Did I KILL my rare vintage Mac clone?!
HTML-код
- Опубликовано: 27 сен 2024
- I think I screwed up big time.
3D-printable CPU bracket: www.printables...
Sources:
"The Mac Goes Multiprocessor," Byte, February 1997.
Power Mac 9500/9600 motherboard photo: fr.m.wikipedia...
---------------------------------------------------------------------------
Please consider supporting my work on Patreon: / thisdoesnotcompute
Follow me on Twitter and Instagram! @thisdoesnotcomp
---------------------------------------------------------------------------
Music by Soundstripe (www.soundstrip...) and Epidemic Sound (www.epidemicso...).
Intro music by BoxCat Games (freemusicarchi....
Maybe reach out to Steve from Mac84 about this, he’s a wizard when it comes to vintage Mac stuff
nah not good as dosdude1, he swaps cpus all the time
@@DatBlueHusky both are equally great.
@@RetroTechChris my thoughts exactly, Chris!
@@esseferio I might have already summoned someone. LOL!!
Dosdude is the real genius bar of macs from late 90s
I would recommend you don't feel bad about this. You did the best you could, took good care of it, and treated it well. It is almost 30 year old technology. The fact it works at all is amazing. And that looks like an intense cpu card anyways. It is possible heat just killed it.
Yep. I wouldn't even swear off preventative maintenance. These things are on borrowed time. If you don't maintain them, they'll fail. If you DO maintain them, you might kill them. Either way...
We're with you, if you need to bury it and get an alibi we're here to help.
Friends help you move. Real friends help you move bodies
I don't know what you're talking about, he was at my place the night this happened
It's no real loss. The machine is useless in the modern day as a multi-processor machine. It only utilizes the extra CPUs in a hand full of applications. If having a multi-processor system is that important, he could put in a G4 multi-processor upgrade card which would run circles around the original card.
@@tarstarkusz Haha, you have no clue what the draw to a machine like this is
@@repatch43 The machine itself is the draw. The fact that one card has been replaced is irrelevant. The proof of this is all these upgrades were used back in the day.
One thing that might be work trying is an oven reflow
The stress of the heat sink may have cracked other solder joints, so heating it in the oven could reflow them, potentially fixing the board
Same issue as the XBox 360!
@@kirishima638 : sort of i've worked on those but they are GBA with none lead solder that was the issue .................
Oven reflow is never a good idea. Reflowing with a heat gun and some flux is much better, although it probably will not work either.
I don't think it'll fix the problem, but I don't see the harm either. As long as it's done properly...
@@RoxGYT Never? That's a bit of a stretch. The heat gun method has its flaws too.
Could you use a thermal camera to spot the processor that is not working?
If it doesn't boot, not sure if any would receive enough power to increase temperature. Maybe he can identify using a voltage level in one of the legs though
@@mmpsp693 without the heatsink he could probably try to run it short enough to see a small difference between processors easily.
@@mmpsp693 it is worth a try. All with equal temperatur: Nothing learned. One CPU hot: Bingo.
Sure, if it does not boot it might not heat up at all, but if he has a thermal camera and the heatshink removed for other reasons, it might be worth a shot.
Dos dude might be the guy to call
was about to comment this! He would have to resist the urge to change it to a quad g3 card though :D
@@SAerror1 He would have to design some sort of board that converts the QFP footprint on the board to a BGA footprint before that's possible, though!
@@SAerror1 He'd probably upgrade it to to something never heard or seen before, still, it might be possible to swap the cpu chips with "new" ones
This was a MAJOR effort to repair the fault. Cleaning, re-soldering, even 3D printing! But sometimes these chips do go bad with no apparent external cause.
I recently put an SSD in my old 2012 iMac from storage and it suddenly stopped booting. Cause was a faulty RAM chip, which I never even touched.
If it makes you feel any better, back in 2015, I restored a beautiful example of a PowerBook 180c. Cleaned, recapped the screen, replaced the drives, new memory, the works. I took it apart to get some pictures of the inside and in doing so, tore the tiny and thin display cable under the hinge that is unique to this model, instantly bricking it.
Unfortunate, but not terminal as I had another 180c for spares with a cable.
And then I tore that one too...
I had to wait many, many years for another working one to come up on ebay at a reasonable price, during which the 180c collected dust. It had to be working one to prove the cable was intact so I didn't just buy another useless machine.
