PS3 Frankenstein PHAT PS3: CECHA with 40nm RSX

[Squiglemouse]

Wow wow so first of all amazing job. The results are looking amazing. It seems like the 90 nanometer CPU might be a sticking point though, is it theoretically possible for us to replace the cell with a smaller process size? Coming until recently in the last 2 years it wasn't general knowledge that this was even theoretically possible on the RSX. Anyhow I'm going to go ahead and @squeept so that hopefully we can start getting some information to him too. Maybe if we're lucky he'll start offering this as a service at least we can dream haha.

 

[Pacorretaco]

Wow wow so first of all amazing job. The results are looking amazing. It seems like the 90 nanometer CPU might be a sticking point though, is it theoretically possible for us to replace the cell with a smaller process size? Coming until recently in the last 2 years it wasn't general knowledge that this was even theoretically possible on the RSX. Anyhow I'm going to go ahead and @squeept so that hopefully we can start getting some information to him too. Maybe if we're lucky he'll start offering this as a service at least we can dream haha.

No way to change CPU with smaller one. But fortunately, there's no point!
Because the CPU is not a problem unlike the RSX. They are not a natural failure point.

Remember this isn't something you do just to get lower temperatures.
You do this when your RSX is beyond help.

If you just want a quieter system, there are things you can do that are way simpler. Even with a 90nm RSX, quiet is possible.

Cheers

 

[RIP-Felix]

Let me first say this is a breakthrough for all PS3's. This will allow repair shops to source chips from other motherboards across almost the entire productline (except the super slim). Before, you would have to harvest a 90nm from a brooken backwards compatible motherboard. However, RSX faults are so common, that there are more dead RSX chips than there are good ones on bad boards. Increasingly so as time goes on. So this mod breaks opens up the majority of the PS3 product line to harvest chips from!

I don't mind sacrificing a slim model to revive a backwards compatible model! There are more of them out there, and if it makes the BC as reliable as a slim, without the worry of experiencing a YLOD again, that is nothing short of GAME CHANGING!

But before this can go mainstream, someone needs to offer this as a service. Even for intermediates this is too difficult.

For example, yesterday I attempted to swap a 40nm from a KTE-001 (3001A model) to PS3#2 (CECHA01 motherboard). After much trouble to get the solder balls attached to the RSX, it was able to solder it on, but it only attached one side! It was lined up right, but there must have been a bridge underneath, because one side settled down and the other side lifted up! When it cooled I could slide a piece of paper under one side! Haven't seen that one before! Seriously, I must be running out of ways to get it wrong by now!? I mean, it's comical at this point...haha! Once I think I got it, something new comes out of nowhere to slap me back to noob.

I removed it this morning and tried again, but failed 3 times to get the solder balls to attach to the RSX. They moved, wouldn't attach to the center of the pad, or stuck to the stencil when I lifted off. Then I had to pick and place balls with shaky hands, causing more balls to shift. SOOOOO TEDIOUS! Once I'd get them in place they squirmed when I tried to heat affix them...Grrr...Ahhh!

Its just not going to work with my equipment!

So ordered a direct heat stencil that's smaller. It won't fit in a jig, but I can tape it to the RSX and heat directly so the balls can't move out of position. That is my biggest problem! This 40nm still reads good resistance, so I haven't killed it yet. I'd rather buy them with balls attached!

 

[vyktormvmpay25]

Let me first say this is a breakthrough for all PS3's. This will allow repair shops to source chips from other motherboards across almost the entire productline (except the super slim). Before, you would have to harvest a 90nm from a brooken backwards compatible motherboard. However, RSX faults are so common, that there are more dead RSX chips than there are good ones on bad boards. Increasingly so as time goes on. So this mod breaks opens up the majority of the PS3 product line to harvest chips from!

I don't mind sacrificing a slim model to revive a backwards compatible model! There are more of them out there, and if it makes the BC as reliable as a slim, without the worry of experiencing a YLOD again, that is nothing short of GAME CHANGING!

