PS3 Frankenstein PHAT PS3: CECHA with 40nm RSX

I hope I've been clear about when I'm making guesses on things here.

Also, I like that you guys are all like "let's solve the mystery!" and I'm sitting here like "This doesn't help me repair things better, so I don't care."
 
RIP-Felix said:
I created a Poll here in case anyone has had a personal experience reballing the 90nm RSX. Please tell us how long the console lasted.
Click to expand...
RIP-Felix said:
BTW: I tend to agree with @DeadEnd and @squeept that there is a bump issue. But that's a belief I have only circumstantial evidence to backup (ATM). I would feel better having direct evidence. IDK about all the other comments, I haven't really been paying that close attention TBH. Kinda busy.
I keep hearing we have no "proof," as if it's our obligation to prove BumpGate. Which is something I don't think will ever be achievable (proof).

First the burdon of proof doesn't fall on us to prove BumpGate was a thing, the "evidence" wasn't provided by us, it was provided by Nvidia during a lawsuit they lost! It's a forgone conclusion that Nvidia produced defective chipsets during the time the RSX was produced. And at the same fab (TSMC). What isn't a forgone conclusion is that the is the RSX was also affected.

Is there circumstantial evidence, yes. Is that "proof"? No. Is there proof for most things? No. So how does the world function without all this proof no one seems to be able to find? It's because humans don't need proof to function. They choose to believe in either what they want to (confirmation bias) or in a "preponderance of evidence" that makes something to be more likely true, than not (reason). And they choose to abandon reason when emotionally charged. This psychological hack is exploited to great effect by politicians.

As a scientist by trade, I know it's important to identify these biases and mitigate them in the experimental design of the study. To limit the effect they have on the results.

As I said, we don't need to prove BumpGate. We just need to show it's likely, that there's an open question. Because if there's any reason to believe the 90nm is not as good as a 65nm or 40nm, then the frankenstein mod is clearly better. That's not even a contested topic...
  • The 65nm and 40nm produce less heat and less strain on the BGA. So for that reason alone it's better. Even if the Bumps weren't a thing.
  • SONY was able to remove the IHS from the 40nm entirely, going with direct die contact because it didn't need the IHS for stiffening anymore. We could leave it off too. That makes changing the paste easier. Another reason frankies are better.
  • Less heat overall in the case means less wear on heat sensitive components.
  • PSU runs cooler and more efficient, since it doesn't have to deliver as much power. Same story with the VRM. They see less current and heat = they last longer. Even tokins don't see as much ripple on the RSX, because the switching noise from buck converters is proportional to the current they deliver. Since the 65nm and 40nm require less current, the noise the tokins see is less. = tokins run cooler and last longer.
Those are facts.

So even if there isn't a bump issue, the entire point is still valid. Frankenstein Phat mods are superior to reballing. I'm not saying reballing is useless, I'm just saying you don't have to take the chance the bumps could be bad. And you get all these other upsides. If you have to reball anyway, why not frankie it? There's only a few more trivial steps.

Because you have to salvage a working 65nm or 40nm from a potentially working or easily repairable no-BC model? That's about the only fair criticism I've heard. And it'll be true when NOS 40nm RSX's dry up. But even then, there are so many more non-BC models that even if you had to kill the 4.6 million working slims to put 40nm on every BC model produced (you can't because lots of BC models were irreparably destroyed), then you would barely make a dent in the 28+ million 40nm RSX installed in slims alone. Or the 21+ million 65nm. Besides, we will never Frankie that many BC models. Not to sound crass, but BC lives matter more. F the slim!

We don't have to agree, but @Pacorretaco makes some useful counterpoints that caution against over simplifying the issue and assuming too much. That's useful to me, because when I present BumpGate theory in the video, it would be prudent to acknowledge the unknowns, so that people can't simply dismiss me as another fear monger pushing Bumps like the tokins were 2 years ago. If people have a reballed console or OG working one, they shouldn't fee discontent. Like they have to go out and prevent this from happening. I just need to be careful I'm not feeding that hysteria (paranoia, as he calls it).

He contends that simply mentioning the bumps will likely cause people to start murdering their slims to get a 40nm RSX and then murder a BC PS3 attempting a mod they have no business attempting in the first place. I contend that I don't want to leave out a legitimate subject "I believe" is an important part of the story. Yes it makes for a good narrative, and yes it may only be a story. But I don't believe it's hogwash and it's my video.

I'm not responsible for idiots with a heatgun that will kill their console trying something they don't understand. I warned them not to! If they disregard the warning, that's on them. They shouldn't believe everything they read or see on the internet anyway.

Even if I am trying my best and have built a measure of trust among the community, at the end of the day you still need to decide for yourself how to best diagnose and repair your consoles. Don't just take my word for it! There are no shortcuts!