I even spent a lot of time trying to make my own connector from scratch.
Could be a bit strange, but have you tried to start the Genesis without the heatsink installed?
Just to see, if maybe the heatsink is attached with to much force, so that the CPU-Board is slightly bended and makes no contact to the mainboard.
Was thinking the same
back then it wasn't a guarantee that cpu's will self thermal limit. if you fire it up without a heatsink it might cook itself. i accidentally roasted an athlon this way once. my heart sank when i saw the smoke. didn't even get to bios.
One of the first things that suffers from PCB warping are SMD ceramic capacitors. These kind of details cracks very easily and when they crack, it almost certainly leads to short circuit. You can find broken one with multimeter in continuity mode.
Might be worth grabbing or borrowing an oscilloscope, grabbing the datasheet for the Motorola MPC952 PLL driver, and check if all frequencies are there. Maybe a broken connection related to the clock signals could keep the CPUs from getting a clock frequency, which would obviously keep it from doing anything.
That is my thought as well. The Problem needs to bei narrowed down on the CPU Board.
Seems to me like a broken trace on the cpu board due to the heavy bending. This would explain the reseat working/non-working issue too. But could be hard to detect
I have a mac laptop, is the only mac I only have in my life, since I live in Argentina and macs are not so commond. I disassembly it a long time ago, and broke some things (I wasn't so experienced nor seen a lot of videos like this) and then, time crashes the screen, and now is just a big paperweight. I still have it, as a reminder that it existed.
Anyway, kisses from Argentina and I enjoy your videos a lot, keep it going.
Sorry to hear about the machine. I was stoked to see that one of my BlueSCSI brackets lives in it!
Interesting video! Have you considered that the slightly warped expansion card may not be properly seating in the card slot? If one or more of those metal connections along the card edge isn't mated with a corresponding pin in the card slot, then that would explain the boot failure. It would also explain why reseating the card worked... until it didn't. Removing and replacing the heat sink may have worsened an existing issue with the warped card. The solution might be to either A) straighten the card somehow or B) tighten the cardslot somehow.
That was my thought as well. 3d printing a shim/spacer to run down the center of the card under the heatsink might be a cheap/easy thing to try before swapping cpus on the card. Figuring out the dimensions would be the hardest part, though measuring the gap near the screws is probably the best way to get the right dimension.
11:24 That's a cool little 3D printing hack, thanks!
I'm wondering if those nylon washers are supposed to go between the CPU card and the heatsink. That could explain the card being warped.
Soul destroying stuff, I really feel for you Colin. I managed to kill my PowerComputing CPU card back in the late 90s. In my case, I suspect it was static electricity that did it.
that's the main reason why i don't take apart electronics anymore.
like my grandfather says if it works fine let it be and don't mess with it
Nah! Don't be afraid to take things apart... sounds like it already had a problem. I don't think you broke it. It's possible that someone could fix it.
I appreciate the honesty sadly its good to show the actual issues that can be faced now a days with these kinda machines you’re really one of the only guys that does this especially with Mac’s keep it up!!
I have a 9500 with a dual processor card. If you run BeOS on it, it uses both processors and runs in SMP mode. Worth doing a video on it.
THANKS for making the STL Available!
Awww I feel for you. I've deffinitly sunk hours into repairs that went nowhere. Could you Try to determine which processor is the primary one and simply swap it with another on the board? Sure it'd only be a 3 cpu card, but at least it might boot?
Or swap it with a 604e from a sacrificial common single CPU board
It seems that there ought to be documentation identifying the primary CPU. But of course I don't know if there is.
Shot in the dark: Remove the heat sink, clean up the cpus. Install it, power it on, just for a minute, and use a thermal camera to see if there is a chip on the board that is hotter than the rest. My thought is that it isn't the CPUs but another chip on the board. If worried about the CPUs, get some individual heat sinks to attach temporarily.
Actually, attaching some individual heat sinks might take the strain off of the board.
You mentioned that the dual processor came from another unit. You mentioned that the motherboard is the same as another unit. Have you tried the quad in another unit? Case wise it might not fit, but what f you took the motherboard out of the other unit, put it on some stand offs, and tried it open air to see if the CPU works in another motherboard?
First, thank you for the isopropyl alcohol soak porn shots.
