PS3 Frankenstein PHAT PS3: CECHA with 40nm RSX

squeept said:
I was paged? I think you all figured it out already. Looks official to me. I doubt anyone without access to some real Sony documentation on the syscon would have just known to do that. Especially since it's been pretty well documented that we couldn't swap between process sizes, so they would have been putting in a lot of work and parts testing it out on a whim that goes against what has already been documented. There was a reference chart on the BGAmods forum about confirmed GPU swaps that work, but I just looked and that place doesn't appear to exist anymore, but I remember they were all the same process size. Even then I think a few didn't work, but that could easily have been a dead chip or operator error.

Just checked the pictures again....that's almost definitely official Sony work. The IHS looks like it was glued back on with the Sony stuff. If you scrape that stuff off is it stretchy and rubbery? The plaster stuff I use is closer to latex feeling when it dries, instead of hard and a little flaky like the Sony stuff. Some people leave the IHS on when they are doing rework, but since the GPU on the Super Slim doesn't have an IHS to begin with, that was definitely added, so they had to put adhesive on it themselves. If it's exactly like the Sony stuff you're used to, it's official. There is nowhere I know of to buy exactly what they use, or I would have bought it myself.

Doesn't look like there's flux residue everywhere? Check the bottom side of the USB connectors where there is always a huge puddle of flux. That will tell you if they put it through a new cleaning afterwards. The only cleaning methods I know for the general public are fairly harsh liquid chemicals, and then they need baked to dry afterwards (for the full immersion that a BGA requires to clean under), which will leave the freshly exposed and cleaned copper around the edge of the board looking like a rainbow as it oxidizes. I can see in the sticker picture that the copper looks normal, so if it was washed, it was washed by Sony.

Not saying it's NOT official if it wasn't washed or you find the latex stuff. However, if it was EITHER washed or glued with the right stuff, then it definitely IS official.

Great find either way, I might have some boards in my scrap pile that I can bring back to life now. I'll sacrifice a Super Slim to save a CECHA01 any day.

edit: just found a super slim already missing the GPU in my scrap pile, grid is exactly the same on both chips, so I don't think there's any magic going on underneath the chip unless they cut a trace or something. I think that one little resistor is the only change.
Click to expand...

The rsx is from slim not superslim.
 
@squeept But doesn't "artic ceramique 2" dries as the original glue used by Sony? I still have my doubts about this. I mean, anybody can do this, with the proper knowledge, of course but leaving it "that" clean (and also that resistor trick) must be common in a proffesional repair store there in Japan. Maybe they're so pro (or too ninjas) to know that little secret about the resistor.

If you have a rework station would be cool to see if this trick works on mobos with a huge time gap like in this case.

I also found the link that I saw many years ago. it has a little more of info about this, I think @sandungas will find it interesting too if he didn't know about it. Here they put a larger list of processors models found in different mobos, and in the comments there're a few that could fix a mobo with a different RSX, but didn't find anything like in this thread. The forum is in spanish, but it's easy to understand.

http://tecnologos.net/viewtopic.php?f=114&t=2131

Luisile said:
The rsx is from slim not superslim.
Click to expand...
https://www.psdevwiki.com/ps3/images/8/83/MSX-001_-_1-886-928-11.jpg
https://www.psx-place.com/attachments/photo_2020-02-02_19-18-44-jpg.23523/

Both RSXs are the same, phisically. That's because there wasn't a huge gap between 3xxx and first 4xxx models. If the majority of 40nm RSXs work on any old mobo with that trick, it's time to celebrate.
 
ElGris said:
@squeept But doesn't "artic ceramique 2" dries as the original glue used by Sony? I still have my doubts about this. I mean, anybody can do this, with the proper knowledge, of course but leaving it "that" clean (and also that resistor trick) must be common in a proffesional repair store there in Japan. Maybe they're so pro (or too ninjas) to know that little secret about the resistor.

If you have a rework station would be cool to see if this trick works on mobos with a huge time gap like in this case.

I also found the link that I saw many years ago. it has a little more of info about this, I think @sandungas will find it interesting too if he didn't know about it. Here they put a larger list of processors models found in different mobos, and in the comments there're a few that could fix a mobo with a different RSX, but didn't find anything like in this thread. The forum is in spanish, but it's easy to understand.

http://tecnologos.net/viewtopic.php?f=114&t=2131


https://www.psdevwiki.com/ps3/images/8/83/MSX-001_-_1-886-928-11.jpg
https://www.psx-place.com/attachments/photo_2020-02-02_19-18-44-jpg.23523/

Both RSXs are the same, phisically. That's because there wasn't a huge gap between 3xxx and first 4xxx models. If the majority of 40nm RSXs work on any old mobo with that trick, it's time to celebrate.
Click to expand...

yeas they may look the same. but superslim RSX have no IHS. and i dont know if yo can use ihs from other rsx as there maybe is height diference between chip and memory. i dont say it will not work but only rsx from first 4xxx series looks the same. 42xx and 43xx have different rsx.

https://www.psdevwiki.com/ps3/images/f/f4/PQX-001_Board.jpeg
 
squeept said:
If you scrape that stuff off is it stretchy and rubbery? The plaster stuff I use is closer to latex feeling when it dries, instead of hard and a little flaky like the Sony stuff.
Click to expand...
Yep, it's hard and flaky. But I don't think IHS was tampered with before me. As other said, it's probably Slim RSX, so it had it's IHS from the beginning.

squeept said:
Doesn't look like there's flux residue everywhere?
Click to expand...
I saw this nail polish like flux residue in abundance around through-hole elements, but no - there is no such residue around RSX. Well, at least not on such scale.
 
