PS3 Frankenstein PHAT PS3: CECHA with 40nm RSX

While I'm waiting for my tantalizers to arrive, I took some resistance measurements on that board and actually wrote them down this time. The chart I referenced is the one by RIP-Felix on this post: https://www.psx-place.com/threads/f...nd-error-reporting.30100/page-115#post-319714

(using the dot as decimal separator and comma as thousands separator)
RSX_VDDIO: 99.22 ohms
FBVDDQ: 382.9 ohms
RSX_PLL: 533k ohms
RSX_VDDC: 2.21 ohms
VDDR: 857.2 ohms
YC_RC_VDDIO: 16.25 ohms
BE_VDDC: 6.45 ohms
VDD_MEM: 18.85 ohms
VDDA: 6.43 ohms
MC2_VDDIO: 15,609 ohms
BE_PLL: 1.817m ohms

Note that this chip is a CXD5301 that was pulled from a slim motherboard that had a dead CPU. Comparing to results from @DeadEnd that he showed in his video at around timestamp 15:40, I can see that some of my measurements are way off. Still am not sure if these measurements are due to a bad or worn chip or poor soldering. I may have to pull this chip back off, reball it again, check resistances on the chip and the board, and try soldering it again. Wish I had an x-ray machine to see what happened underneath there, haha.
 
iComputer7 said:
While I'm waiting for my tantalizers to arrive, I took some resistance measurements on that board and actually wrote them down this time. The chart I referenced is the one by RIP-Felix on this post: https://www.psx-place.com/threads/f...nd-error-reporting.30100/page-115#post-319714

(using the dot as decimal separator and comma as thousands separator)
RSX_VDDIO: 99.22 ohms
FBVDDQ: 382.9 ohms
RSX_PLL: 533k ohms
RSX_VDDC: 2.21 ohms
VDDR: 857.2 ohms
YC_RC_VDDIO: 16.25 ohms
BE_VDDC: 6.45 ohms
VDD_MEM: 18.85 ohms
VDDA: 6.43 ohms
MC2_VDDIO: 15,609 ohms
BE_PLL: 1.817m ohms

Note that this chip is a CXD5301 that was pulled from a slim motherboard that had a dead CPU. Comparing to results from @DeadEnd that he showed in his video at around timestamp 15:40, I can see that some of my measurements are way off. Still am not sure if these measurements are due to a bad or worn chip or poor soldering. I may have to pull this chip back off, reball it again, check resistances on the chip and the board, and try soldering it again. Wish I had an x-ray machine to see what happened underneath there, haha.
Click to expand...

Hmm, some of the resistances can be higher than others . VDDR is quite high though... It almost looks like some of your measurements weren't taken correctly. it's unlikely that Cell has 6.45 ohms and YC_RC VDDIO is 16 ohms... Cell is normally 2-4 ohms and YC_RC VDDIO is 12 ohms. You can test the meter by putting probes together to see what's the resistance of the probes. Also I would try another multimeter.
 
DeadEnd said:
Hmm, some of the resistances can be higher than others . VDDR is quite high though... It almost looks like some of your measurements weren't taken correctly. it's unlikely that Cell has 6.45 ohms and YC_RC VDDIO is 16 ohms... Cell is normally 2-4 ohms and YC_RC VDDIO is 12 ohms. You can test the meter by putting probes together to see what's the resistance of the probes. Also I would try another multimeter.
Click to expand...

I touched my probes together and they measured 2-6 ohms so I switched them out for different ones and I now get about 0.4-0.6 ohms. I took new measurements on the board this time and here's what I got:

RSX_VDDIO: 96.4 ohms
FBVDDQ: 372 ohms
RSX_PLL: 37k ohms
RSX_VDDC: 0.8 ohms
VDDR: 796 ohms
YC_RC_VDDIO: 13.3 ohms
BE_VDDC: 3.3 ohms
VDD_MEM: 16.4 ohms
VDDA: 4.5 ohms
MC2_VDDIO: 3,135 ohms
BE_PLL: 956k ohms

RSX_VDDC seems extremely suspicious to me. I've been suspecting that my soldering didn't hold and I think this confirms it. Going to wait to hear confirmation before I pull the chip back off and try again though. At least it'll be easier this time since I used leaded solder.
 
