PS3 Frankenstein PHAT PS3: CECHA with 40nm RSX

[Workz_777]

Is it just possible to delid keep fan speed to 40% and not perform the 40nm RSX mod as it requires expert skills a donor console for RSX?
Will taking all these precautions keep my COK-02 fine?

Also delidding can be risky too, regardless of which approach you go with, preparation is key:- "slow and steady wins the race".

With practice it becomes easier, but for the first time it can be risky. I wrap electrical tape (many times) around a butter knife tip / end, so it can just fit that through the RSX + IHS gap, then lever it off. This way there is little risk to scratching the RSX substrate under the IHS.

 

[tungatung]

Also delidding can be risky too, regardless of which approach you go with, preparation is key:- "slow and steady wins the race".

With practice it becomes easier, but for the first time it can be risky. I wrap electrical tape (many times) around a butter knife tip / end, so it can just fit that through the RSX + IHS air gap, then lever it off. This way there is little risk to scratching the RSX substrate under the IHS.

TYSM for the advice, but that still dosent answer my ques, will it be ok to do all this and not perform the 40nm RSX MOD

 

[truemaster]

temprature is not electronic friendly. the lowest you keep it the longest will last. personly i will try to keep both cpu gpu at max 55celsious no matter how high the fan will go.

 

[tungatung]

temprature is not electronic friendly. the lowest you keep it the longest will last. personly i will try to keep both cpu gpu at max 55celsious no matter how high the fan will go.

70 degree is the max i go, it is sufficient.

 

[truemaster]

just remember when the temprature reach 75c the default syscon fan control is starting to rump up the fan 70c is very high is nearly as before anything touched-modified

 

[Pacorretaco]

Hmmmm, I've heard this talk before, it's a running joke in Computer Booter's Live Stream chats, so i'm guessing you too are part of the COK Club? What is your Youtube name bro? Let me guess... are you... Jason?

Ahhh, no not really a Jason no. Im just Paco hahaha.

Also I am just random person, perhaps with a weird hobby haha, but I am not running any business or youtube channel like them or offer any services or products. I am not affiliated and I dont know who Jason is haha.

But Cheers anyway!

 

[Workz_777]

Ahhh, no not really a Jason no. Im just Paco hahaha.

Also I am just random person, perhaps with a weird hobby haha, but I am not running any business or youtube channel like them or offer any services or products. I am not affiliated and I dont know who Jason is haha.

But Cheers anyway!

Lol, no worries, but if ever you want to watch @Computer Booter do some Frankies live, he has a Youtube channel and streams for hours most days / nights (PDT time zone). A few of us from here are regulars, like @RIP-Felix, @vyktormvmpay25, and others, here's the link if ever you wanna come see some COKs being manhandled (waka! waka!)
https://www.youtube.com/channel/UCPWknjnQNOwSD15nNh4rhLg

 

[St3v3b82]

Also delidding can be risky too, regardless of which approach you go with, preparation is key:- "slow and steady wins the race".

With practice it becomes easier, but for the first time it can be risky. I wrap electrical tape (many times) around a butter knife tip / end, so it can just fit that through the RSX + IHS air gap, then lever it off. This way there is little risk to scratching the RSX substrate under the IHS.

I've read somewhere in this forum that a delid of the rsx isn't necessary and could maybe bad too if the heatspreader isn't re-glued back. I guess it was a post from @RIP-Felix

Quite interesting but I don't like the idea to reglue the lid back, cuz I don't know how the glue 'works' if it gets hot, maybe the gap between the IHS and the die gets bigger, expands during playing/heating... and I guess an untouched original glued PS3 will get a BGA damage too at some point.

Though I'm really curious if the BGA of a non-glued only repasted RSX will fail faster or not..
Do we have users arround with problems like that?

Best regards

 

[RIP-Felix]

I've read somewhere in this forum that a delid of the rsx isn't necessary and could maybe bad too if the heatspreader isn't re-glued back. I guess it was a post from @RIP-Felix

Quite interesting but I don't like the idea to reglue the lid back, cuz I don't know how the glue 'works' if it gets hot, maybe the gap between the IHS and the die gets bigger, expands during playing/heating... and I guess an untouched original glued PS3 will get a BGA damage too at some point.

Though I'm really curious if the BGA of a non-glued only repasted RSX will fail faster or not..
Do we have users arround with problems like that?

Best regards

It's inferred from my research into thermal design of processor packages. But no, we don't know how much shorter the BGA lifespan will be when the IHS isn't stiffening the package. And it's a subjective test that would be impossable to quantify.

The "glue" you choose to reattach the RSX is important. The thermal epoxy you use needs to have a CTE as close to 25 as possable to avoid warping stresses. I just applied MG chemicals 9460TC to PS3#13 (40nm RSX Frankie). That RSX I delidid before I knew better. It has leaded balls and the console is working fine. But I wanted to keep it that way and try out this Thermal Epoxy. So on it went.

Thing is I sealed up the console and don't intend to open it again until the paste needs changed again (5-7 years). Assuming it lasts that long.

