A spudger tool might be your best bet, or even an old license or credit card.

Have an image/video?

This has the basic technique, though you may want to adjust the tools used to minimise the chance of damage

https://web.archive.org/web/20000815213023/ht … oval/index.html

Thanks a lot for that link. The operation sounds too risky for a 1 GHz Thunderbird. My T-bird was giving all sorts of memory errors - himem, cachechk, speedsys, then some odd smell, then no more boot. I was thinking I might take the enclosure off to see if some external component fried, but this sounds risky. Do you think the most probable cause of failure is inside the CPU itself rather than some PCB component?

The main problem with this type of cartridge (Pentium II is practically identical to Athlon) is that one of the plastic holding plugs is much tighter than the other three and that separating them is most likely going to cause damage.
There is an overview of a procedure here:
Hello, World! :D - Robert B's PC builds - oogle away freely :)
scroll down to:
2016-07-09, 05:59 The K7 slot A - build
1st Step - Slot A Cartridge removal/disassembly/teardown/dismantle

I did two of such operations in the past and did -not- manage to complete them without damage to the plastic shell but imo using circlip pliers instead of regular pliers gives better control.

Here is a russian video (Pentium II) which shows the above mentioned problem but the way how this operater handled the separation is obviously not meant to end up with re-usable parts.
https://www.youtube.com/watch?v=-cdWYEARJYo

Was going to suggest Robert B's approach, but it's been covered 😀

What I can confirm from my limited experience is that the thermal pad used in these CPUs dries and turns to some solid crumbly mass that likely has very poor thermal properties. To make things a bit worse, there's some sort of a grid/mesh made probably of fiberglass, embedded in the pad. I think the purpose of this was to keep the pad material from spreading out (migrating) from under the die, due to the pressure and heat. But now, with the pad completely dry, this mesh only serves to keep the die and heatspreader slightly apart.

Noticed all of the above on one of my Slot A processors, one that I got with its plastic cover already removed/missing. I have 4 other CPUs but what I've seen in that one makes me wary to use them for anything but a quick test, until I open each of them for cleaning and applying fresh thermal paste. I'm actually thinking of ways to improve Robert's method, maybe by using two jaws or plates that can be slowly spread apart by turning a screw.

My theory about your CPU is that it slowly overheated and killed itself, so I don't think you have much to lose from opening the case. Not much to gain either, but at least you can confirm what went wrong.

quicknick wrote on 2020-05-06, 12:02:

Was going to suggest Robert B's approach, but it's been covered 😀
8><CUT
I'm actually thinking of ways to improve Robert's method, maybe by using two jaws or plates that can be slowly spread apart by turning a screw.

If I ever have to do it again (unlikely) I would first drill out the plastic on the location on the shell where the tightly fitting holding plug sits. The other three are relatively easy to separate. This way the damage is controlled and it is not difficult to make a fix for the missing part.

So, you mean if things go wrong the damage is so big that it becomes obvious from the outside? I thought that maybe the prong breaks or gets too loose and the shell won't fit that tightly afterwards, wasn't expecting holes to be ripped apart in the shell. The whole deal just became scarier for me 😀

Thanks for the additional information. I guess my concern with ripping off the cover is: what’s the collection value of a defective, but in good visible condition, slot A 1 GHz thunderbird? Seems like CPU collectors aren’t concerned much with working samples.

quicknick wrote on 2020-05-06, 12:19:

So, you mean if things go wrong the damage is so big that it becomes obvious from the outside? I thought that maybe the prong breaks or gets too loose and the shell won't fit that tightly afterwards, wasn't expecting holes to be ripped apart in the shell. The whole deal just became scarier for me 😀

It depends on what the plan is. If you want a working cpu that can be easily taken apart you could probably live with a small hole in the shell. If you want a pristine looking case you need a different approach.
RobertB's effort showed that only 1 in 4 survived his method of operation without any damage.
Your idea of making a special tool is probably the best but I doubt that it is worth the effort if you only have one or a couple of cpu's to fix.