iComputer7 said:
I touched my probes together and they measured 2-6 ohms so I switched them out for different ones and I now get about 0.4-0.6 ohms. I took new measurements on the board this time and here's what I got:

RSX_VDDIO: 96.4 ohms
FBVDDQ: 372 ohms
RSX_PLL: 37k ohms
RSX_VDDC: 0.8 ohms
VDDR: 796 ohms
YC_RC_VDDIO: 13.3 ohms
BE_VDDC: 3.3 ohms
VDD_MEM: 16.4 ohms
VDDA: 4.5 ohms
MC2_VDDIO: 3,135 ohms
BE_PLL: 956k ohms

RSX_VDDC seems extremely suspicious to me. I've been suspecting that my soldering didn't hold and I think this confirms it. Going to wait to hear confirmation before I pull the chip back off and try again though. At least it'll be easier this time since I used leaded solder.
Click to expand...

This seems more realistic. But yea, under 1 ohm VDDC is very low. , Especially with 0.4-0.6 on the probes. Almost shorted. Let's see if the second time around something changes.
 
DeadEnd said:
This seems more realistic. But yea, under 1 ohm VDDC is very low. , Especially with 0.4-0.6 on the probes. Almost shorted. Let's see if the second time around something changes.
Click to expand...

Second attempt went well up until I went to put it back on. I noticed that the chip was a little bit warped but it had not been an issue until now. On the first attempt it was mostly flat but now it's so warped that only the balls on one corner are actually making contact. I think this chip has gone through too many rework cycles. At least it wasn't a new old stock chip, it was one I pulled from a slim board that had a dead CPU.

It doesn't help that my hot air station doesn't seem to be heating the chips evenly. I should have done this earlier but I'll invest in an infrared top heater and use it for the whole process next time (removing GPU from donor board, reballing it, removing GPU from the CECHA board, soldering donor GPU)
 
iComputer7 said:
Second attempt went well up until I went to put it back on. I noticed that the chip was a little bit warped but it had not been an issue until now. On the first attempt it was mostly flat but now it's so warped that only the balls on one corner are actually making contact. I think this chip has gone through too many rework cycles. At least it wasn't a new old stock chip, it was one I pulled from a slim board that had a dead CPU.

It doesn't help that my hot air station doesn't seem to be heating the chips evenly. I should have done this earlier but I'll invest in an infrared top heater and use it for the whole process next time (removing GPU from donor board, reballing it, removing GPU from the CECHA board, soldering donor GPU)
Click to expand...

No worries, I killed about 5 of them before I started having some kind of success. Also it's important to have a proper jig for the motherboard.

But I always believed that hot air is not easy to use for this task. I am trying to design an upgraded station , the safer version of the one I used in the video. Well, it still implies you would have a T8280 for the bottom, but the top would be using IR with a precise temperature profile... If all goes well, I would eventually finish it and have a prototype.

Btw I put one frankie to ebay for an auction... If interested.
 
Anybody knows a service like computer booter in the EU? I got a CECHB00 with a partly YLOD (works with ps1 and ps2 but ylods after some minutes in ps3 games. Xmb works fine always) now i interested in "franking" her. But shiping a nearly 10kg console to the us would cost nearly as much as the service computer booter would charge on top of it, so a service in europe or the EU would be greaaat.
 
In the EU theres Victor in Romania for some time, and I am in Spain.
Better than shipping around to America for sure!

But CECHA and CECHB are more difficult and risky. Many get other problems (common example PS2 blackscreen), and then it is also difficult to find parts because they didnt exist in Europe either. So it is very hard to give any guarantees for those.

COK-002 are much more reliable design and less risk
 
What about CECHA/COK-001 systems makes it prone to issues after a RSX Frankenstein compared to other SKUs/Motherboards?
Is the PS2 blackscreen an issue with the RSX of 65/40 nm not being able to communicate wit the PS2 EE/GS?

On a side note, is CELL Frankenstein ever going to be a thing? Taking the 90nm CELL to a 45nm one for example?
From what I understand, it would require an interposer like the 28nm RSX.
 