I don't really think 40nm RSX need the IHS epoxy. SONY evidently didn't think so. In 40nm Super slims they don't have one. And they do fine. But the 65nm and 90nm do.

 

[Pacorretaco]

TYSM for the advice, but that still dosent answer my ques, will it be ok to do all this and not perform the 40nm RSX MOD

You may be misinterpreting some things...
"40nm RSX MOD" is not something you do "to get better temperatures"...

I dont understand what you are trying to compare. If your machine is fine, you dont need to do anything. This is not some kind of "upgrade". This is a last resort repair path.

If you are trying to get a "quiet" machine and are considering the 40nm RSX, you may get disappointed too because normally is not about the RSX getting "hot". It is always about the CPU anyway.

In that case, the answer is yes. You can most definitely get OK temperatures without "40nm mod". You dont need 40nm for that.
Standard machine can run cool too, even without modifications. CPU may need delid one day but RSX normally doesnt even need it.

PS:
When you delid both, this is what you normally see:

In a normal machine,
The CPU has lost almost all contact with the IHS while the RSX is completely fine.

What this means in practice, is that the Fan speed is pushed higher and higher by the hot CPU to try to compensate for the poor thermal transfer, while the RSX is OK.
This indirectly makes RSX run even cooler than normal, even in machines that were never opened or maintained. You can check this yourself.
In fact often when people mess with them, is when things get worse, not better especially for RSX. Or like in this case, get very small or no improvement.

 

[ElGris]

My thumb is hurting..

@vyktormvmpay25 about your finding:

 

[squeept]

I don't really think 40nm RSX need the IHS epoxy. SONY evidently didn't think so. In 40nm Super slims they don't have one. And they do fine. But the 65nm and 90nm do.

Keep in mind, the heatsink in the super slims is mounted physically to some stuff that keeps it locked in to place. If you don't use an adhesive of any kind somewhere in that sandwich, *theoretically* the IHS can migrate from heating cycles. It is not mechanically linked to anything.

And the thermal plaster will definitely help to better move heat off of the chip. Anything is better than nothing. Didn't Anandtech show that toothpaste was only a few degrees worse than the cheapest thermal paste?

I'll try the MG stuff again, but I vaguely remember it being just straight up epoxy. Like...you won't be able to service it later because the RAM chips are going to rip off.

 

[ElGris]

The RSXs on SS are directly connected to the main mini block/fan combo via a heat pipe, this improves the heat disipation a lot. It's made of copper as everyone knows, but this is not only a copper bar, I recently discovered that there's something else about this:

https://www.msi.com/blog/laptops-10...esigning-an-efficient-laptop-cooling-solution

Asumming Sony used on BCs and SS these same kind of heat pipes, then it could be the SS has a better RSX cooling overall than the BCs, taking in count everything, fans, heatsinks, IHS, etc.

They didn't use a IHS because it wasn't necessary for this low cost console, it could be better for the cooling aspect.

In the other hand, If I'm not wrong 40nm RSXs need a copper sheet in between die/IHS, because there's a little gap that wasn't in the 90nm RSX. So you need to put something in there apart of the thermal paste, and then epoxy if you plan to let it in just one piece.

 

[Gulag]

I am offering the 40nm swap, join our discord if you'd like to chat about it, disclaimer: it's $500 for the service with a 6 month warranty.
Discord: https://discord.gg/RAfmRC4ZDR

Thanks! I will need to save up a bit for that but I will contact when I am ready

 

[tungatung]

You may be misinterpreting some things...
"40nm RSX MOD" is not something you do "to get better temperatures"...

I dont understand what you are trying to compare. If your machine is fine, you dont need to do anything. This is not some kind of "upgrade". This is a last resort repair path.

If you are trying to get a "quiet" machine and are considering the 40nm RSX, you may get disappointed too because normally is not about the RSX getting "hot". It is always about the CPU anyway.

In that case, the answer is yes. You can most definitely get OK temperatures without "40nm mod". You dont need 40nm for that.
Standard machine can run cool too, even without modifications. CPU may need delid one day but RSX normally doesnt even need it.

PS:
When you delid both, this is what you normally see:
View attachment 36901
In a normal machine,
The CPU has lost almost all contact with the IHS while the RSX is completely fine.

What this means in practice, is that the Fan speed is pushed higher and higher by the hot CPU to try to compensate for the poor thermal transfer, while the RSX is OK.
This indirectly makes RSX run even cooler than normal, even in machines that were never opened or maintained. You can check this yourself.
In fact often when people mess with them, is when things get worse, not better especially for RSX. Or like in this case, get very small or no improvement.

I was thinking that the 40nm RSX will run cooler and also cause of smaller die size will possess fewer risks of failure.
However, if it's okay with the stock 90nm RSX with higher fan speed and deliding, then I wont perform the mod.
Also, are there chances that the CELL can fail as well [90nm], if yes then will deliding and setting high fan speeds save it & keep it in good healthy condition.
My motherboard is COK-02

 

[RIP-Felix]

Keep in mind, the heatsink in the super slims is mounted physically to some stuff that keeps it locked in to place. If you don't use an adhesive of any kind somewhere in that sandwich, *theoretically* the IHS can migrate from heating cycles. It is not mechanically linked to anything.

