PostalDude__
Member
Broken bumps can easily lead to VRAM problems, since all it takes is for a bump to go bad on a memory signal line and boom ,you got artifacting, its what commonly happend to GPU's that massive memory busses
Pacorretaco
Member
This is basically the same argument that was used to try to "prove" the Tokin story.
"These capacitors are failing all the time in all kinds of devices... Look at all those Toshiba laptops!
They are "defective". It's an "industry known" fact!
They were so bad, the company behind them ended up going out of business!
Etc, etc...
So obviously they have to be defective in our boards too. Let's get rid of them, and we'll get rid of all the problems.
Yeah... Maybe.
As I'm sure you know, is not so simple...
PostalDude__
Member
execpt there is proven and backed evidence on the whole underfill issue, to the point where nvidia had to go and release revised chips for laptop GPU's as they ran so hot, they'd die within their warranty period. the 8600M GT for example, you could see it slowly failing as it lost PCIE lanes over time
DeadEnd
Senior Member
Accusations that I put the idea into his head as if he has no brain of his own...
I am not even sure what I am supposed to argue about. I'm not dismissing anything, but you keep going on about how something isn't right about the idea and that because some of us agree on it - it's supposedly a harmful thing to say.
Well, if you don't want to look at the evidence, then what is your argument exactly?
I'm not trying to prove anything to anybody. I never said "all chips are defective". Some 90nm chips can last better than others, but many will YLOD. Felix has covered that part better. I won't go digging into every explanation right now. In the end the evidence seems to suggest that many chips are defective. It's a logical conclusion. I'm not forcing you to believe that, I'm only sharing opinions and information. It's not my dark agenda to get people to join my side. There is no secret society of 90nm deniers and bumpgate believers. Everybody is free to think what they want. Everybody is also free to agree with you if they wish so. I never cared to "promote" that all 90nm are evil to the people all over the world. The fact that Felix decided to change his mind is his own business. You do realize these links and other information have been out there long before I brought it up to him... In any case, If you consider sharing an opinion and logical conclusions as a "danger" then that's your problem. Because I don't see it like that.
Last edited:
truemaster
Member
guys you do realize that we argue for a gpu ics that are nearly 20 years old. yes thay have a design flaw and most of them will suffer premature death. thats all. furher analisis and argument about what is the specific problem wont cure those dead ics and this threat is not made for this analisis. to anyone who still have gpus still working from that era congrats. the other can swap to more reliable ic ex 40nm (if its a console) also i have a friend with cechb he bought it second hand from usa in 2010 its factory sealed. he ignore my advice to do some cleaning repaste and this machine is still working exept laser issues is not reading even a cd. well thats all
Pacorretaco
Member
Hmm, Okay.
But then how come are we still seeing these kinds of problems even in 40nm?
And not as rare as you may want to think.
Wasnt the "defect" already "fixed"?
Not just naturally on slim and superslim, but also on frankenstein COK, where the chip has almost no chance of "getting hot" or anything. Even with a very new chip and an official rework.
I dont have to dig too much to come up with this stuff. Is literally the first post of this wonderful thread for example.
The official 40nm Frankenstein from Sony Japan developed similar issues. But obviously that must have been a fluke.
Apparently now is me who doenst want to look at the fatcts and has no argument.
I dont want to turn this into a petty argument so I will not answer more to stuff like this.
Of course we all know Felix is a grown person that can come up with his own conclusions. I dont see any accusations. I know is not even your own idea either.
RIP-Felix
Senior Member
This is in your mind. No one is saying that. Cognitive dissonance. This is just your personal struggle to come to the same conclusion.
It took me awhile to believe it too. So I feel your pain brother!
No one wants to believe this. It's not comforting. If anything, I wish we were wrong. But these chips do have bad underfill. That's a fact.
What isn't a fact, and I'm on your side about this, is the idea that defective underfill explains why 90nm RSXs fail. You are correct that VRAM dies, VDDIO balls on the BGA tend to crack, FBVDDQ balls too. VDDC dies from electromigration (both open and short faults). And how many of those faults are actually due to the underfill vs just normal wear and tear that affects all processors is unknown. That's the point you and @squeept were making. And its a good one.