The preponderance of evidence at this time and IMO tips toward BumpGate theory. If we're being objective, the only evidence I've heard Paco, Alty, and victor hugo alverez offer is that Alty is a very experience repair tech with satisfied customers. And that proves reballing works. Because if it didn't they would be unsatisfied.

I get it. Does it count for nothing...hell no! It's visually impressive. I'm sure it convinces lots of gullible customers. People used to care about that kind of thing, because back in the day people left honest reviews.

No offense intended, but I could make the argument that is BS and we need to hear it from a random sample of actual customers. Like a survey that called them up and asked if their PS3 kept working and how much time they've put on it since having it reballed. I could poin to the fact Alty doesn't have becounts to prove his claims that consoles lasted that long after a reball. But we've only had the SYSCON diagnostic for what..2 years? That's noit fair. I can point to the fact people don't like being blackballed for leaving bad reviews and instead of giving bad ones, they just don't (response bias). It's a quid pro quo system and everyone knows it. But that's not fair either.

These are not a practical request. It's criticism, but it's not constructive.

See, the burdon of proof can be applied both ways. I can say Ratings are BS. Prove to me that the consoles kept working, then dispute whatever you say next. And we can go back and forth forever never getting anywhere.

Have I asked you to believe me because I'm the great and powerful Felix? No! It's officious when people tells us to believe them credulously. Instead I present my theory and what led me to that conclusion. Then engage productively in the debate. If I'm wrong, I'm wrong. I value that feedback because my goal is the truth, not being right.

We can move the needle forward when we set our egos aside.

All this, "prove it's the bumps"..."no, you prove it's the BGA" stuff is nonsensical and won't lead anywhere but to a toxic counter productive culture.

Let's instead challenge our assumptions and see where that takes us. And in the future when I post the video presenting bump theory, you can take it with a grain of salt if you believe I'm wrong. Just don't say I'm trying to mislead people, because I have a body of work that proves the exact opposite. Very few people here have manned up to being wrong when presented with demonstrably false information like I have. That's my modus operandi. Blunder in, get corrected.

I don't claim to be an authority. Just a hobbyist interested in figuring this out. SONY or Nvidia could come in here and just just tell us what happened. But that don't. So instead of blaming each
other and back biting, why don't we instead place the blame where it belongs. With the multi billion dollar companies that left our beloved consoles out to die so they could upsell us the next one that isn't backwards compatible!
Click to expand...

@RIP-Felix Man you're amazing, you seen much more scientific and logical, much much much more than Richard Dawkins, You're an Amazing Scientist Felix, you're much more scientifical than lots of Scientists that I see, i've Reach scientifical thinking due to Inteligent Designe theory even said is pseudoscience but for me Evolution is pseudoscientific.

So I've Been listening to lots of talks by big Scientists and Felix you've seen for me as an extremely impressive scientist, superior in knowledge to much academic Scientists regardless of their theory of choice or believes so it's Epic to have you on Psx-place.

Edit: i'm so bad at language, Really bad not only cause of my english being my second language but because im bad overall at language but im extremely great at logic so I shall be usefull at least for something small on PSX-PLACE.
 
Last edited:
DeadEnd said:
Full on off topic . I'm done explaining common sense to you and your mind tricks. You can think what tf you want , keep your b$llsh$t to yourself. It's because of people like you this community is a toxic environment . I'm done with this .
Click to expand...

@DeadEnd Bro all respect to you too, Youre a super cool Dude, and your vídeos and details Repairs are Great, Great Quality, Great voice!! But why are you so Angry? Yes maybe Alty was geting out of respect a bit but dont mind, you dont need to "be done with this" i'm sorry if I was disrespectful, I think no but sorry anyways.
 
Last edited:
Can we all just agree on data collection, though? That's science, and that helps us all fix shit. Should we start a new thread that is just posting an agreed upon spreadsheet after every repair and nothing else? Google docs?

edit:
 
Last edited:
Hey Felix Did you said on your Vídeo that PS3 Firstly Rsx had only on Cok-00xs had an Problem of Bump Being Destroyed by excessive Current on its Units Bumps, and the problem was fixed by sharing the Power current or voltages across multiple bumps and reducing the Current on individual Bga Bump and by That reducing the Power and heat on individual Bga and resulting in avoiding of Bumps being burn ed and destroyed?? Or you said that happen to other gpus and not PS3?? (I cant remenber) But I can A sure That Never Happen to PS3.
 
PS3 Coks nearly 100% of the time always come back to life after an Reflow on Rsx so logically absolutely the Rsx had never an Problem of its Bumps Eletrical Engineering being dumb designed and taking more Eletrical Power than what its body individual micro Bga or bump could take and being destroyed by excessive power, that never happen.
 
