...should we still be using 3x 470uF caps when installing a 40nm on a frankenphat? or should we drop it to 330x3? with or without the extra MLCC caps? BC units
The frequency resopnse curve of the combined filter is affected by the choice in caps you make. 330 responds slightly differently. It'll work, but for convenience and because the extra capacitance (within reason) doesn't hurt, it's easier to use a blanket recommendation. Replace each tokin with 3x 470uF polymer caps. Can be TaPol or AlPol, but they need to be low ESR processor decoupling caps. That'll work by itself.
However the MLCCs bridge the gap and attempt to broaden the frequency response to account for the fact that the tokins were better at reaching and filtering higher frequiency noise/ripple that the TaPol and AlPol alone cannot. If you look at the way SONY transitioned to a polymer array, they used 470uF polymer (tantalum or Aluminum) AND 10uF/22uF MLCCs, in addition to the 0.1uF array. This is a more standard method nowadays. The proadlizers (tokins) were a neat product, but not necessary. And old now. As they die, we need an alternative filter array that fills the role they played. While Polymers alone will "work," it's not a 1:1 replacement. A fully populated Tantalizer is designed to be.
If you want to use 330uF, use 4x. That'll also "work."
Why my system got ylod'ed again and can i get rid of it with replacing more necs
And for my last question is there a way to get the error codes without the test pads (my console isn't running long me to open a web browser etc.)(I heard about some service ports or smth like that but i dunno where it is or even its eve usable?)
You'll have to repair the pads, connect to a VIA, or the syscon pins themselves.
As for what caused it to fail again, we'd just be guessing. You didn't post a picture of your work, so I'll assume your soldering wasn't great. A cold joint can cause that. If you didn't replace all the tokins, the bad apple could be spoiling the bunch. Or while disassembling/assembling you may have caused the MB to flex and break a connection. Like the BGA, or knocked off an SMD...we can't know without the codes.
BUT...If it YLOD is greater than 4s it's probably the caps (my guess).
...clean CECH-2001A slim...Ran great...130d run time. I tore it down, and delidded both chips...Now it doesn't start and I get the following in the error log:
# A0403034 2DD0C299
# A0404411 2DD0C298
I can see exposed copper where you likely nicked a trace. Can't see it well enough to be sure. And the paper towel is obscuring the view of the upper right hand corner. Those errors are common after a failed delid attempt and that model is not prone to them from natural causes.
Out of curiosity, what tool did you use to delid? I'd reccomend so modifications to prevent the chance it can dig into the substraite. Also, never start from the corners.
...can I replace my caps with a 1000uF and a 330uF TaPolys? I think the 1000uF one should be fine given its ESR/impedance characteristics (sorry, can't post a link yet, it's Kemet's) – I think they are very decent with an ESR of 5mOhm at critical freqs. The 330uF one is a bit worse at higher frequencies impedance-wise. I'd also throw in a couple of ceramic ones (100nF, 10uF, 20uF).
I just hope 3001 doesn't mean the board is dead ...
The 4x tokins have a combined ESR of 0.375 mOhms. As an EE I'm sure I don't need to remind you, but the reason you choose multiple 470uF low ESR (4.5 mOhms is ideal) is so that their combined ESR divides by the number of them used in parallel (4.5mOhms / 12 caps = 0.375 Combined ESR of the resulting filter array), so you can meet or exceed this specification. Using a single 1000uF of 5 mohms will result in a combined ESR that's higher than the original array. Which is within reason may not become an issue, it's just not recommended and those kemet caps are more expensive. Cost's more for worse performance.
...K with 423 days of uptime showing signs of NEC/TOKINs going bad. ...1002 codes...I have 330uf caps...Can I use these?
Use 4x 330uF per tokin replaced. Be aware that old caps that have been installed before may degrade when reflowed several times. Not such a big deal, but old caps from slims are still old. They can go bad too. New caps resets the timer.
...cechk04, DIA-002...gets a YLOD after a few seconds and turns off...I run bringup is a bunch of a0202121...
Code:
Power-on: 342day 17hour 51min 49sec
ofst[ 12]:err_code:0xa0202121, clock:0x2de90d59 2024/05/28 21:47:05
ofst[ 16]:err_code:0xa0202121, clock:0x2de90d59 2024/05/28 21:47:05
ofst[ 20]:err_code:0xa0202121, clock:0x2de90d59 2024/05/28 21:47:05
ofst[ 24]:err_code:0xa0202121, clock:0x2de90d59 2024/05/28 21:47:05
ofst[ 28]:err_code:0xa0202121, clock:0x2de90d59 2024/05/28 21:47:05
ofst[ 32]:err_code:0xa0202121, clock:0x2de90d59 2024/05/28 21:47:05
ofst[ 36]:err_code:0xa0202121, clock:0x2de90d59 2024/05/28 21:47:05
ofst[ 40]:err_code:0xa0202121, clock:0x2de90d59 2024/05/28 21:47:05
ofst[ 44]:err_code:0xa0202121, clock:0x2de90d59 2024/05/28 21:47:05
ofst[ 48]:err_code:0xa0202121, clock:0x2de90d59 2024/05/28 21:47:05
ofst[ 52]:err_code:0xa0802021, clock:0x2de90d5f 2024/05/28 21:47:11
ofst[ 56]:err_code:0xa0801002, clock:0x2de90d5f 2024/05/28 21:47:11
10x A0202120 / 1x A0801002/A0802120 error combo is a newer reported one, but I've seen it be the RSX tokins. It seems to be more common in nonBC phat models due to a difference in how they report error codes. Your's is a K model (DIA-002). I believe I saw this same error in one of my DIA-002's. I used it as a pretext to steal it's 65nm RSX, instead of replacing the tokins which were what likely killed it. For me the 65nm RSX is more valuable to harvest and use to revive a BC model, than reviving the DIA-002. Even though it's a relatively easy diagnosis.