But the point @DeadEnd, @PostalDude__, and I are making is that when you reball a 90nm RSX you are replacing the BGA, but not the bumps. The bumps also "reflow" and can reestablish a broken connection. But as we all know reflows are inferior and prone to cracking sooner. So the reball may appear tk have worked, but because the underfill is still defective, the bumps will eventually fail again...and sooner.
So if, AND ONLY IF, you need to reball (3034/4xxx), it make more sense to replace the 90nm RSX with a NOS 40nm RSX instead of reballing a 90nm with defective underfill. Now that we know how to instruct the SYSCON to bit calibrate the 65nm and 40nm RSX, this is a no brainer. Why the hell not?
I whole heartedly agree! Keep fighting the good fight brother!
ElGris
Senior Member
Apart of filling the thread of unnecesary feline confrontations, I think this whole deal is irrelevant as @truemaster said, a 90nm has 16 years old, and if it's still alive in a fat PS3 (A/B/C/E/G/H) it's because it survived all the issues these consoles always had, the minimum percentage. It's a total waste of time arguing which was the reason of such high percentage of failures, having also in hand tokins and bgas. How do you know how many were due die bgas, tokins or pcb bgas? Which are the main cause of YLOD. Remember that these consoles were overheating directly from factory, allowing the temps go to +75 ºC in some cases. It's a whole soupe of crappy design with Sony and Nvidia as cheffs. And yet, people discusing the truth after 16 years, for what? Noone has the real answer.
Old bgas on fat will fail with time? YES
Tokins? YES
RSX? It could be if it isn't dead already.
CELL? Anyone heard about dead CELLs? I only read @vyktormvmpay25 saying that on super slims.
There are BCs still alive and running good, I have a few, but also have a fats cemetery (most of them already touched previously), but the funny thing is that most SS I bought for repair that had a 40nm RSX were all good with this chip, and were nicely sold. My OG 2013 SS is my main since always and is still smiling.
The main issue here was always temperature. Overheating took out of business a lot of PS3, and not only because of Nvidia.
P/S: Also, IMO, the BCs I bought are still running because those owners were casual gamers, by seeing what games they played. It's not the same having the PS3 turn on for hours watching netflix, as playing medium/heavy games. BCs with healthy tokins are something rare, and if you have one that was used during years with games, a tokins replace will be needed in no time.
Old bgas on fat will fail with time? YES
Tokins? YES
RSX? It could be if it isn't dead already.
CELL? Anyone heard about dead CELLs? I only read @vyktormvmpay25 saying that on super slims.
There are BCs still alive and running good, I have a few, but also have a fats cemetery (most of them already touched previously), but the funny thing is that most SS I bought for repair that had a 40nm RSX were all good with this chip, and were nicely sold. My OG 2013 SS is my main since always and is still smiling.
The main issue here was always temperature. Overheating took out of business a lot of PS3, and not only because of Nvidia.
P/S: Also, IMO, the BCs I bought are still running because those owners were casual gamers, by seeing what games they played. It's not the same having the PS3 turn on for hours watching netflix, as playing medium/heavy games. BCs with healthy tokins are something rare, and if you have one that was used during years with games, a tokins replace will be needed in no time.
Pacorretaco
Member
Thats great. If it was in my head and nobody was saying it, then DeadEnd was right and there was no argument. Perhaps also had a point about the crazy part too.
And to the people that see these last 3 pages as offtopic, I'll ask them to think or read again because I was just trying to answer a question which I think is relevant here:
What issue exactly are we avoiding when we put different RSX?
Because yeah the 90nm one might be "inferior". But by how much? And why?
Perhaps more importantly, how significant can this actually be for the end user?
I dont want to put Squeept in difficult position but he explained it very well:
Now that there are new options, he has "altered his threshold for discarding a 90nm chip", for practical reasons. That's great.
But he still stands by his well rounded estimate that in a machine that has failed he expects about:
33% problem under RSX
33% problem with RSX
33% any other problem
Lets assume thats accurate for a moment (which very well might).
