PS3 Frankenstein PHAT PS3: CECHA with 40nm RSX

[Supermaripro]

Ok so I retry today and It didn't not give MMI0 access error so I trained syscon with 40nm EB command... No luck, tried with EC command.. No luck so either board /or reball not working.. I will try another

 

[Pacorretaco]

Hi paco I have made 60nm cok 002 ,

I have also try another 40nm on a cok 002 the reball i think is very good, but when I try syscon training I'm getting an MMIO access error? Is this bad other component? View attachment 40537 View attachment 40537

Good job on that 65nm COK 002. Happy to hear that.

Those MMIO access errors are probably because the address is wrong, w 194... probably a typo.
Should be something like
w 3254 21 EB

But remember the training data only matters when you have RSX:FLEXIO_ID error. If you have some other error, you need to find it first

Good luck

 

[Supermaripro]

Good job on that 65nm COK 002. Happy to hear that.

Those MMIO access errors are probably because the address is wrong, w 194... probably a typo.
Should be something like
w 3254 21 EB

But remember the training data only matters when you have RSX:FLEXIO_ID error. If you have some other error, you need to find it first

Good luck

Thanks paco I have made 2 x cok002 with 65nm, but I have failed 2 x 40nm cok002 but I'm so enjoying doing this I have to stop for a couple of days as I have burned my eyes from heat of reballing machine .. I will report back in a couple of days

 

[vyktormvmpay25]

W 194 didn't write nothing so don't worry about it, that address is not accesibile via uart on CXD model. Only SW models can write on addresses from 0x00 up to 0x13ff so nothing was writing on 0x194.
Via uart we can w/r from 3000 to 7400 only on CXD models
Only access from 0x00 to 3000 is with buspirate when ic is out of board.

Sent from my M2007J20CG using Tapatalk

 

[Supermaripro]

W 194 didn't write nothing so don't worry about it, that address is not accesibile via uart on CXD model. Only SW models can write on addresses from 0x00 up to 0x13ff so nothing was writing on 0x194.
Via uart we can w/r from 3000 to 7400 only on CXD models
Only access from 0x00 to 3000 is with buspirate when ic is out of board.

Sent from my M2007J20CG using Tapatalk

It's all good. I got both 60nm cok002 working, it's only the 40nm I'm having some issues with but I will get one done next week 100%

 

[vyktormvmpay25]

With 40nm rsx if they get glod is a big chance, not really sure, they are in low fw? Test with sb uart debugging for fw version.
If 3034 and 4002 something from setup 50/50 chances, but can be cell need some attention as well.


Sent from my M2007J20CG using Tapatalk

 

[Supermaripro]

With 40nm rsx if they get glod is a big chance, not really sure, they are in low fw? Test with sb uart debugging for fw version.
If 3034 and 4002 something from setup 50/50 chances, but can be cell need some attention as well.


Sent from my M2007J20CG using Tapatalk

I did have glod with the 40nm.. on 1st of the attempts, firmware was 4.89... Interesting information to know though. Thanks

 

[vyktormvmpay25]

Well considering under 3.40 fw it may not work to display. We forgot and I made one unit DECHA00A. I didn't see image but sb uart log was telling me that fw is running. Then it was on 3.10 or something like that. I went for dex fw 3.55 in fsm mode, blind installed . All went fine.
I can tell not always glod is a good reball. I did find removed rsx ram and reball all 4 rams will work. When sb is telling software is running, there may be a ram that have bleed issues or pooped without we can see.
So condition is it must work or another rsx if you don't have cnc to scrap those rams.

Sent from my M2007J20CG using Tapatalk

 

[Supermaripro]

Update, woke up did another 65nm 1st attempt , decided to retry a 40nm, took it from a superslim that was working and BOOM now successful with 40nm, both fully working on 1st attempts, very pleased

 

[vyktormvmpay25]

Well done for progress.

 

[Supermaripro]

Well done for progress.

Thanks, I got this on lock now

 

[azlox9]

Hello,
I have a motherboard from a 2004 slim, 45 nm RSX.
CXM4024R encoder is dead, but RSX is likely fine.
Anyone interested in it to salvage the ship ?
I'm in the EU, drop me a PM if you are.
Thanks.