But before this can go mainstream, someone needs to offer this as a service. Even for intermediates this is too difficult.

For example, yesterday I attempted to swap a 40nm from a KTE-001 (3001A model) to PS3#2 (CECHA01 motherboard). After much trouble to get the solder balls attached to the RSX, it was able to solder it on, but it only attached one side! It was lined up right, but there must have been a bridge underneath, because one side settled down and the other side lifted up! When it cooled I could slide a piece of paper under one side! Haven't seen that one before! Seriously, I must be running out of ways to get it wrong by now!? I mean, it's comical at this point...haha! Once I think I got it, something new comes out of nowhere to slap me back to noob.

I removed it this morning and tried again, but failed 3 times to get the solder balls to attach to the RSX. They moved, wouldn't attach to the center of the pad, or stuck to the stencil when I lifted off. Then I had to pick and place balls with shaky hands, causing more balls to shift. SOOOOO TEDIOUS! Once I'd get them in place they squirmed when I tried to heat affix them...Grrr...Ahhh!

Its just not going to work with my equipment!

So ordered a direct heat stencil that's smaller. It won't fit in a jig, but I can tape it to the RSX and heat directly so the balls can't move out of position. That is my biggest problem! This 40nm still reads good resistance, so I haven't killed it yet. I'd rather buy them with balls attached!

Yes added some photos with direct heat stencil and preparation for reball rsx. In 4 years never killed any 65mm or 40 nm rsx.
https://www.psx-place.com/threads/reball-ps3-cell-rsx.32376/page-2#post-290650

What power of preheater do you have on rework station? what do you use for top heating?

 

[DeadEnd]

You must put extremely thin layer of flux so that balls will not shift.

At least on my system, games work fine. I haven't tested it for 1 hour though, just 15 min. Cell is not actually going that hot after all (room temperature is 24 C)

Some screens (Ofc smearing happened)

https://drive.google.com/drive/folders/1k24gX0pYQxZ-rn5vqm9etiDINn21ETqZ?usp=sharing

 

[RIP-Felix]

You must put extremely thin layer of flux so that balls will not shift.

Tried 4 time in total, each time spredding it as thin as I could. One time the flux that went over the edge wicked back up onto the balls near the edge and they swam. I had to wipe the whole thing off and start over.

I tried the thinnest layer you could hope to achieve, then I tried heating it to smooth out the surface before placing balls. They still swam. Some would snap to neat rows, others would stick to the side of it's pad, not centered! I could move them manually with fine forceps while holding my breath, but with hundreds of tiny balls and shaky hands I'd rather pay someone else to do it. I'm hoping the direct heat jig solves this problem. Otherwise, I'm giving up on reballing.

Maybe it's the flux that came with my cheap reballing kit. At this point I'm going to have to buy more solder balls because the vial they sent me only has enough left for one more attempt. I've probably tried and failed 10 times now. BTW here's my ghetto setup...

The preheater and 853D Hot air wand with 45x45mm BGA Nozzle work great. 2hrs at 100C is effectively drying the boards. I can tell if it needs more drying if it's taking longer to preheat to 180C (meaning evaporating liquid is holding the temps back). I just let it go longer at 100C, but 1-2hours is working well. As I said before, the preheat (not to be confused with drying) is to 180C, I apply flux before turning on the top heat @240C. It takes about 5minutes to get up to flowing temp. I can verify by nudging the chip, which springs back by the cohesion of the now molten solder balls.

Reflowing is perfect with this setup. Topside and bottom side temperature only vary by 5C, so I'm not woried about warping anymore. I can't perfectly simulate a reflow profile with this setup, but it's adequate to get the job done with experiance. And it didn't cost $550, like the cheapest BGA rework stations.

If I had chips with the balls already attached it wouldn't be a problem at all. I can easily reflow using this setup, I just can't get the GORAM solderballs to line up neatly on the RSX!!! I really hope the direct heat stencil solves this problem.