In this case, the 2120's do not indicate the HDMI or DVE IC, because the number of A0202120's is always 10x and followed by a 1002. Check the timestamps, they occur in the same YLOD event.
This is going to be tokins IMO.
...CECHK01 model, with about 11 days of uptime...After replacing the thermal paste, it acted like GLOD. ...Here are the codes.
00000000 A08014FF 2DF7E0D6
00000000 A0801701 2DF7E0D6
00000000 A0232102 2DF7D7F5
00000000 A0232102 2DF7D7F1
00000000 A0232102 2DF7D7ED
00000000 A0232102 2DF7D7E9
00000000 A0902120 2DF7D77B
00000000 A0403034 2DF7D77B
00000000 A0404002 2DF7D77B
00000000 A0902120 2DF7D775
[/CODE]
"after replacing thermal paste" it developed an error typically associated with BGA defects in this model. Check your method of breaking the suction between the HS and IHS. If you flex the Motherboard too much during that separation, it can cause BGA defects. It scares me every time I have to do it, but I use a plastic spudger to try and lever it up from the side, where it causes the board to flex less.
Also check for drop damage and any signs the board has been tampered with. If a BGA defect is suspected I always find it's impossible to rule out user error. In other words, I usually find evidence it didn't happen naturally from thermocycling. Sealed, errors predate any separation of HS from IHS, etc...
Likely, the GPU needs reballed.
I read Syscon and got error 3039, but there is not much described about this error, besides the beginning of the code being different from everything shown so far A052.
Model is CECHL04
Below is the full reading:
Code:
00000000 A0523039 FFFFFFFF
00000000 A05114FF FFFFFFFF
Any idea what it could be?
Facinating! That is an as yet unreported code! A new code!!!
These things get me excited. BAsed on the step number it's likely related to byte-training. This is the process immediatly after Bit-training, which is what fails in the case of Processor failures, like BGA or Bumps defects in either the CPU/GPU. This is when the processors' high speed interface on the FlexIO, th traces connecting them, are fed a series of bits or bytes, and the expected timing and response is measured to calibrate the phase lock loop and bring the alignment to a center position. This improves high speed signal integrity.
When there's an issue, like a poor solder joint, or a worn out die, this calibration procedure has a harder time completing, and when it finally can't...BEEP...BEEP...BEEP! Byte training come immediately after bit training. BitTraining causes 3034/4xxx errors at step# 40. ByteTraining causes 3035 errors at step# 50, and finally we've seen 3041 at step #52. That's the closest to your 3039 we've seen. We never got a resolution to it, nobody knows what caused it. But I have a guess and your's may help with this...
Your 3039 is right at the last stage before the bootloader. Between ByteTraining at step number 50 and when the StarShip 2 controller loads the bootloader stored on flash at step 60. Issues with the Nand/Nor cause A0603040 errors. But botth you 3039 and that 3041 occured at a previous step number of 52. The 14FF is a clue as to what's going on. That's a Checkstop. This is usually a hardware error during a software check. It discovered a fatal HW error during a check, and stopped because of it.So the poignant questions are...
- What comes after ByteTraining has determined the CPU and RSX are calibrated and ready for action? (assuming they are)
- What critical HW is needed for the bootloader to start?
I speculate it may be the Flash or controller itself, the starship 2 controller, because it's similar to 603040, but occuring earlier. So perhaps 3041 is an issue with the StarShip 2 controller. And perhaps your 3039 is too. It gives you a place to look at anyway. However, this is not going to be easy to troubleshoot.
Could you please provide as much information about the condition of the console as you can? Like describe if it was sealed, dusty, has evidence of drop damage, water damage, what work have you done, or what work can you see was done. Like, does it have new thermal paste? Is there flux residue around the RSX/CPU/ Has it been delided? And so on.
"I have a PS3 series 20 and have a NEC/Tokin connected to the RSX, NEC/Tokin with the series 0E108 totaling 4, if I replace it with NEC/Tokin series 0E128, can all NEC/Tokin series 0E108 be replaced with 0E128?"
It's exceedingly difficult to replace the proadlizer with a proadlizer. You will have to subject the board to the same temperatures needed to reflow the solder under the processor. It's more adviasble to remove them and replace with polymer caps soldered to a carrier board like the
Tantalizer, because it makes soldering easier and can be done using an Iron. It won't risk the BGA. That was my main objective in making designing it.
) into your posts and findings, 
Although these posts of yours might not have been there yet. Or they might have just come out, I don't know.
Don't know for how long, though ... There are no signs of bad RSX bumps/broken BGA balls yet. Before my fuckup there were only 0xa0801001 errors in the errlog. Someone did delid both big ICs before, however, so I don't know 
![>](http://<a href="https://postimg.cc/47KwRDsm" target="_blank"><img src="https://i.postimg.cc/47KwRDsm/error-1002.jpg" alt="error-1002"/></a><br/><br/> <a href="https://postimg.cc/wyS2qwjF" target="_blank"><img src="https://i.postimg.cc/wyS2qwjF/IMG-20240624-204641.jpg" alt="IMG-20240624-204641"/></a><br/><br/> <a href="https://postimg.cc/v12XjYRJ" target="_blank"><img src="https://i.postimg.cc/v12XjYRJ/IMG-20240624-204658.jpg" alt="IMG-20240624-204658"/></a><br/><br/>)
