PS3 Frankenstein PHAT PS3: CECHA with 40nm RSX

well i completly agree back then gpus where fail. even a rebal was a gamble because most of them had bad conections beteween die and substrate. and other on substrate and main pcb
 
In a way, it's true. If the console works , perhaps there is no need to go above & beyond to rip out your og RSX for the preventative maintenance sake. However, some of us believe the stock gpu is inherently flawed. It was manufactured at the time when Nvidia struggled to get the packaging formula right. I doubt Sony was immune to the problem. There were reports of bumpgate, inter-layer delamination and underfill issues. It took them a while to perfect the process, but by the time 40nm rolled around the packaging technology had improved. But this is just my own conclusions, everyone is free to think otherwise. Having said that, there are examples of untouched systems working for a decent amount of years as well. So not all 90nm are equally terrible, but noticeably many seem to die sooner rather than later.
It is true that the BGA/Bump/Underfill situation in 2006 wasn't as good as today, but I argue that FCBGA itself is flawed. I honestly do not think BGA will ever beat the reliability of QFP, and that for the sake of pin count and lower BOM costs, the consumer industry is fine with that. Why we (the public) don't demand sockets is beyond me. Not only can you replace a defective chip, the socket itself provides strain relief. I've never worried about my computer CPU running 70-90C. They thermal throttle at 100C FFS! Your GPU will flex itself to death reaching those temps regularly. And the PS3? Forget it!
 
A slightly offtopic worthy mention of anybody who is planning to play ps2 games on a frankenstein or any other model for that matter. It is known that many of the ps2 games do not support widescreen. While you can use the patches in the pcsx2 emulator to correct the situation, it isn't quite as simple if you are playing isos/discs on a ps3. However, there is a way to get them to display correctly in widescreen. Just download ps2 widescreen patches archive and use a tool called Ps2 patch engine to patch the iso. That way you can play almost all of the games in a correct aspect ratio through the ps3. But keep in mind, ps2 patch engine is blocked by an antivirus as a false negative.
 
That's pretty interesting. I'm torn about this kind of patch/mod. The purist in me says that's not the way these games were intened to look. They were intended to be played on a CRT in 4:3. OTH the devs would surely have added a widescreen mode if HDTVs were standard at the time. And some games did have widescreen options, even 1080i. So I think it's really a matter of preference. If you like the way it looks on a CRT, get a CRT or use a RetroTink5x with a CRT filter. If you like HD and sharp pixels, go HDMI. And if you hate black bars then the patch is for you. It's good to have options.

But I hate the whole false positive antivirus hits. I really can't understand why some people have programs that have low volume and don't trigger antivirus alerts, and others do. What really sucks is when you have a really cool tool that is great, but the things the antivirus finds sound sketchy as hell. You don't know who to trust, wouldn't trust the tool, except you want to use the tool so bad you ignore your better judgment. I've come across a few of these over the years and it's aggravating that they can't figure out a way to make thier program legit pass these scans.
 
In my experience, it is quite enjoyable to play most ps2 games through ps3 on a CECHC/CECHE model particularly and 4k TV at around 2 meters distance. But I can't claim that they will look as good on ALL HDTVs. Mine is LG and it seems to have a decent scaler inside. I use smoothing option and widescreen patches makes most games look as good as they possibly can. Why only CECHC/CHECE ? Well, there is a noticeable sharpness difference in a lot of games in comparison to CECHA. C and E models simply produce sharper image. I will eventually make a comparison video to prove that.

But speaking of quality. If anybody has heard of PSX DVR, then here's an interesting find. It appears nobody has compared its output to a regular ps2 or ps3. It seems to have better smoothing as well.

https://www.reddit.com/r/ps2/comments/fty5dn/ps2_vs_psx_clearly_there_is_a_winner/
 
Last edited:
What I've also noticed is that CECHA displays ps2 games in a smaller frame compared to CECHC. Resolution might be the same but the image is "smaller".
 