 

[RIP-Felix]

Some screens (Ofc smearing happened)

That smearing was your camera moving during the exposure, not the console right?

 

[marciolsf]

Tried 4 time in total, each time spredding it as thin as I could. One time the flux that went over the edge wicked back up onto the balls near the edge and they swam. I had to wipe the whole thing off and start over.

I tried the thinnest layer you could hope to achieve, then I tried heating it to smooth out the surface before placing balls. They still swam. Some would snap to neat rows, others would stick to the side of it's pad, not centered! I could move them manually with fine forceps while holding my breath, but with hundreds of tiny balls and shaky hands I'd rather pay someone else to do it. I'm hoping the direct heat jig solves this problem. Otherwise, I'm giving up on reballing.

Maybe it's the flux that came with my cheap reballing kit. At this point I'm going to have to buy more solder balls because the vial they sent me only has enough left for one more attempt. I've probably tried and failed 10 times now. BTW here's my ghetto setup...
View attachment 33019
The preheater and 853D Hot air wand with 45x45mm BGA Nozzle work great. 2hrs at 100C is effectively drying the boards. I can tell if it needs more drying if it's taking longer to preheat to 180C (meaning evaporating liquid is holding the temps back). I just let it go longer at 100C, but 1-2hours is working well. As I said before, the preheat (not to be confused with drying) is to 180C, I apply flux before turning on the top heat @240C. It takes about 5minutes to get up to flowing temp. I can verify by nudging the chip, which springs back by the cohesion of the now molten solder balls.

Reflowing is perfect with this setup. Topside and bottom side temperature only vary by 5C, so I'm not woried about warping anymore. I can't perfectly simulate a reflow profile with this setup, but it's adequate to get the job done with experiance. And it didn't cost $550, like the cheapest BGA rework stations.

If I had chips with the balls already attached it wouldn't be a problem at all. I can easily reflow using this setup, I just can't get the GORAM solderballs to line up neatly on the RSX!!! I really hope the direct heat stencil solves this problem.

You almost have me convinced to try this. How much (ballpark) did you pay for the preheater? I was already planning on buying the air gun, and I think I have pretty much everything else.

 

[vyktormvmpay25]

Tried 4 time in total, each time spredding it as thin as I could. One time the flux that went over the edge wicked back up onto the balls near the edge and they swam. I had to wipe the whole thing off and start over.

I tried the thinnest layer you could hope to achieve, then I tried heating it to smooth out the surface before placing balls. They still swam. Some would snap to neat rows, others would stick to the side of it's pad, not centered! I could move them manually with fine forceps while holding my breath, but with hundreds of tiny balls and shaky hands I'd rather pay someone else to do it. I'm hoping the direct heat jig solves this problem. Otherwise, I'm giving up on reballing.

Maybe it's the flux that came with my cheap reballing kit. At this point I'm going to have to buy more solder balls because the vial they sent me only has enough left for one more attempt. I've probably tried and failed 10 times now. BTW here's my ghetto setup...
View attachment 33019
The preheater and 853D Hot air wand with 45x45mm BGA Nozzle work great. 2hrs at 100C is effectively drying the boards. I can tell if it needs more drying if it's taking longer to preheat to 180C (meaning evaporating liquid is holding the temps back). I just let it go longer at 100C, but 1-2hours is working well. As I said before, the preheat (not to be confused with drying) is to 180C, I apply flux before turning on the top heat @240C. It takes about 5minutes to get up to flowing temp. I can verify by nudging the chip, which springs back by the cohesion of the now molten solder balls.

Reflowing is perfect with this setup. Topside and bottom side temperature only vary by 5C, so I'm not woried about warping anymore. I can't perfectly simulate a reflow profile with this setup, but it's adequate to get the job done with experiance. And it didn't cost $550, like the cheapest BGA rework stations.

