesn't happen immediately, but years of exposure to high operating temperatures can do the trick.
    2. It can cause an interconnect to fail from electron migration.

  58. 2 weeks ago
    Anonymous

    You know nothing, anon that doesn't reply. IC designers limit heat for plenty of other reasons. Explain impacts of the silicon thermal limit on circuit design, without reaching for Google or parroting another post. How about the impact of heat on non-ic components? Can't wait to hear

  59. 2 weeks ago
    Anonymous

    Heat is everything. The actual silicon in the chip is in no danger of meltic as silicon can handle extremely high temperatures. As anon said, hot chips can fail from the dopants moving out of place, and every 10C of temperature you knock off will double its lifespan.

    • 1 week ago
      Anonymous

      Sorry anon, I was in a mean mood earlier. The cycle of anons repeating ignorant BS here is just shitting me to tears.

      No heat isn't everything. Conventional knowledge (not horse shit here) is *the packaging* is the most likely element to fail, which can happen in a few ways (moisture absorbtion, cracking from shock, distortion causing stress on the silicon being a major cause etc). Epoxies used to package ICs vary in composition and, importantly, the fillers used to cheapen their manufacture, improve elasticity etc. Ceramic is better but rare.

      Heat rates a very distant second. It affects electromigration RATE, but EM will eventually kill any chip that doesn't die from something else first at room temp. Heat can aid package decay but is unlikely - the epoxies used are less susceptible to heat than previously mentioned problems.

      Like packaging composition, each process node can involve different proprietary methods and will have different limitations. It is difficult to nail down generic IC properties and limitations but silicon itself has an unusual conductance/temp curve that imposes a relatively low practical temp limit with current technologies (this may well change). Silicon at 30 degrees and 300 degrees F have roughly the same conductivity and between here is lower - a safe zone for the silicon itself, ignoring all else. Although the resistance actually starts dropping at 200 deg and temp can be much harder to control at this point. Too low and thermal stress can shatter the core, too high is MUCH worse because...
      1) cheap packaging will more likely distort / fail.
      2) lower resistance will likely cause excessive load on power AND signal circuits connected to it, meaning the system will simply start to break down e.g corrupting data. This state may be just temporary, unlike...

  60. 1 week ago
    Anonymous

    3) results of interrelated effects of dielectric breakdown, thermal runaway (lower resistance draws more current, creating more heat, lowering resistance, drawing even more current and so on)

Your email address will not be published. Required fields are marked *