Second, you're really good at making us feel good about a bad ending.
Appreciate you and all of your content.
I confess that I enjoyed that soak too
check the tantalum caps on the back side of the cpu card. use a continuity tester, should chirp at most, if its a hard short then they need to be replaced. Common for old tantalum to fail short.
This seems like a job for dosdude1 to tackle. They could probably fix it, not sure if they are headed to VCFMW maybe you could discuss it with them? Seems like it be a decent collab/commission
Is it possible that voltage regulation is the culprit? I don't know enough about add-in cards to say for certain but it might be an external problem to the CPUs themselves
You gave that computer a life for it's last golden years. More than you can say about 99,% of other computers
Nothing ventured, nothing gained. There's always a risk (not RISC) with old machines that this can happen. But the educational value of the videos and what you personally learned about the machine makes it worth it. Besides, one day you may have a technician's epiphany and suddenly realize that the problem was hiding in plain sight. Also, whether it's a successful repair, or a vain attempt -- I'd rather watch someone who is being honest.
I recommend testing each trace with a multimeter. If you can find the broken one, you might be able to use a jumper wire and fix it that way. Also, I recommend modifying the heatsink so that it causes less stress. (Seems like the legs on the side are too long, maybe sand them down a bit?)
Wow colin big wow, amazing effect buddies. Best video so far for me. PS does PCBway fix boards
I suspect your diagnosis is correct. One or more of the CPU's on the card has failed. Perhaps send the card to a repair shop with surface mount tooling (Northridge Fix or similar) that could either remove and resolder the CPU's or perhaps find four matching ones to replace them?
IBM chips in that blue package are pretty fragile. Not only PPC CPUs die, but also those IBM DACs which were used in many professional boards, like Number Nine Revolutions and such.
IBM RAM chips from that time are fragile too.
Did you check all the 'via'? the connections between the two sides of the PCB? Almost certainly one or more of the vias near the connector is broken.
Electronics Engineer here. I don't see anything obvious that you've done that would have killed that CPU card.
That sucks but don't let it get you down. And don't let it discourage you from working on older tech in the future.
I'm surprised the CPU card from the 9600 is compatible. I know Apple licensed the tech from Daystar, but didnt realize they used it without any chonges.
All of the CPU cards from the 7300, 8500, 9500, Daystar and some other clones are interchangeable. There are also some other models, like the 7600 can I believe use the cards, but came with an older processor. The original 9600 with tsunami logic boards are also compatible, but the later Kansas ones changed the architecture and are not compatible.
I'd like to see you demonstrate one or more of the Orange PC cards. They had a long history from the 68K Nubus years almost until Apple switched to Intel CPUs. Most of the Orange PC cards contained all the bits and pieces of an IBM clone, but the final PCI slot models were mostly just an x86 CPU onna card and used software to emulate the rest of the hardware due to the G series CPUs and RAM capacity of the later Macs being able to handle those tasks.
Of course with the switch to Intel, a PC card was no longer needed to run PC software since one could simply dual boot the Mac with Windows.
Plot twist it was a broken solder joint on the processor card
One (stupid) question ; Did you try to put the CPU card in the machine but without the heatsink ? Maybe the weight of the heatsink (i don't know of what material it is made but it's look quite heavy) is one of the possible culprits of the failure since the warping of the board. You don't have to let it run for long, just enough to let the chime sound if removing the heatsink does something.
After, one of the possible solution, but it's really long and with the possibility of "unobtainium" components (I don't know if we can get BR new PPC 604E chips and there is an unidentified IC on the board), it's to do a Reverse Enginering of the card and replicate it new.
Did you check over the Passive Delay Line's or the 44mhz oscillator ?, My moneys on the oscillator. Put it on the scope when its powered up and use the frequency counter measurement function.
I
edit: Similar delay line spec sheets say they have an internal fuse. I would honestly start by checking them and the clock source if I was in your shoes. I probably would have ripped the oscillator off and feed it externally off a signal generator during testing.
Mabye some traces got cut/lost connection because of the bending of the cpu card.
Watched a video of PCB damage that can occur with very heavy modern GPUs and it reminded me of this. In that case the pads of the chips had been ripped off by rough handling and the weight of the thing. Maybe there could still be something like that happening in this case as well instead of a whole CPU being dead.
that computer is huge, you could fit your whole arm inside the case
nothing ventured, nothing gained.... don't feel too bad about what happened!!