And the thermal plaster will definitely help to better move heat off of the chip. Anything is better than nothing. Didn't Anandtech show that toothpaste was only a few degrees worse than the cheapest thermal paste?

I'll try the MG stuff again, but I vaguely remember it being just straight up epoxy. Like...you won't be able to service it later because the RAM chips are going to rip off.

I press the IHS firmly into the HS to make a close bond first. That helps limit squirming. But just to be sure I screw the leaf spring tensioners in one twist at a time for even pressuer. Then complete a 2 hour burn in with the console on its side. I let it cool overnight for the gasket maker and paste to cure. This particular eposxy (if used) has a burn in cure time. I forget off the top on my head, something like 80C for 1 hour. It's listed on the product page. But it'll cure slowly over the course of a few weeks with normal use temps. Maybe faster with the burn in.

I don't think delidding again will be necessary, never seen a 1201 error (RSX overheat). And the bigger concern than overheating is BGA stress from warping. Even with lead balls, the gain is elasticity will be outweighed by the greater warmping stress in ea h thermal cycle. The IHS needs to stiffen the package to limit warping stresses and the epoxy is rigid for that reason. Silicone is not good enough. I'd love to know what plaster SONY used, but this MG chemicalsthe epoxy had the closest CTE to 25 and the color was similar. I thought I'd give it a go and see how it does. I'm not necessarily recommending it to everyone just yet.

But we really should be keeping the lid on the RSX. And if not, putting it back. It's just a matter of finging the right plaster/epoxy.

 

[sandungas]

Dont miss this tutorial for COK lovers

I press the IHS firmly into the HS to make a close bond first. That helps limit squirming. But just to be sure I screw the leaf spring tensioners in one twist at a time for even pressuer

Auch, thats a new level in COK modding, just thinking in it makes me feel the pain

 

[Pacorretaco]

I was thinking that the 40nm RSX will run cooler and also cause of smaller die size will possess fewer risks of failure.
However, if it's okay with the stock 90nm RSX with higher fan speed and deliding, then I wont perform the mod.
Also, are there chances that the CELL can fail as well [90nm], if yes then will deliding and setting high fan speeds save it & keep it in good healthy condition.
My motherboard is COK-02

My crystal ball got the YLOD too, so I'm afraid I dont have a definite answer to your question now.

But the good news is, you dont need to know the future either. You just enjoy it while it lasts, like everything. People get problems as they age too.

However I can still give you my guess.
When these machines fail, I dont even think is fair to blame it always on the "heat".
I think way more machines have been damaged than saved by the so called "preventative maintenances". Yes, with successful delid you may make it quieter or cooler, especially the CPU.
But is hard to say if it will actually extend the life of the whole machine at all, and by how much if so. First of all, CPU failures for natural causes are rare. Second, as you saw before, the RSX may likely be cool and fine already and see no improvement at all.
And most importantly though, you also need to understand that is not just about CPU or RSX. These boards are very complex and is hard to say how long they will last because many other things in the board could also fail too.

Even the RSX itself is a complex thing which can fail in many different ways and for different reasons, possibly but not necessarily because of heat. They even has no less than 4 VRAM chips on top, if they already werent complex enough.

Just relax and dont be afraid to deal with the problems "if" and when they come. Is impossible to try to anticipate everything. Also unnecessary. Do what you can do. Get the eid_root_key and dont be afraid. You probably wont even get to see any problem, like most people in the past. And even if you do, is no big deal.

 

[Workz_777]

Dont miss this tutorial for COK lovers

Auch, thats a new level in COK modding, just thinking in it makes me feel the pain

Ahahaha!!

 

[RIP-Felix]

I was thinking that the 40nm RSX will run cooler and also cause of smaller die size will possess fewer risks of failure.
However, if it's okay with the stock 90nm RSX with higher fan speed and deliding, then I wont perform the mod.
Also, are there chances that the CELL can fail as well [90nm], if yes then will deliding and setting high fan speeds save it & keep it in good healthy condition.
My motherboard is COK-02

The CELL doesn't appear to be prone to BGA defects and wasn't affected by Bumpgate. IBM seemed to know what they were doing. That's not to say it can't go bad, it can. It's just not defective like the 90nm RSX is.

Of course the CPU does overheat because of old thermal paste between DIE/HS. So that does need changed. This needs to be done with a good tool to adoug killing the CPU. And the IHS needs to be glued back on (with high temp silicone, not super glue). Otherwise you would be creating a defecitive CPU that would eventually get a BGA defect.

The 90nm RSX will eventually go bad. There is no way to tell how long it'll be. But it's not as simple as replacing a defective RSX with a good one. The heat needed puts everything on the board at risk! It should not be looked at as an "upgrade." It is a "repair." You should only consider it to ressurect a dead console. That's why we call it the Frankenstein mod.