 

[yanvbraz]

So I'm trying to learn all of the procedures of the Frankenstein mod and I watched one of the Computer Botter streams. I would like to know if someone can explain what are these resistor seen in "PS3 CECHA01 90nm to 40nm RSX Swap Frankenstein Backwards Compatible PS3" YouTube stream (can't link here), from 1:33:34 to almost 1:40:00.... it seems that they being resoldered to ground points and not attached to anything at all? They are the resistors put around the orners of the RSX chip. If I were to buy a "RSX itself" and don't salvage it from another motehrboard, I assume they don't come with the chip itself and I need to salvage it from another board? Thanks in advance.

 

[Mates1500]

I just recently discovered this whole crazy journey through Felix's video in attempt to see what's wrong with my friend's PS3 Fat (CECHC), it still boots up, but occasionally, the entire screen freezes and corrupted green squares pop up, I assume it's the RSX going bad, not quite to the point of YLOD, but quite close.

To get to the point, the frankenstein mod seems like the most feasible long term solution, we found a decent repair shop that can transplant the RSX itself, I have an acquaintance that can do some resistor resoldering.
However, if we want to do the proper VDDR voltage mod, we need a spare BD3504. Does anyone know if the slim boards use those, since my friend is getting also a cheap working slim to do this, or are we only left with voltage regulator hack option? The slim is already on its way, but I'm still quite curious before I delve into it all.

Anyways, thank Felix and everyone else involved for the insane amount of work and research behind all of this!

 

[RIP-Felix]

Im partially responsible of what i publish in wiki and i dont feel confident in publishing that, sorry, it looks more like a pareidolia

What makes me think this is very unprobable is what i mentioned before


It would mean there is (at least) 8 revisions of the CXD5300xxxx series. Even considering some of the intermediate models im mentioning could be prototypes of rarities that was not produced massivelly, 8 is too much

 

[Octal450]

Yes, it real and not fake, I am not blind you know!

 

[RIP-Felix]

For the information effort, a better photo always makes vetting information more concrete. While I trust Josh, I understand where @sandungas is coming from.

 

[Supermaripro]

Hi for anyone who's interested I have a Cech H06 90nm rsx white Japanese ps3 with glod and replaced the rsx with a 65nm worked flawlessly

The guide for cok002 resistor mod is exactly the same for the dia-001 motherboard - confirmed today - photos available if needed for updating the guide by @felix

 

[truemaster]

ylod returned to me after having ps3 inactive for 8 months and i get syscon error 3034.
edit i assembly it same thing ylod. i hit hot air with hair dryer and it worked again. that sounds like bumb gate issue. but victor put 40nm gpu how can it be?

 

[Mates1500]

An update to my previous post, I couldn't really find a spare BD3504 on the slim board KTE-001 no matter how hard I tried.

More importantly, we tried to do the Frankenstein mod on a COK-002 but no success so far. Resistors have been modded as per Felix's tutorial, the RSX has been transplanted with the CXD5301A1GB model and I rewrote the EEPROM through syscon as noted, voltage mod not done yet to ease up diagnostics, which was a good call. I can't get the console to turn on.

As noted, I tried using both EEPROM training set variants `w 3254 21 EC` and `w 3254 21 EB` (and fixing the checksum afterwards), but neither seems to work.

ofst[ 92]:err_code:0xa0404002, clock:0xffffffff
ofst[ 96]:err_code:0xa0403034, clock:0xffffffff

ofst[100]:err_code:0xa0404412, clock:0xffffffff
ofst[104]:err_code:0xa0403034, clock:0xffffffff

I get the code 4002 with EB and 4412 with EC. From what I've gathered, 4002 usually means wrong training set, and 4412 that something is wrong with the RSX on a hardware level. My friend paid a non-miniscule sum for the job to a guy with a proper BGA rework station, does that mean he just did not do the job properly, or is there another explanation? The RSX chip was harvested from a perfectly working slim unit.
One curious thing is that even though the 3034 failure is usually called YLOD, I've never actually seen a yellow light while turning on the console, it's just red.