That leaves at most 33% chance that when the board failed, the RSX is actually dead. This is for non working machines, and maybe including ones that may have been previously opened etc.
But then the bad RSX may have also failed for more reasons other than the "bumpgate"/"underfill" story too.
Internal shorts, VRAM problems, "electromigration"...
Again I dont have a proper statistic for that, but I'm pretty sure is not that close to 100% at all (of the previous slice). I'm not even convinced it will be over 50.
So yeah I can "believe" it too. They may actually have the bad underfill or something. But this is not the full story.
Especially for those who have surviving machines that are still working fine, worrying about this specifically may be not so important. And is not 2007 anymore... They are also way past their original life expectation. They dont need to survive the same battles as they did before either.
So actually, is not really like that I think. If anything, we have now reached a point where we can actually deal with most of the problems that these machines can have. Even the previously unfixable ones like a truly dead RSX. There is no need to worry so much about that now.
You're right. Thank you
I've ditched the idea to get a working BC unit from japan for now. Need to think and also to read a bit more about it.
I keep going on and off this thread, last time I checked it was about 2 months ago and look how far the whole thing went, again.
Very amazing.
I found the particular cxd9208 exchange the most interesting. The PS2 playback issue on early models is getting more common I think, and I'm so glad that we know which part is faulty now. The question is why, and if it can be fixed somehow without taking the chip from the donor.
I had this fault myself on the Japan unit and the PS2 playback only worked at higher temperatures, around 70+ on CELL or RSX, so I assume it's something about the soldering pads on cxd9208? Not really sure I'm not a technician at all..
Can someone explain it better for me? Would be awesome to revive even more cechas with new RSX and solid cxd9208, and keep these system alive for as long as possible
Very amazing.
I found the particular cxd9208 exchange the most interesting. The PS2 playback issue on early models is getting more common I think, and I'm so glad that we know which part is faulty now. The question is why, and if it can be fixed somehow without taking the chip from the donor.
I had this fault myself on the Japan unit and the PS2 playback only worked at higher temperatures, around 70+ on CELL or RSX, so I assume it's something about the soldering pads on cxd9208? Not really sure I'm not a technician at all..
Can someone explain it better for me? Would be awesome to revive even more cechas with new RSX and solid cxd9208, and keep these system alive for as long as possible
sandungas
Developer
Im not sure is is getting more common but im curious too, in this forum there was already several people mentioning it, we are not sure if is always that chip though
One of the problems we have to try to figure it is his pinout is unknown, as squeept said some posts ago by knowing the voltages used by it we could get an initial idea of how much are "stressed" his internal circuits... initially we should assume it works with a low voltage but dunno, also we need to keep in mind is related to video so probably there are some high frequency signals passing throught it (this could be another cause of internal circuits fryed by the normal use... you know the most "flytime" it had the most probable is it went burned)
Is interesting what vyktor mentioned about the BGA balls having the same diameter than syscon... so is not like they used a weaker technology (when compared with syscon), is the same stuff and syscon doesnt have this kind of problems
I said that looks weak, but mostly because i think is located in a area of the motherboard more prone to wrapping... but im not completly sure about this either
The point is... all motherboards wraps a bit with the time (is like the materials gets adapted to a specific tension), and when we remove the heatsink sometimes it happens that the thermal paste dryed and grown some "adhesive" efect... so we need to pull from it with force and this could wraps the motherboard a bit
If that wrapping happens under the CXD9208 then yeah... a few BGA solder balls could break (it could be just one or a couple)
Thats why i suggested to do some pressure tests on it... but this kind of pressure tests are not definitory
It works if the broken ball is located in the peripheral (and if you press in a angle into that border)
But if the broken ball is located at the center then the pressure test is not going to be able to "reconnect" the broken ball because is surroundedd by a lot of balls that handles the mechanical deformation... lets say... you are trying to press in the center but the pressure is spreaded all along the other balls (that resists the pressure well without any deformation)
What you mentioned is interesting though... if it works when the console is "warmed up" thats probably because all the materials expanded a bit (micrometers) and the broken ball reconnected just by that small expansion of materials that happens naturally
Sadly, after desoldering the CXD9208 there is no way to see visually if it had broken balls, see the photo uploaded by vyktor some posts before... all the balls looks "meh" (a few of them was looking a bit darker than the others), but is hard to tell because all them was melted and had big amounts of flux
Last edited:
Computer Booter
Member
Felix brought up an interesting point, I just did a Frankie that had amazing temps on both cell/RSX. I never delid the RSX but I do delid the CELL when necessary. My question is what would happen to a board I ultrasonic clean that hasn't been delidded? Would they then need to be both delidded? I'd assume USC would destroy the paste under IHS. If so I'd rather not ultrasonic boards unless I delid both RSX and cell. That would mean I'd start delidding RSX's off board since it's easier for me to do and also less risky to the board and RSX. I would also be delidding cell after "Franking" then ultrasonic cleaning and resealing both.