Another interesting thing to note when installing a rsxmod. If after installing a new revision RSX (65 or 40nm) you run a "bringup" command and see a bitraining error such as [POWERSEQ] Error : BitTraining RSX:RRAC:BX0:BX:FLEXIO_ID , do go ahead and cross reference it with this https://www.psdevwiki.com/ps3/Rambus_Registers#RSX_RRAC_Registers As you can see, your bittraining routine is stuck in the very first line, so chances are your modchip is not active (not connected to the pcb properly) and syscon cannot even begin to ID the RSX. You can also check that by disconnecting the mod and the error should stay the same. Basically this is the error you get when there is no modchip connected. So pay attention to the pcb joints where you are connecting the wires. Sometimes it may seem like the wires are holding and the solder is there, but if you hadn't exposed enough copper when soldering then the contact will not be made properly. Same thing if you accidentally exposed too much copper in the surrounding contact and your modchip wires may be touching it. It could be very difficult to see what is going on without microscope.

However, in case you see bittraining errors that match with any other lines in rambus sequence, then chances are that BGA/Bumpgate is at fault. For instance I got errors such as BitTraining BE:RRAC:RX0:GLOBAL1:RX_STATUS and BitTraining BE:RRAC:RX2:GLOBAL1:RX_STATUS when my board was flexing and the RSX was not soldered properly. These kind of errors could also show up in ylod boards with stock 90nm. Basically anything that is not the first line in rambus sequence would indicate an issue with RSX and its connection with the board (It even points you to a specific pin where the problem occurred).

This is probably nothing new, @RIP-Felix has discovered the bittraining references a while ago and @M4j0r has posted the links earlier on as well. But I just wanted to remind about it once more, because we have just been trying to help somebody figure out this exact issue.
 
Last edited:
While on the topic, I have talked to @botakompong once more. And I think it's become more clear now about certain discrepancies in our thinking. While his techniques and ways are no doubt well analysed and tested, there are still some differences in the reasons for the mod. First of all, ps2 games on ps3 is not a common practice in Indonesia. He has not done this mod for cok boards for a long time, while we are mainly focusing on those particular boards over here. In fact, it is the sole reason I have been active in the community. So just keep that in mind. Secondly, his reasoning is to repair large numbers of consoles with easy solutions. He may not necessarily strive to stick to the most perfect way, but a rather efficient and working way.

Having said all this. I found it a bit ironic that some of us in the thread have been trying to measure the resistor modifications that were made by Sony. Well, where do you think @botakompong has come up with those numbers for the resistors? He has studied those same frankenstein boards by Sony and has analyzed the changes before we even knew of their existence. So you guys just didn't think to ask :D There is no need to desolder and measure them several times. But here's the thing some may have failed to notice. The changes made by Sony concern more than just 1.2 to 0.95v adjustment (FlexIO core voltage). While the simpler method suggested by our friend was to simply add a 0.95v IC, Sony changed the RSX Core VDD well. And yeah, this has already been discussed, but just to remind you. In the picture below, notice that changes to pwrgood signal are not related to other adjustments. It is changing a different voltage (RSX VDD). Now, I don't remember what I may have been babbling about in the past. But this kind of does seem important as well.

voltage changes.jpg

However, Sony has also done the same to Cell BE section in here:

cellbe changes.jpg

And this was confirmed by him as well. Remember, he has analyzed the refurbished boards himself. But he doesn't know why it was done to Cell as well. I haven't looked into that either.

Now @Icferrum had also mentioned earlier that these extra changes are not prerequisites to RSX replacements , but yet beneficial to phats. So you can make your own conclusions if you really want to follow the sony way precisely or not.

But again, it's hard to measure resistors value. And what is interesting, measured resistance is the same as for CELL/BE VRM.
So, I can't tell for sure what are real values, but R6207=R6112 and R6205=R6111. Though R6112/R6111 may have also been replaced by Sony, who knows.

DeadEnd already mentioned that these resistors are for adjusting PWRGODD signal lower threshold. What he didn't mention is that this threshold is primarily depends on Vout_noload/2 value and (R1+R2)/R2 is just a modifier. Vout_noload is defined in NCP5318 datasheet for every selected Vout, which is selected by RSX itself via VID. So if RSX needs 0.95V VDDC, Vout_noload will be 0.9310V and so Vlower will be already set to 0.4655V w/o resistor adjustments.

View attachment 35027

I don't say that resistor mod is not needed at all. With stock R1(R6207)=15kΩ and R2(R6205)=20kΩ, Vlower would be 0.8146V for 0.95V VDDC - about 0.1 V difference (for 1.2V VDDC this difference would be roughly the same). With modded R1(R6207)=10kΩ and R2(R6205)=27kΩ, Vlower would be 0.6379V for 0.95V VDDC - about 0.3 V difference. I think, if Sony has really done this mod, the reason for it was to have less tight tolerance for PWRGOOD signal after they found out what resource heavy games can do to voltage ripple on both CELL/BE and RSX. So it should also be beneficial to standard Phats. But it's definitely not a prerequisite for RSX replacement.
 