Don't feel too bad about taking the card out. It was neat to see how they built quad core boards back in the day.
If the computer is very old but still working don't touch it let it keep working. That's what I always think when i working with old system at my job. I hope you find somonoe that can help you with this awesome machine. Thanks for sharing. 💪
$14,400 USD in 1994 is $29,313 USD in 2023. People balk at a $7000 Mac Pro these days, but in reality, Macs have never been cheaper.
Just wondering, did you ever try the card without the heatsink? Even at least seeing if it fit a bit better since it was bending from the pressure of the heatsink?
If I may be so forward: You seem like the type of guy to have a mate, even if they're hidden in the recesses of your FB friends list, that can examine hardware at that level and test it on a bench or something. I wonder if there are clones of those chips somewhere . . .anywhere . . .even in an FPGA arrangement . . . to help breathe life back into that system.
if it problem with one of the dies, heat/cold could help locating the problem.
I assume you checked all the other components on the cpu card? there could be cracked SMD's, and those are really hard to detect
Have you ever watched Adrian's digital basement? He's a wizard with fixing this sort of stuff and loves to collaborate. Might be worth an email.
Although Adrian is really brilliant at repairs, he doesn't have the required expertise in smd soldering. In fact he repairs almost only through-hole components based motherboards, such as commodore 64s, amigas and computers of the same era.
my condolences for your loss.
A valiant effort, for sure, but one thing you could do that would be easy to verify is to inspect the pins on the bottom of the CPU card and see where it's bowed out the most. If you were to place a dollop of solder on those pins so that it would effectively create a bit of height on them, it could very well make full contact with the slot on the motherboard. It is very much within the realm of possibility that there are hairline fractures or other such issues as suggested by other people, but with the massive heatsink cranking on the PCB, I believe it is quite likely that it's just not fully connecting all the way. Macs aren't consoles, but my experience with working on vintage game consoles is that the connector will lose their ability to clamp after decades of service.
This is a bit of a problem in and of itself, but when you've inserted a card that's effectively flexing the entire PCB, the problem compounds upon itself.
Measure and inspect the resistors on the bottom side of the board. The warping could have caused a resistor to fracture or its solder to crack. I would pay particularly close attention to the solder joints on those resistor packs, the like to crack on the edges.
It looks like the debug connector at J3 is an SPI programming header for the Lattice PLD, you could trace out the pinout using the Datasheet online. The other connector is probably some sort of manufacturing test header, it looks like some pins go to U6, but I can't see what that part is.
Ok Final suggestions, Did you check the solder joints on the power connector to the CPU card? (I imagine you did, but didn't mention it in video.) I'm banking that you are most definitely right with the cause being from the card warping. My "broken for parts only" Neo-Geo MVS board had failed from that, luckily in it's case it only broke SMD solder joints, was just a 4-layer PCB (bottom, gnd or pwr plane, opposite plane, top), and was easily fixable. Hopefully this isn't a 6+ layer PCB and you don't have broken traces in the middle of the PCB. Also I'm not sure what happens to the decoupling capacitance if the middle planes break from flex. It's a very weak capacitor, and is only used to filter out voltage spikes/noise, but maybe it's enough to cripple the CPU. Would be neat if there were someone willing to "x-ray" the PCB for you. Oh, You never really gave me a good view of the back of the card (unless I missed it), although I'm not sure if there's really anything more to spot. From the end shot it looks like it's just a bunch more inductors.
During video comment scratch: I keep pre-typing suggestions, then you do them. 😅 8:34 That cap directly next to it looks ready for replacement too. Though I suppose in the perfect world you re-cap the entire board with SMD Tantalum. 10:20 Tantalum should also last you forever. I'm surprised you have electrolytic caps on hand, since they have a shelf life that eats into their service life.
Heat up each core with a an air solder to around 320celsius for few minutes, then try again.
Just a thought... Perhaps you could 'rotate' the CPUs so that one of the other, hopefully good, ones gets used by the system.
he is not equipped for that, but could work
@@FlnSkr Perhaps if someone did have such equipment they could do so. I think that the rarity and value of this machine makes trying well worth it.