If I had chips with the balls already attached it wouldn't be a problem at all. I can easily reflow using this setup, I just can't get the GORAM solderballs to line up neatly on the RSX!!! I really hope the direct heat stencil solves this problem.

As long as you can desolder in good condition one 65mm or 40 nm rsx you can solder it back even easier. Once out remove ihs as I do. Find any small aluminium plate 4mm 10x10cm, attach one thermocouple with foil on that plate and use 50° on bottom. Plate should have like 70 after few minutes,
Start hot air with medium tube nozzle and 20 percent of air at 210°. First will move on each area at distance of 3cm about, each place 3~5 seconds. Once cumulated temperature will show for your bottom plate 120, start keeping hot gun in each place 10~15 seconds until you see balls will start melting.
First process is kind dried, you have applied from beginning only a thin on rsx. Once plate will show it is cooling apply second stage of flux with one paint pencil that won't let behind any tiny parts as hair. Once reached again 120, keep again hot air gun on each place and you will start see right shape of melted alloy. Once cooled at about 100, remove stencil and you can see all perfect soldered. Don't cross to much from 135. Stop or move rsx from plate, or stop hot air /move aside.
If you see all in perfect line you can cross more flux over all while rsx at 80 and is another way to see if all balls are in place.
I'm pretty sure you will get this easily.
Oh if any ball is unsoldered, just put back stâncil, this time, not necessarily to stick it with foil, just place back on preheater and heat same process for that ball. Sometimes you may have to push it down because some pads are oxidised .
Edit
Remember rsx 40 /65nm ihs best remover will be old Gillette blades. They are the only thin blades I have atm.

 

[ThePolyesterPimp]

So uhhhh, how much are y'all charging to swap the RSXs?

 

[DoublesAdvocate]

@Ph4nton @feel2death No, not quite. The 28nm RSXs use a completely different pad layout/design and are therefore incompatible, there may be some versions of the 40nm RSX that are incompatible but I don't think we've determined that yet.

Wow wow so first of all amazing job. The results are looking amazing. It seems like the 90 nanometer CPU might be a sticking point though, is it theoretically possible for us to replace the cell with a smaller process size? Coming until recently in the last 2 years it wasn't general knowledge that this was even theoretically possible on the RSX. Anyhow I'm going to go ahead and @squeept so that hopefully we can start getting some information to him too. Maybe if we're lucky he'll start offering this as a service at least we can dream haha.

No chance, the CELL is married to several other components on the board which are completely incompatible. There's also no history of Sony ever doing that either.

 

[DeadEnd]

That smearing was your camera moving during the exposure, not the console right?

Of course it was the camera, do not fear . The console is running fine so far. I could even record a stream video from me gaming on it if you have doubts.

About the balls, hmm. My method is slightly different than Victor's. Hopefully you aren't using that hot air stand and the gigantic tip I'm seeing for the reballing? I could only get it to work when I'm holding the handle with my hand at an angle and with a medium-sized nozzle, then slowly moving it from there. Not even sure what angle exactly (you must have seen my video?), but I noticed that the side where the screw is holding the tip cannot face the the balls. Might be that some air is escaping through it, not sure. You have to at first test it out by moving it veery carefully to the edge and see what happens. The moment some balls get messed, you must stop because something's wrong. Fix them up and try a different angle or your air isn't hot enough yet.

I'm using Kingbo Rma-218 from aliexpress, it smells bad but gets the job done. The temperature on the hot air station is 340 C. I couldn't get them to melt when it was less and it's not actually that hot at the distance i'm holding it. It won't instantly get hot enough so you must let it heat up for some time in the beginning, and only then move it closer to the balls. It sounds like you're not using enough heat. It has to create some type of a heatwave attack with instant melting temperature.