Seraphic said:
What about CECHA/COK-001 systems makes it prone to issues after a RSX Frankenstein compared to other SKUs/Motherboards?
Is the PS2 blackscreen an issue with the RSX of 65/40 nm not being able to communicate wit the PS2 EE/GS?

On a side note, is CELL Frankenstein ever going to be a thing? Taking the 90nm CELL to a 45nm one for example?
From what I understand, it would require an interposer like the 28nm RSX.
Click to expand...

I talked about it in the video, it's unlikely to be a thing... At least for the moment. I mapped out how many connections needed rewiring, and there are just too many of them. It's a lot of work with questionable results. Not to mention other difficulties.

As for 28nm RSX, I nearrly finished the pinout for it too. But... It doesn't have analog inputs so even if we made an interposer and installed the chip onto COK boards, you would lose ps2 games support. In essence, the chip itself has been stripped of PS2 games support (EE/GS cannot connect to it). That's partly how they have been able to reduce the size of it so much.
 
Seraphic said:
What about CECHA/COK-001 systems makes it prone to issues after a RSX Frankenstein compared to other SKUs/Motherboards?
Is the PS2 blackscreen an issue with the RSX of 65/40 nm not being able to communicate wit the PS2 EE/GS?
Click to expand...
We dont really know yet. But theres something dodgy about COK-001. Something about the design that makes it prone to those issues. Doesnt have to be our fault when doing the frankenstein process, as we have seen it in all kinds of machines.

But It's just not normal that so many COK-001 are getting those PS2 blackscreens, while COK-002 are mostly all fine.
For example Felix got PS2 blackscreen even on his first frankenstein. But not just him... Pretty much all others got some as well. Booter got at least 4 or 5, shawn also, Victor a couple too. Botakompong (Aki), ACE... And many others including people that had official frankenstein from SONY as well.

My educated guess is that it seems something related to EE/RDRAM, Some sort of communication issue between them, at least for some of the cases. Or possibly communication with RSX too, but it really is hard to say

Seraphic said:
On a side note, is CELL Frankenstein ever going to be a thing? Taking the 90nm CELL to a 45nm one for example?
From what I understand, it would require an interposer like the 28nm RSX.
Click to expand...
And yeah those are a... dead end:adoration:
 
DeadEnd said:
I talked about it in the video, it's unlikely to be a thing... At least for the moment. I mapped out how many connections needed rewiring, and there are just too many of them. It's a lot of work with questionable results. Not to mention other difficulties.
Click to expand...

Indeed, sounds like a lot of effort for a very old console with smaller and smaller user base.
Does this pin-out connection issue remain for the 65nm CELL as well or only 45nm process?
And in the end, does the CELL even get that hot after a delided (with fresh thermal paste even on 90nm) to make it worth all the trouble?

DeadEnd said:
As for 28nm RSX, I nearrly finished the pinout for it too. But... It doesn't have analog inputs so even if we made an interposer and installed the chip onto COK boards, you would lose ps2 games support. In essence, the chip itself has been stripped of PS2 games support (EE/GS cannot connect to it). That's partly how they have been able to reduce the size of it so much.
Click to expand...

Not surprised they would work to remove EE/GS support after the chip has been long gone from the board itself, but confused about the "analog inputs"? Is that how PS2 games are communicated from the EE/GS <- to the -> RSX? Analog not Digital?
 
Pacorretaco said:
We dont really know yet. But theres something dodgy about COK-001. Something about the design that makes it prone to those issues. Doesnt have to be our fault when doing the frankenstein process, as we have seen it in all kinds of machines.

But It's just not normal that so many COK-001 are getting those PS2 blackscreens, while COK-002 are mostly all fine.
For example Felix got PS2 blackscreen even on his first frankenstein. But not just him... Pretty much all others got some as well. Booter got at least 4 or 5, shawn also, Victor a couple too. Botakompong (Aki), ACE... And many others including people that had official frankenstein from SONY as well.

My educated guess is that it seems something related to EE/RDRAM, Some sort of communication issue between them, at least for some of the cases. Or possibly communication with RSX too, but it really is hard to say
Click to expand...