@@denniseldridge2936 of course. a good equipped repair shop should be able to do it no problem. it just requires a hot air station with a big chip nozzle, maybe. ;)
@@FlnSkr One of my first jobs was at a factory testing expansion cards for PCs and other systems in the mid 80s. I used some of the early surface mount soldering stations much to my fingers' dismay lol.
Sorry, you just brought back some cool memories 👍
@@FlnSkr low melt solder and a little hot air would remove them very quickly and safely.
I know this feeling too well, but don't feel bad because it wasn't your fault. You did the best you can and maybe someday you'll find a fix. I'm still holding on to my childhood 486 motherboard which recently crapped itself in hopes that I might one day resurrect it.
With a PCB bend like that, it was gonna give up the ghost sooner or later. I think it's great you got the chance to document the system before that happened.
it'll give up the ghost, maybe bite into the grass even.
Sorry you’re having these problems. This looks like such an awesome computer that I’ve never heard of before and I want to try to find one very badly lol with four processors even though it’s from 1995. Is this able to play RUclips videos or run Linux? I’m not sure how much testing you did while it was still working. Thanks I love your show.
Given that these CPUs are older than the first G3 processors, I believe they would have a hard time running Mac OS X - let alone stream video from RUclips. About a decade ago, I was able to get RUclips videos to load on a 350MHz iMac G3, but it was excruciatingly slow and laggy. It is unlikely that modern RUclips would be able to load on any Mac OS 8 or 9 browser, but I could be mistaken.
I went on a spree trying to find rca lyra m200 series mp3 players for nostalgia reasons, but barely found out that there was an expiration date on the screens they used. Now i have three technically working mp3 players with disfunctioning screens. Additionally, the m2000 series are the rarest as none are being sold online.
At 7:30 when testing a power supply you should test it under load because if an output capacitor goes bad (and those go bad a lot) it might provide the correct voltage to the multimeter but not to the board. Hope this helps!
Thinking further on this issue I’d also say that if a Power Supply output capacitor was going bad it would have done the same intermittent failure where removing the card AND the power would have been the thing that “fixed” it. Don’t despair! Power supply issues, capacitors and other discrete components are far more common failures than CPUs! You should try to reach out to Adrian at Adrian’s digital basement as he is very good at troubleshooting this issues and hopefully he could throw a couple of ideas before even having to ship anything.
Just some sparse thoughts:
- i see another big chip other than the 4 cpus on the board. Did you rework it too?
- to test if could be some cold boot issue you can try to leave it on for 10-30minutes to heat up and then restart it, or heating the board while on.
- did you check the voltages with the board installed? the psu may be too weak for 4cpus but still enough for 2. Measuring with no load is almost pointless.
I'd love to see if some other hardware youtubers can help debug this issue. I think there's hope of at least learning what's wrong. That debugging port _looks_ like a pretty standard serial interface.
I KILLED my 2012 Macbook pro yesterday replacing the battery I know how you're feeling. ugh
Did you check the clock signals for the main processor?
of the 4 processors do you know which one was the shall we say the lead chip' is it possible to move one of the chips to the main chip position ? you have hot air could you flote one off to see if any trace's are under the chip ??
you might want to swap the CPU places on the board by re-soldering. this would verify your theory.
For me this looks like a perfect example for hairline cracks. Have you checked all traces on the card? Is it a multilayer PCB?
Colin, did you try to reflow the entire cpu card? Another thought.... does the angle of the processor card return to normal when you remove the heatsinc? Maybe a slightly bad connection on the motherboard is made worse with the bent cpu card and has gotten to a point of not making contact. Maybe try to run the cpu card without the heatsinc attached and see what happens. Just a thought. Thanks for the great video's and don't feel too bad about it. We've all been there.
Why not swapping the CPUs so that the primary processor gets replaced and at least get it “working”.
Could it be that the CPU card has any cracked traces that could be fixed with bodge wires? Just a hunch.
I would have desolder all the CPU's and switch their places, so the primary one would be elsewhere and you could isolate the CPU that way and maybe replace it even with one from another clone. If you do all that you would be a hero. I used to work for Apple during those days and I used to own a Power Computing.
Rossmann repair group, try them
Don't feel bad mate. Like you've always said, 30 year old tech cannot last forever. The CPUs probably finally died and you just need a new old stock to bring it back. I hope there's a solution soon.