Yeah, and I still used 0.55 mm balls. Someone said it would cause issues because the RSX would then sink lower. Well, hasn't caused any issues for me. The heatsink is clearly doing its thing since the temps on the RSX are fine. 0.05 mm is probably negligible. Although I can't claim it makes reballing easier either, because I haven't compared to how it goes with 0.6 mm (My 0.6 mm vials had an expiration date?)

 

[feel2death]

@Ph4nton @feel2death No, not quite. The 28nm RSXs use a completely different pad layout/design and are therefore incompatible, there may be some versions of the 40nm RSX that are incompatible but I don't think we've determined that yet.

Yeah i think 40nm from super slim are the limit of swaping since the video we share here are he use from super slim one

So uhhhh, how much are y'all charging to swap the RSXs?

They charge me 50 $ here in indonesia including the rsx

 

[RIP-Felix]

You almost have me convinced to try this. How much (ballpark) did you pay for the preheater? I was already planning on buying the air gun, and I think I have pretty much everything else.

Around $100. One problem though is that I had to buy a 15A rated power cable, as the one that comes with it is only rated for 10A and got hot after only a few minutes. After I got the 15A cord, however, it's been a great purchase. Just big enough to heat the phat PS3.

...they charge me 50 $ here in indonesia including the rsx

Must be nice! I'm jelous! I wonder what it'd cost to send a console half way around the world? Would it arrive in and make the return trip intact? I imagine customs and taxes are a nightmare they would rather avoid. Maybe I should plan a trip to Jakarta!

Customs: "Why are you traveling with so many PS3's sir?"
Me: "E-sports?!"

I think at some point we need a middle man to procure mod chips and distribute to vetted installers that do quality work. People like @squeept (USA) or @vyktormvmpay25 (Europe). Sounds like Indonesia has it covered...lol! Problem is these guys already have more work than they can handle, not sure if they can/want to take on more. I don't want to speak for @squeept, but I imagine the same reason he doesn't accept reball repairs is the same reason he won't be offering this as a service. High chance of the swap not going well and people accusing you of stealing or breaking their console. Probably only worth it to sell pre-modded consoles.

Just noticed there was some new activity here. @sandungas @M4j0r @littlebalup @DeViL303 or anyone else with some credentials that wants to take a crack at figuring out a 65nm CECHA01, just send me the two boards needed (make sure the 65nm is known working, but the COK-001 can be YLOD) and I'll swap the RSX and send back for free.

Does this offer still stand? I have a Sealed A01 with 2s YLOD that I haven't diagnosed yet. I did pull a 40nm from a 30xx that I would prefer to use, but if it doesn't work I'll settle for the 65nm from a working 20xx I don't mind sacrificing if I have to. But I would prefer scavenging from the dead, instead of the living.

 

[DeadEnd]

You obviously wouldn't need to send the whole console, just the board. Surely there is some type of a reballing service still operating in US/Canada?

I could contribute to the cause as well and if push comes to shove, I could lift some RSXs from slims for the greater good. Last time I bought a dead slim console for only 10 euros (oof, inbefore the word gets out on what we're doing here ). Sending over a tiny package is easy as well, so it shouldn't be that pricey (depends on the sellers ofc). I can place the balls on the chip myself (So @RIP-Felix if all else fails, I can help you). Buying RSX from Aliexpress is a gamble... All I need is to figure out a few kinks with how to improve the fume extractor so that the smell doesn't flood the whole damn place.

Edit. I MIGHT even be able to perform the whole thing for a fee in case somebody desperately needs it (I'm in EU too). But I am not quite at the professional level like Viktor is. I bet @squeept and @vyktormvmpay25 would be first choices either ways. I could come in third, as a potential "backup".

Edit 2. I read that lead solder is banned in EU. Last time I ordered from China, the vials still arrived. Not sure will they keep arriving if I order more...?

 

[Ph4nton]

Edit 2. I read that lead solder is banned in EU. Last time I ordered from China, the vials still arrived. Not sure will they keep arriving if I order more...?

the European Union has decided to ban everything that has lead in its composition. Everything that has RoHS you are sure there is no lead in its composition, lead is harmful to health, but we all know that lead-free soldering is sucks.