Last edited:
@RIP-Felix I have noticed you've been adding more text here https://www.psdevwiki.com/ps3/Syscon_Error_Codes#3034

I think you could also add that modchip is not the only way to install a different model RSX. You can also do it without the modchip by using a flashable syscon or a syscon version 304GB (if I remember right). It requires patching with bus pirate however.
Yeah I should mention that SONY had an official method, but I don't really want to get into the details of it, since we never actually figured out how to replicate it. Even better would be it's own page that we could link to, in order to keep the error codes section clutter free. It would be beneficial to have this documented "officially" on the wiki. You should sign up! You have as good an understanding of how it works as I do, if not better.
 
...In the picture here. Notice , the changes to pwrgood signal are not related to the changes in the FlexIO core. It is adjusting a different voltage. Now, I don't remember what I may have been babbling about in the past. But this kind of does seem important as well.

View attachment 35402

However, Sony has also done the same to Cell BE section in here:

View attachment 35403

And this was confirmed by him as well. Remember, he has analyzed the refurbished boards himself. But he cannot tell why it was done to Cell as well. It doesn't seem to do much to it.

Now @Icferrum had also mentioned earlier that these extra changes are not prerequisites to RSX replacements , but yet beneficial to phats. So you can make your own conclusions if you really want to follow the sony way precisely or not.
Yes, this is one of the things that got me researching the power control topology. It's been ongoing for the past few months because it's a difficult topic to unpack. I just made a post introducing this topic on the SYSCON thread. Impeccable timing!

This is the next Part in that series and I haven't gotten that far into it yet, but will be soon. Here's what I do know. That voltage divider selects the setpoint for the Voltage sensing pin for Power Good's lower threshold. My current working hypothesis is it was done to reduce the frequency of false 1001 and 1002's errors. That would explain why they did it to both buck controllers. A sort of admission of guilt that they set it too aggressively at the factory. It's one of the questions I want to answer in "PS3 Power Control Topology - Part 2." It could be particularly important, because we're seeing a lot of unexplained 1001's recently. I have intentionally held off recommending this mod to people, because I'm only freshly aware of it's potential. I really want to systematically research the topic before I feel comfortable with a conclusion. It's all speculation, hypothetical at this point. And I'm not an electrical engineer!
 
@sandungas @M4j0r do you know if there is a way to check the VID setpoint for the Cell/RSX at Idle?

The Vid pins VID0-5 on the buck controller form a 6-digit code corresponding to the Vout No load setpoint. Power Good Vmin and Vmax thresholds are relative to that set point. With stock COK-00X voltage divider values (15K and 20k), Vmin = -163mV. Vmax is always +100mV. The Vout voltage cannot deviate more than that. If it does power good goes low and the SYSCON will error.

What SONY's new resistor values (27K and 10K) do is change Vmin = -400mV. So the Low Voltage Drop-Out (LVDO) is now more than twice as low, allowing much more ripple before it triggers an error. That is consistent with my hypothesis above.

But before I can conclude it is correct I would like to know what the VID at idle is, so I can compare actual voltage measurements with the LVDO setpoint. On a console with 1001/1002 error, Vout should drop more than that before it experiances a YLOD.
 
Last edited:
It is true that the BGA/Bump/Underfill situation in 2006 wasn't as good as today, but I argue that FCBGA itself is flawed. I honestly do not think BGA will ever beat the reliability of QFP, and that for the sake of pin count and lower BOM costs, the consumer industry is fine with that. Why we (the public) don't demand sockets is beyond me. Not only can you replace a defective chip, the socket itself provides strain relief. I've never worried about my computer CPU running 70-90C. They thermal throttle at 100C FFS! Your GPU will flex itself to death reaching those temps regularly. And the PS3? Forget it!
I hate BGA too, in small chips with few contacts doesnt makes much sense, and in big chips is a lot less relliable than a socket
This makes me remember a PC motherboard i had that was completly bent by the clamps of the CPU heatsink, it was amazed that was working fine, the CPU socket in the center of the curvature and no problem, lol. I know this was a bit extreme but a BGA cant resits that
BGA is flawed by design because the contraption and expansion of materials, soon or later the balls are going to fail