You seem pretty proficient at soldering, you should try to locate some 604s and test swapping them out. Even if you have to pull the off a more common board, this particular computer is rare enough to justify it.
It's an interesting philosophical question with anything vintage. Do you upgrade it, mod it, preserve a previous mod, or return it to factory standard? Do you pack it up so dust can't get at it, or use it?
If you preserve it by not using it, who cares if it works or not? It's just to look at, and not to be used. If you use it, there is always the risk that it will break.
Everyone will have a different idea of what to do with something vintage, but whatever is done, should be done to the enjoyment of the owner.
It died by being enjoyed, not by being thrown away. Probably the best way to go rather than a screw up.
I kept waiting for the part where you killed it…. it never came. I think it was already bad/failing. Don't beat yourself up on this one, tried everything you could (way more than I could imagine most doing) and it still worked with all those other cards. I'd be curious if you checked the voltages WHILE load is applied to the system however. Voltage may be fine with no load but may dip with a load.
Just an idea: it can be a marginal power supply that browns out under a higher load. 4 CPUs surely take more power than just 2, and this can hit the inflection point of the PSU.
@ThisDoesNotCompute a quad CPU setup is going to need some quality power. Was the power supply tested under a load for sagging voltage or other issues? Perhaps run the processor card from a secondary power supply? Booting on the second power up is also something that happens with failing/edge case power supplies.
Bake the CPU board @ 300-500C for 10 minutes. A good baking brings back many chips. I would just give it close to 500C untouched for 5 minutes first and let it cool off, worth trying.
Aw man. I feel your pain. I've buggered up old tech by taking it apart because I'm a ham-fisted oaf! It happens, sadly.
you can try a crazy idea, swap all the cpu chip around, probably one of the 4 processor is dead, but as you said, the system boot on one cpu and only next can use the other 3 cpus, so, if you swap the cpu probably start working again
Should test the PSU underload
I’m sorry you’re so disappointed about this special machine dying, but thank you for sharing it with us
Don't feel guilty, there was nothing you could have done to save it. Re-inserting it to get it to work for a while didn't kill it, it's age did. Remember what Ben Franklin said about Computers "The only thing you can count on is death and collectability"
... taxes
I'm suspecting cracked solder joints, it's old and has served many years. That bend on PCB looks nasty and I wonder if it's not making good contact with the mobo.. Could you check the voltages when the card is in place?
Maybe someone like NorthWestRepair could reflow the whole board and check the traces, then transplant working CPU from a card that has same CPUs
if the failure is 'on-die', it could need to be put on a heat bed and swap out each chip and have them tested, and replace the possibly offending piece with a healthy harvested part
F in the chat
Press F to pay respects
F
F 😬
F
F
DOSDude to the rescue!!! If anyone can perhaps fix this issue, he can. Guy is a genius!! Good luck man!!
Looks like that CPU card is the only one out of the bunch that uses additional power, did you check the PSU?
A PSU is a lot more likely to die than a CPU.
He checked the PSU but without Load. Maybe rechecking with an oscilloscope under load?
@@rubberduck4966 He did? I must have missed that part somehow
Great work again.
You need to find pinout of CPU chips, and riser cable to test outside of case. Check VCC and ground. Check for clock signal going in on each CPU. Then probe data lines coming out. If there's signals going in and nothing coming out, then there's the problem. You'll have to carefully remove that chip, so as to not rip tracings or lift any solder pads. Pre-tin all legs on new(ish) CPU with fresh solder. Wick and clean pads. Flux and drag-solder pads. Then place CPU, and reflow each leg, one at time. Time to clean off flux and test board. Good luck.
I wonder if it's possible to disable or hault certain CPU's within the bios or something on this computer? I'm sure your idea on a fried CPU is correct!
Macs don’t have a BIOS, and any open firmware configuration needs a booting machine. The CPU cards are plug and play from what I understand.
That is thermal compound not paste, and it is a little bit too much, doesn't need to overflow off the die. This is not the cause of death though.
Doesn't matter if you did, you can't take it with you. There are no pockets in shrouds.
Didn't see any mention of the power connector on the CPU card in the vid. Was that reflowed in case the solder joints developed cracks from all the connect and disconnects?
The next step is to hook up a JTAG debugger and hope that the boundary scan files are available. 😆