Last edited:
@squeept , I would love to see you have a run at this on cech a01 to see if anything else needs to be done other than a swap or with that resistor he noted. You are the only person I know of here that can do this , but would be very professional about it with pics and all that good stuff. You can even give some of your attitude along the way......lol. Would be cool to see an in depth about though
 
CGRESET is not grounded on SS, don't have a Slim on hand. Original post said SS, so I said SS, deal :P

I got nothin' then. Yeah, anyone can do things right, but..... they don't. The horrors I've seen... Anyway, only thing I'm going on now is that I can't see any logical reason someone would have just thought to do that on their own without knowing what that signal was used for. Just looking at the service manual, there's an SPI interface between the RSX and syscon, so I would have just assumed it asked for a chip ID or something. There's a million ways it wouldn't work, so at the very least they had some inside info that encouraged them that it wasn't a waste of a ton of time.

I don't remember using ceramique, but anything meant for metal to metal contact on a modern processor usually acts more like an insulator when it's in a thick gap, so there's a different set of products. Don't ask me, thermodynamics is black magic. I know heat spreaders work, but if you ask me how, my eyes will gloss over.

I'm buying a working slim right now to keep in waiting, and I'll report back next time I have a CECHA01 that I think has a dead GPU.
 
Last edited:
Give me a post address via pm. I've got a 3k I'll send you that you can rape for parts. It works, only thing wrong with it is it's very slow in taking discs like the drive is jammed up on it. I bought it just to rip the idps from it. Been sitting ever since. Not sure if a 3k will work though as it seems this one was taken from a 2k. Let me know man. @squeept
 
nCadeRegal said:
Give me a post address via pm. I've got a 3k I'll send you that you can rape for parts. It works, only thing wrong with it is it's very slow in taking discs like the drive is jammed up on it. I bought it just to rip the idps from it. Been sitting ever since. Not sure if a 3k will work though as it seems this one was taken from a 2k. Let me know man. @squeept
Click to expand...

I'm an idiot and I just looked on the shelf. I have one that I keep to test drives. I also have a CECHA01 in the drying oven right now, and I'm getting kind of antsy because this would be huge news if there was suddenly a supply of known working chips available. Even at the cost of a full slim system, it's still worth it. I'm sure I could charge more for these monsters, too. I may just go jam the slim in the oven now...
 
Kot1LXa.jpg


Alrighty, minimum 4 hours of bake, an hour to lift and clean, another 4 hours of bake, then an hour to reball and cool. I'll see you folks at 9 or 10 tonight if I don't popcorn it from cutting corners on the bake time. It feels like Christmas! I wanna open my presents!
 
May I ask why you bake them 8 hours? To soften the balls? It's the first time I see that. Well, at least no in a youtube video lol. I wonder if RSXs from 4000 series are fully compatible. I have two boards waiting to be sacrifice.
 
ElGris said:
May I ask why you bake them 8 hours? To soften the balls? It's the first time I see that. Well, at least no in a youtube video lol. I wonder if RSXs from 4000 series are fully compatible. I have two boards waiting to be sacrifice.
Click to expand...

Drive out moisture to prevent delamination/damage/popcorning to the board/chips/components from the high temperatures needed to melt lead free.

If this works, I'll definitely be trying every RSX with the same footprint afterwards.
 
The most awesome thing would be to "transfer" the RSX 28nm from a CECH-43xx to a CECHA
I dont even know if they have the same footprint though, but for the missing IHS there is no problem, you can adapt one from other PS3 model

Something like that would be epic because are the 2 extremes of the "PS3 familly" :D
A CECHA with the RSX of a modern PS3 is a lot better than a simple CECHA... so yeah that frankencecha's are worthy
 
sandungas said:
The most awesome thing would be to "transfer" the RSX 28nm from a CECH-43xx to a CECHA
I dont even know if they have the same footprint though, but for the missing IHS there is no problem, you can adapt one from other PS3 model

Something like that would be epic because are the 2 extremes of the "PS3 familly" :D
A CECHA with the RSX of a modern PS3 is a lot better than a simple CECHA... so yeah that frankencecha's are worthy
Click to expand...
the rsx from 43xx series is way different. it has only 2 memory chips and the main die is not in center. so you can not adapt the ihs. look at my previous post there is image of the 43xx rsx
 
Luisile said:
the rsx from 43xx series is way different. it has only 2 memory chips and the main die is not in center. so you can not adapt the ihs. look at my previous post there is image of the 43xx rsx
Click to expand...

Makes no difference how it looks. You could easily put an IHS on that using Silicon Thermal adhesive and a bit of know-how trickery to make it sit flush with the RSX die, memory and the heat sink of a Phat PS3. Also the mem chips make no difference either as all that been done there is 4 has been compacted down to 2 using better tech.

All that matters is the pin_outs on the RSX, if they are close enough and if it can be done without too much invasive modifications done.
 
Yeah, the reduction of memory chips number to 2 (from the previous 4) doesnt matters for software, this change is not going to create incompatibilities
Also, the fact the superslims doesnt have an IHS in RSX means that memory chips doesnt really needs an IHS

But if you really want to add an IHS in it... i guess the better thing you can do is like what AMD did in some of the EPYC processors that have some dummy DIE's with the only purpose of serve as the base of the IHS
Same stuff... but in this case the dummies should have the same height than the memory chips

And for the IHS dimmensions... dunno, but you can work on it with sandpaper or files to change it
...or take a piece of copper and build your own :D
 
You must log in or register to reply here.

Similar threads

Back
Top