Well you give me some pause now after purchasing having already purchased a CECHA00/COK-001 that I wanted to have a 45nm RSX installed on. Doesn't the COK-002 remove the EE/GS chip though and perform PS2 using software?
 
Seraphic said:
And in the end, does the CELL even get that hot after a delided (with fresh thermal paste even on 90nm) to make it worth all the trouble?
Click to expand...
The CPU is considered reliable and not "prone to failure." There is no reason to worry or even think about its temperature if it has fresh paste. There is no reliability reason to frankie the CPU. It was always just an interesting challenge. Even if we could make an interposer to adapte it, solve the voltage issues, and remarry it, the interposer would represent a new point of failure that likely would cause the console to be less reliable. So all that work and expense for a worse console. The only reason to do it is for the swag.
 
Seraphic said:
Well you give me some pause now after purchasing having already purchased a CECHA00/COK-001 that I wanted to have a 45nm RSX installed on. Doesn't the COK-002 remove the EE/GS chip though and perform PS2 using software?
Click to expand...
To be clear. There is no reason to believe the COK-001 Frankie will get the no PS2 issue. And it's not caused by the RSX swap. It could be related to the RSX connections to the board, but that could happen with any RSX.

COK-002 is a good console too. Not that it doesn't have it's own subset of drawbacks. It uses partial emulation. It has the GS, but emulates the functions of the EE using the Cell CPU. This leads to a significant loss in backwards compatability. reduces it from the COK-001's 99% to somewhere around 80%. Many of the games that do work can have minor graphical artifacts and inaccuracies. Most of the games it can't play have HD remasters, which are perhaps a better way to play them anyway. So that's not necessarily a deal breaker.
 
Last edited:
RIP-Felix said:
the interposer would represent a new point of failure that likely would cause the console to be less reliable.
Click to expand...
Yep, i always was concerned about that, when thinking about how the pressures are spreaded is obvious that adding another layer on the cake cant make any good, thats for sure

The first time i wrote something about the cell frankies we was not sure how many pads mistmatches, and i was considering also to take advantage of the fact that there is a squared hole in the motherboard at the cell back. so being optimistic, i thought maybe it was posible to "fix" the mistmatching pads around the hole (that are mostly power lines, relativelly easy to mess around with them) in a guetto way with a copper piece, wires etc
And incase there was some more mistmatching pads not accesibles from the hole that needed to be remapped it was posible to rebuild a trace, or mod the motherboard by scratching it to expose inner traces, etc... Of course this tricky pads that required "electronic artist" skills should not exceed lets say... 3 or 4 pads
All this without interposer (and being optimistic that the amount of mistmatching pads was small)

But after we realized there are lot of mistmatching pads (and the only realistic way to remap them is with an interposer) then the story changes a lot
Personally i would not do it, i dont trust in BGA, so adding another BGA layer is against my principles, lol

It was a nice research thought, it was the kind of thing that is needed to investigate before making an oppinion about it
 
can someone help me?
My CECHL00 always blanks black every time I open any program, my current position is CFW 4.85.1 Pro Citra Mulia, I want to upgrade to 4.89 evilnat, but an error 8002f169 appears, it says I have to enable the QA flag, so I install toggle QA, but when opened only black blanks, as well as others like multiman or rebug toolbox, only black blanks, is there any solution?
Sorry for my bad English
 
Seraphic said:
Indeed, sounds like a lot of effort for a very old console with smaller and smaller user base.
Does this pin-out connection issue remain for the 65nm CELL as well or only 45nm process?
And in the end, does the CELL even get that hot after a delided (with fresh thermal paste even on 90nm) to make it worth all the trouble?

Not surprised they would work to remove EE/GS support after the chip has been long gone from the board itself, but confused about the "analog inputs"? Is that how PS2 games are communicated from the EE/GS <- to the -> RSX? Analog not Digital?
Click to expand...

Every new revision Cell has been changed in some way. We don't have pinouts for 45 nm , although they would be more similar to 65nm.

Perhaps I shouldn't call them analog inputs, but rather connections going to EE/GS or GS alone. The 28nm does not have the inputs/outputs needed.
 
You must log in or register to reply here.

Similar threads

Back
Top