The thermal pad theory is a plausible one. At this time AMD didn't have a reliable thermal protection circuit in their CPUs. See this classic video as an example: https://www.youtube.com/watch?v=NxNUK3U73SI. They didn't totally fix the issue until the Athlon 64. The video shows Socket A CPU's, but I can't imagine the Slot A chips have it if the Socket A chips did not. There would be no reason for AMD to remove it.

I'm sorry to hear your 1 GHz Slot A T-Bird is likely dead. It's such a rare CPU. Now I'm worried about all of my Slot A CPUs, but don't want to risk opening them up.

Katmai500 wrote on 2020-05-06, 15:48:

The thermal pad theory is a plausible one. At this time AMD didn't have a reliable thermal protection circuit in their CPUs. See this classic video as an example: https://www.youtube.com/watch?v=NxNUK3U73SI. They didn't totally fix the issue until the Athlon 64. The video shows Socket A CPU's, but I can't imagine the Slot A chips have it if the Socket A chips did not. There would be no reason for AMD to remove it.

I'm sorry to hear your 1 GHz Slot A T-Bird is likely dead. It's such a rare CPU. Now I'm worried about all of my Slot A CPUs, but don't want to risk opening them up.

Yeah, good to be worried. I found 2 other dead slot A's in my pile.

So does the dead slot A 1 GHz t-bird have no collection value then?

feipoa wrote on 2020-05-06, 12:21:

Seems like CPU collectors aren’t concerned much with working samples.

Ohh ? I beg to differ ;p -> insert joke
I have a few Slot A and am thinking of a way to redo the thermal paste. One is in New condition, the other the guy got it from tried to jump the thing to make it work/seem like a faster processor and cut a portion off the top. IMHO any Slot A that works, no matter what the plastic looks like, is mega-better than a dead pristine one. I liken it to a full bottle of very expensive wine/liquor/etc with bad label versus an empty pristine bottle. So what if you have a pretty non-functioning part...... just my opinion.
edit: Sorry What I really mean is: save your Slot A's even if you have to drill a hole, crack plastic, whatever it takes to keep it working !

Yeah, in my opinion as well. I don't collect broken parts. Sounds like my 1 GHz T-bird is valueless now and I'm not the kind who lists broken stuff on eBay. Was just wondering what those CPU collectors might value it at.

Concerning cracking the plastic - what's the chance that there is a broken PCB part though? Seems like a greater probability that the CPU itself fried.

If it worked, bad plastic or not, it is worth a lot due to the rarity. It could be a burned cache chip too, that could prevent boot on any Slot cpu. I would risk opening it.. but that is just me.

Horun wrote on 2020-05-07, 01:55:

If it worked, bad plastic or not, it is worth a lot due to the rarity. It could be a burned cache chip too, that could prevent boot on any Slot cpu. I would risk opening it.. but that is just me.

On Thunderbird Slot A CPUs, the cache is integrated onto the CPU die, unlike the Orion Slot A's which have it on the PCB.

Oops my bad. Yes forgot about that. Guess this lock down stuff has effed up my memory (or just old age). Had a service call at a clinic today, the manager and two staff were there but no one else because of the "emergency only" protocol. They were very happy to see me, like I was the first person in days that they had seen outside of direct family or was not an emergency patient. I know them fairly well because I usually see them many times a month and have for years. Just felt very odd and gets more odd as the days turn to weeks and now months..

My only iteration is with the cashier at the grocery store.

EDIT: Were there not any Socket A to Slot A adapters? That way we could run a 1.4 GHz Thunderbird on our Slot A boards.

feipoa wrote on 2020-05-07, 02:42:

My only iteration is with the cashier at the grocery store.

EDIT: Were there not any Socket A to Slot A adapters? That way we could run a 1.4 GHz Thunderbird on our Slot A boards.

Yes there were, I had one at one point and pricewatch had stacks of 1.2ghz slot A processors for sale (presumably slotketted) around 2002.

I am told they are extremely rare and extremely unstable, makes me sad mine probably got recycled during the great downsizing of 2010