Victor Hugo Alvarez said:
PS3 Coks nearly 100% of the time always come back to life after an Reflow on Rsx so logically absolutely the Rsx had never an Problem of its Bumps Eletrical Engineering being dumb designed and taking more Eletrical Power than what its body individual micro Bga or bump could take and being destroyed by excessive power, that never happen.
Click to expand...
Of course they are going to come back to life, as the heat from the reflow is going to cause materials to expand and move, it only takes the slightest amount to cause a bad bump to reconnect for a perioid of time. How long? who knows.

Its the reason the oven method/hairdryer/heat gun works on all the "defective" 7000,8000 and 9000 series nvidia chips. it is also why it only lasts on the short term.

For example, the 8600M GT infamous for this. slowly looses PCIE lanes over time, until it can no longer interface with the PCIE bus, blast it with 150C of air for 10 mins your back upto 16 lanes on the short term, until it slowly starts to loose them again.

Not saying BGA issues dont happen, they absolutly do. but its not BGA nearly 100% of the time.
 
Hello,
Today, I completed fpr the first time, Frankenstein PS3 CECHA-01 with 40nm CXD5301 RSX.

unknown.png

Screenshot_1.png


Special thanks @RIP-Felix for his amazing guides and @Computer Booter for the RSX.

Kind Regards,
Josh
 
RIP-Felix said:
EDIT: Had to remove due to legit oversights I didn't catch before release. Lets call it a failed soft release. I'll get back to it eventually, but ATM I'm gonna have to cool it. I rushed it because I wanted to get it done before summer was over and I won't have as much time to work on it. Oh well, it wasn't ready for the masses. It'll just have to wait until it is.
Click to expand...
Okay, let's try this again...
 
I'd never seen like half of the videos you found for this. Great work! But that was a really long way to say "the chips suck, lead free sucks, and the caps are probably fine."

Should have called Rossman a twat, though.

You mention the "death grip" of the misconception, and I knew that was going to happen, so that may have contributed to how much of an asshole I was early in the TOKIN thread. I REALLY wanted to get ahead of it before it was too late, so I wanted to absolutely stomp on any misinformation.

edit: I'm sold, 2982 65nm Franks are getting priced as if they have a 90nm and only a one year warranty from now on.

edit edit: I'm still laughing at your timing and choice of one of my comments to scroll in the background where I basically called everyone idiots.
 
Last edited:
Or.... another in between price with an in between warranty. That sounds right. Lower TDP is always good for everything.
 
Last edited:
squeept said:
Or.... another in between price with an in between warranty. That sounds right. Lower TDP is always good for everything.
Click to expand...
Ahh, I would not really worry about the 65nm, certainly not because of the video...
Felix did a great job, though as he rightfully says, especially about that part... mostly are educated guesses and speculations, all open to discussion.

I think you can relax. In the real world they survive just fine... Just as good as the 40nm if not more.
Why else would Sony have been confident enough to make them all run hotter (not cooler) by design?
https://www.psx-place.com/threads/syscon-fan-settings-coordinate-graphs.31188/
They clearly werent scared of any possible "underfill defect" that might react to temperature. Because they did the opposite.
And werent they right indeed when they are mostly all still surviving today?

There is not much that suggests there may be anything wrong with the 65nm. If anything, the opposite.
If people havent found yet the 90nm RSX with "broken bumps"... It will be a while after that, to find them in the 65nm.
 
Pacorretaco said:
Why else would Sony have been confident enough to make them all run hotter (not cooler) by design?
Click to expand...
It's also important to remember that just because the fan tables allow the temps to go higher, that doesn't mean the chips actually do. I haven't done any temps testing on the 65nm models to see, but recently someone (I forget who) posted their temps with a 65nm and it didn't even break 65C at minimum fan speed.

I do have an L model and intend to do some tests with it at some point. Donn't have time ATM. I just was suprised, because from looking at the fan tables you would think they run hotter.
 
RIP-Felix said:
It's also important to remember that just because the fan tables allow the temps to go higher, that doesn't mean the chips actually do. I haven't done any temps testing on the 65nm models to see, but recently someone (I forget who) posted their temps with a 65nm and it didn't even break 65C at minimum fan speed.

I do have an L model and intend to do some tests with it at some point. Donn't have time ATM. I just was suprised, because from looking at the fan tables you would think they run hotter.
Click to expand...
Yes, if you happen to have an L model... The best would be to do your own tests and see, with standard settings from SONY.