Also, what is the preferred brand for RSX thermal glue?? Haven't resealed an RSX yet and not sure what thermal epoxy to get…. Suggestions on that? [mention]squeept [/mention][mention]vyktormvmpay25 [/mention] [mention]RIP-Felix [/mention]
Also, what is the preferred brand for RSX thermal glue?? Haven't resealed an RSX yet and not sure what thermal epoxy to get…. Suggestions on that? [mention]squeept [/mention][mention]vyktormvmpay25 [/mention] [mention]RIP-Felix [/mention]
vyktormvmpay25
Senior Member
I really can't find something that is similar to Arctic AluminaThermal Adhesive. Since 2018 I had to stick with that AGThermoglue.
I wanted to approach as factory, at least for rsx but most good thermal epoxy are from mg chemicals. Not easy to get them here so I can't really say what else to suggest. I will apply for this reply one reference link but I have to read and compare all specs.
I wanted to approach as factory, at least for rsx but most good thermal epoxy are from mg chemicals. Not easy to get them here so I can't really say what else to suggest. I will apply for this reply one reference link but I have to read and compare all specs.
Isn't Epoxy a complete mess-up in terms of a possible re-visit, if the new owner decides to re-apply some new paste at some point in the future?
Wouldn't do some basic thermal paste like MX2 instead of epoxy on the RSX-Ram chips do almost the same?
Maybe it drys out quick, fill the gap and may prevent the RSX-PCB from warping too?
vyktormvmpay25
Senior Member
AGThermoglue is easy to get removed from both rsx and cell. On epoxy adhesive is what factory did. For silicone I have not found any with thermal proper documentation for w/cm cubic.
I'm using thermal silicone AGThermoglue on both.
I just like simple things and can not bother with expensive things anymore as I did those investigations over years, I'll probably obtain same results.
I will just reply with my mini tutorial for that when I'm asked what I do on all. For the moment is just fine, if anyone think of improving my work can do it. I will just share my way, don't need to follow if someone believe is wrong.
https://s.go.ro/93xitgz7
I'm using thermal silicone AGThermoglue on both.
I just like simple things and can not bother with expensive things anymore as I did those investigations over years, I'll probably obtain same results.
I will just reply with my mini tutorial for that when I'm asked what I do on all. For the moment is just fine, if anyone think of improving my work can do it. I will just share my way, don't need to follow if someone believe is wrong.
https://s.go.ro/93xitgz7
truemaster
Member
victor in case delit its needed again. how to do it?
Had a look on the pictures, he linked in the last posting...
Very nice method
from my point of view may be better than the factory epoxy method.If I see it clear, instead of epoxy he use AGThermoglue, not on the ram but on two alternative points... so RSX delidding is possible and easier I guess.
@vyktormvmpay25 Thanks for the pics
vyktormvmpay25
Senior Member
Is just long painting knife in those sets we buy for delid, also same time use ipa on blade will pass easy.
Also for everyone advice and safety working with board on jig support to avoid flexing.
Similar threads
-
-
PS3 PS3 CECHA RSX Swap (40nm) – JP Limited – SYSCON Patch, VDDC Mod – Need Detailed Help- Started by jw0710
- Replies: 4
-
-