A slightly offtopic worthy mention of anybody who is planning to play ps2 games on a frankenstein or any other model for that matter. It is known that many of the ps2 games do not support widescreen. While you can use the patches in the pcsx2 emulator to correct the situation, it isn't quite as simple if you are playing isos/discs on a ps3. However, there is a way to get them to display correctly in widescreen. Just download ps2 widescreen patches archive and use a tool called Ps2 patch engine to patch the iso. That way you can play almost all of the games in a correct aspect ratio through the ps3. But keep in mind, ps2 patch engine is blocked by an antivirus as a false negative.
The default ManaGunz instalaltion contains a collection of the PS2 widescreen patches taken from the official PCSX2 forum (includes more patches than the ones approved by PCSX2 project), and it have an option to apply the widescreen patch to the ISO (and another to remove the patch), you should try it

@sandungas @M4j0r do you know if there is a way to check the VID setpoint for the Cell/RSX at Idle?

The Vid pins VID0-5 on the buck controller form a 6-digit code corresponding to the Vout No load setpoint. Power Good Vmin and Vmax thresholds are relative to that set point. With stock COK-00X voltage divider values (15K and 20k), Vmin = -163mV. Vmax is always +100mV. The Vout voltage cannot deviate more than that. If it does power good goes low and the SYSCON will error.

What SONY's new resistor values (27K and 10K) do is change Vmin = -400mV. So the Low Voltage Drop-Out (LVDO) is now more than twice as low, allowing much more ripple before it triggers an error. That is consistent with my hypothesis above.

But before I can conclude it is correct I would like to know what the VID at idle is, so I can compare actual voltage measurements with the LVDO setpoint. On a console with 1001/1002 error, Vout should drop more than that before it experiances a YLOD.
This is over my understanding, im just a tinkerer and im still discovering new things, a couple of weeks ago i didnt knew anyting about the "clock generators" of the PS3, and i still dont get the full picture of how works the CELL/RSX power control
Good work btw, you are digging really deep into PS3 power control
 
I'm located in the USA and interested in obtaining a 40nm RSX modded CECHA01. I already have one that is YLoD, but has never been opened. Can someone point me in the right direction please?
 
I'm located in the USA and interested in obtaining a 40nm RSX modded CECHA01. I already have one that is YLoD, but has never been opened. Can someone point me in the right direction please?
First you need to diagnose the board with syscon uart - to see the error log that will point what exactly triggers the ylod on your board.

I doubt anyone in USA offers any modchip repairs though, so make sure to diagnose it first before coming to the second step - solution.
 
First you need to diagnose the board with syscon uart - to see the error log that will point what exactly triggers the ylod on your board.

I doubt anyone in USA offers any modchip repairs though, so make sure to diagnose it first before coming to the second step - solution.

I have a FT232RL on the way. Fingers crossed that Sony already replaced the RSX, and something else is causing the YLoD. Haha! more like 20 RSX errors in SYSCON log.
 
But here's the thing some may have failed to notice. The changes made by Sony concern more than just 1.2 to 0.95v adjustment (FlexIO core voltage). While the simpler method suggested by our friend was to simply add a 0.95v IC, Sony changed the RSX Core VDD well. And yeah, this has already been discussed, but just to remind you. In the picture below, notice that changes to pwrgood signal are not related to other adjustments. It is changing a different voltage (RSX VDD). Now, I don't remember what I may have been babbling about in the past. But this kind of does seem important as well.

View attachment 35402

However, Sony has also done the same to Cell BE section in here:

View attachment 35403

And this was confirmed by him as well. Remember, he has analyzed the refurbished boards himself. But he doesn't know why it was done to Cell as well. I haven't looked into that either.

Now @Icferrum had also mentioned earlier that these extra changes are not prerequisites to RSX replacements , but yet beneficial to phats. So you can make your own conclusions if you really want to follow the sony way precisely or not.
That picture is confusing. It shows the VDDR voltage mod being done using both a BD3504 MOSFET driver (which uses external resistors to select the output voltage as shown) AND the MOSFET (which I thought was only necessary if you DIDN'T want to change the MOSFET)?
voltage-changes-jpg.35402

On my Frankie (PS3#8) I replaced the MOSFET with one from a 3000 model slim, as shown in the bottom right of that picture. But I did not replace the driver, since I didn't think it was necessary at the time. Was I supposed to to change it?
 
Back
Top