And yes you are right that the fan curves alone dont tell the whole story. For example the 90nm RSX also doesnt necessarily go that far from 65c in the real world either. Even unmodified machines.
But the fan curves are still very significant.
65nm machines have higher targets for RSX but also for CELL. And in the end it does make a difference in real temperatures too. The fan will wait longer.

Again, how much this actually matter for the life of the machines, is very questionable too. That was the main point
 
Pacorretaco said:
Again, how much this actually matter for the life of the machines, is very questionable too. That was the main point
Click to expand...
Exactly. At least as far as the 65nm RSX is concerned, I don't think heat would have comparible results to the 90nm failure mode. Bump cracking is not the theory for how 65nm's fail. Delamination of the dies stacked dielectric layers is. Without the poliamide stress layer to protect it from the stress transferred there by using High Tg underfill (which protects the bumps, but shifts that strain the joint between underfill and die).


Assuming the following the chip is defective. But whether or not the defect is exploited and actually leads to faults is the question.
  • Assumption 1 = No PI layer, like G9600 GPU's released around the same time.
  • Assumption 2 = High Tg underfill
  • Assumption 3 = High-Lead bumps
  • Assumption 4 = 65nm RSX produces sufficient heat to strain the Underfill/Dielectric layer's and exploit the defect.
The stage for a defect is set. But whether or not all the pieces came together for the 65nm RSX hasn't been proven. Would need to slice a chip and have an electron micrograph taken of the bump/die attachment site, to see if they did leave off the PI layer. And then we'd need to somehow match the underfill chemistry used to show it's high Tg. And bumps to show they are High-Lead. Then we'd need to simulate the TDP of the 65nm chip using CAD software to see if the heat produced could exploit the defect. And lastly, we'd need real world examples of failed chips.

By design, clearly speculation and conjecture is all we've been allowed to do. This is the result of positioning and strategy. The cost of such analysis is prohibitvly expensive. Nvidia doesn't release this data, they aren't required to as a matter of public record. They only release what the courts decide is necessary, insensitive data, during the course of a law suit. Or what little is discovered by investigative journalists with a moderate budget to have samples analysed for news articles. And that's only when theres enough smoke to investigate a fire.

Nvidia, TSMC, IBM, ATI, all of them say it's proprietary information, even though those materials are used by entire industry, nothing special. So such claims are really about limiting liability, when someone like us claims they screwed up. We can't prove it unless we had access to a silicon failure analysis lab. And there's no consumer protection agency that investigates such claims AFAIK. Even if there were they'd be in the industries pocket. The FDA or DEA... most government prorection agencies are. The only time their findings work to consumers advantage is when it also works to industries advantage, or when the offense is too agregious to ignore (like the opoid epidemic, or salmonella outbreaks, or toxic paint making kids sick...that sort of thing). And even then it's reactionary. Meaning they don't try to stop it before it occurs, they give them a slap on the wrist after it happens and then the government and industry goes on about it's merry business. Meanwhile the damage is done, tough luck.

We're just talking about video games here. It's not like people's houses are burning down. But still. We only know of this problem because there was a significan enough problem with the 90nm RSX to take notice. Then the investigations into other, more defective chipsets gave us insight into a Nvidia's normally closed design principals. And only as a result of a law suit and outrage that spurred investigations by news outlets willi g to spend some cash to generate the data (do the expensive science).

The 65nm RSX didn't generate as much smoke, so we don't know theres fire. Not because there isn't any, but because no one has botherd to look. It's expensive to do so.
 
PostalDude__ said:
Of course they are going to come back to life, as the heat from the reflow is going to cause materials to expand and move, it only takes the slightest amount to cause a bad bump to reconnect for a perioid of time. How long? who knows.

Its the reason the oven method/hairdryer/heat gun works on all the "defective" 7000,8000 and 9000 series nvidia chips. it is also why it only lasts on the short term.

For example, the 8600M GT infamous for this. slowly looses PCIE lanes over time, until it can no longer interface with the PCIE bus, blast it with 150C of air for 10 mins your back upto 16 lanes on the short term, until it slowly starts to loose them again.

Not saying BGA issues dont happen, they absolutly do. but its not BGA nearly 100% of the time.
Click to expand...

You Dont Get what I said, I was talking about an Observation or fact that Felix was Talking About, it was not about Cracked Bumps or Cracked Bgas But it was about Gpus that had its Bumps turn into Ashes due to bad Design in Eletrical Power Means, one Single Micro Bga/one single Bump would take too much power and then it would be destroyed by excessive Eletrical Current and Burn to Ashes.

And I said of Course PS3 Dont has that problem cause in absolut majority PS3 just come back to life after an Reflow, Detail I do Reflows on PS3 since I was 15 y.o at 2012.

You have to read Everything we said if you wanna argue.
 
You must log in or register to reply here.

Similar threads

Back
Top