You can continue shopping, the European Union just doesn't need to know about that purchase.

 

[DoublesAdvocate]

the European Union has decided to ban everything that has lead in its composition. Everything that has RoHS you are sure there is no lead in its composition, lead is harmful to health, but we all know that lead-free soldering is sucks.

You can continue shopping, the European Union just doesn't need to know about that purchase.

The worry over lead free solder is overblown, the main fault with the RSX seems to be an issue with the chip itself, not necessarily the solder underneath it.

 

[Sampsonay]

Tried 4 time in total, each time spredding it as thin as I could. One time the flux that went over the edge wicked back up onto the balls near the edge and they swam. I had to wipe the whole thing off and start over.

I tried the thinnest layer you could hope to achieve, then I tried heating it to smooth out the surface before placing balls. They still swam. Some would snap to neat rows, others would stick to the side of it's pad, not centered! I could move them manually with fine forceps while holding my breath, but with hundreds of tiny balls and shaky hands I'd rather pay someone else to do it. I'm hoping the direct heat jig solves this problem. Otherwise, I'm giving up on reballing.

Maybe it's the flux that came with my cheap reballing kit. At this point I'm going to have to buy more solder balls because the vial they sent me only has enough left for one more attempt. I've probably tried and failed 10 times now. BTW here's my ghetto setup...
View attachment 33019
The preheater and 853D Hot air wand with 45x45mm BGA Nozzle work great. 2hrs at 100C is effectively drying the boards. I can tell if it needs more drying if it's taking longer to preheat to 180C (meaning evaporating liquid is holding the temps back). I just let it go longer at 100C, but 1-2hours is working well. As I said before, the preheat (not to be confused with drying) is to 180C, I apply flux before turning on the top heat @240C. It takes about 5minutes to get up to flowing temp. I can verify by nudging the chip, which springs back by the cohesion of the now molten solder balls.

Reflowing is perfect with this setup. Topside and bottom side temperature only vary by 5C, so I'm not woried about warping anymore. I can't perfectly simulate a reflow profile with this setup, but it's adequate to get the job done with experiance. And it didn't cost $550, like the cheapest BGA rework stations.

If I had chips with the balls already attached it wouldn't be a problem at all. I can easily reflow using this setup, I just can't get the GORAM solderballs to line up neatly on the RSX!!! I really hope the direct heat stencil solves this problem.

I got that blue POS and yeah i see why you don't like it. Just went through half of my solder balls and ive never gotten them to fall into place quite right. They're super close, but i cant risk this. Will be trying to do that sanding mod mentioned earlier.

Also, do you or anyone else have any tips for cleaning solder balls? Once they get the tiniest amount of flux on them, they stick together and cant be re-used. Im not too keen on buying more when im running through them so quickly.

 

[Ph4nton]

I got that blue POS and yeah i see why you don't like it. Just went through half of my solder balls and ive never gotten them to fall into place quite right. They're super close, but i cant risk this. Will be trying to do that sanding mod mentioned earlier.

Also, do you or anyone else have any tips for cleaning solder balls? Once they get the tiniest amount of flux on them, they stick together and cant be re-used. Im not too keen on buying more when im running through them so quickly.

You can use a desoldering wire. but be careful, in the first tests I scratched the chip with the soldering iron

 

[Sampsonay]

You can use a desoldering wire. but be careful, in the first tests I scratched the chip with the soldering iron

Thank you, but i actually meant the lead solder balls im trying to attach to the chip. They're easiest work with new out of the bottle, the second i drop them onto the GPU and get flux on them, they stick to each other if i messed up at all.

Anyway, I found it easy to work with them again by gently swishing them around with a brush, rather than hoping they roll into their respective hole, because that's just not happening when they're sticky. Perhaps there's a better way to work them again, idk.

Also, enjoy a picture of my current suffering: