PS3 Using a Digital XRAY to inspect boards

Yanitee

Yanitee

Member
Greetings! My friend @DeadEnd sent me a few dead boards for practice, he used to practice on them himself and they had moisture, popcorning, and other strange damages. He gave them to me with the message that they are probably not fixable, but I am devoted to the COK so I want to attempt to fix them anyway! Anyways, this one board I am trying to fix already has a 40nm on it, so I thought it would be a nice starting point to find and figure out the problem. I do not know the full story of this board, but the cell has been reballed and the error code is currently 3032, after a long time of probing and testing, it seems that all is fine, resistances are OK, a lot of components we're checked and they we're also okay. I decided I have to reball the cell once again just in case something went wrong during the original reball, before I did that a friend of mine (@Workz_777) gave me the bright idea of xraying the board! I thought that was an awesome idea, and I was also in luck since my father is a vet and owns a digital xray. I took a reference board and I took the 3032 board, xray scanned them both and I got some awesome results!

xray.png


You can clearly see that the 3032 board had some abnormal merging of balls or damaged solder mask. Since the resistances on the CELL measured ok, I think that it is just damaged solder mask and the xray seeing the exposed copper. Regardless, I think this is a pretty interesting way to take a closer and deeper look into your boards, a lot of things can be seen and perhaps conclusions made about your board!
 
That is indeed extremely interesting, your lucky you have the tools at your disposal to get the xray images.

It's great your devoted to the COK! as I refurbish these consoles (not fix) as prevention is most definitely better than the cure. However along the way I have found mods that highlight that the CPU is undoubtedly the cause of a lot of YLOD consoles and when I have pointed this out, I'm met with only reasons why its still the RSX even though I have strong evidence pointing to the CPU.

GLOD, Graphical Artifacts, and some YLOD are mostly CPU BGA issues NOT the RSX! I have watched Felix's YouTube video and he believes that the Frankenstian mod is a pathway to repair, I simply dont think we can come to this conclusion as we don't have enough Frankenstian consoles in production to test the longevity of the mod. He does point out heat can cause false positives and this could also be the case with the Frankenstian mod. In other words the heat generated to remove the old RSX and then reball the new RSX could actually be flexing the board to reconnect BGA defects on the CPU. I have discovered that using thermal pads to create pressure behind the CPU boots the console, and not only that, it lasts as all 10 I have performed the mod on are still working.

So this is another example of the CPU being the possible cause, a lot of people say 90 percent of the time it's the RSX, I believe it's a lot lower and we all need to be open to the possibility that the CPU causes more system errors than we previously thought. (I'm fully aware this post may not be popular because it's not on trend...... But remember the tokin hysteria and no-one was allowed to point out it may be something else)
 
Timbo9876 said:
That is indeed extremely interesting, your lucky you have the tools at your disposal to get the xray images.

It's great your devoted to the COK! as I refurbish these consoles (not fix) as prevention is most definitely better than the cure. However along the way I have found mods that highlight that the CPU is undoubtedly the cause of a lot of YLOD consoles and when I have pointed this out, I'm met with only reasons why its still the RSX even though I have strong evidence pointing to the CPU.

GLOD, Graphical Artifacts, and some YLOD are mostly CPU BGA issues NOT the RSX! I have watched Felix's YouTube video and he believes that the Frankenstian mod is a pathway to repair, I simply dont think we can come to this conclusion as we don't have enough Frankenstian consoles in production to test the longevity of the mod. He does point out heat can cause false positives and this could also be the case with the Frankenstian mod. In other words the heat generated to remove the old RSX and then reball the new RSX could actually be flexing the board to reconnect BGA defects on the CPU. I have discovered that using thermal pads to create pressure behind the CPU boots the console, and not only that, it lasts as all 10 I have performed the mod on are still working.

So this is another example of the CPU being the possible cause, a lot of people say 90 percent of the time it's the RSX, I believe it's a lot lower and we all need to be open to the possibility that the CPU causes more system errors than we previously thought. (I'm fully aware this post may not be popular because it's not on trend...... But remember the tokin hysteria and no-one was allowed to point out it may be something else)
Click to expand...
I agree with you. I already had several PS3 FAT consoles and I could see very high divergences where CPU is about 15 to 20 degrees above the RSX. I've had a FAT PS3 for 2 years with a thermal pads 3mm pressing on the CPU and it's perfect. CPU and RSX temperatures almost always work together.
 
The delid is a bit risky if you don't have good experience, I prefer to venture into the delid only if there is no other way.
 
Timbo9876 said:
That is indeed extremely interesting, your lucky you have the tools at your disposal to get the xray images.

It's great your devoted to the COK! as I refurbish these consoles (not fix) as prevention is most definitely better than the cure. However along the way I have found mods that highlight that the CPU is undoubtedly the cause of a lot of YLOD consoles and when I have pointed this out, I'm met with only reasons why its still the RSX even though I have strong evidence pointing to the CPU.

GLOD, Graphical Artifacts, and some YLOD are mostly CPU BGA issues NOT the RSX! I have watched Felix's YouTube video and he believes that the Frankenstian mod is a pathway to repair, I simply dont think we can come to this conclusion as we don't have enough Frankenstian consoles in production to test the longevity of the mod. He does point out heat can cause false positives and this could also be the case with the Frankenstian mod. In other words the heat generated to remove the old RSX and then reball the new RSX could actually be flexing the board to reconnect BGA defects on the CPU. I have discovered that using thermal pads to create pressure behind the CPU boots the console, and not only that, it lasts as all 10 I have performed the mod on are still working.

So this is another example of the CPU being the possible cause, a lot of people say 90 percent of the time it's the RSX, I believe it's a lot lower and we all need to be open to the possibility that the CPU causes more system errors than we previously thought. (I'm fully aware this post may not be popular because it's not on trend...... But remember the tokin hysteria and no-one was allowed to point out it may be something else)
Click to expand...

Well, you say we don't have enough frankies in production to test their longevity, in which case it'd be interesting to see statistics of your consoles that you refurbished and how long they have lasted, what were the original errors, etc. Please tell how many consoles with 3034 errors started working after CPU pressure mod. I suspect some of them may not be BC consoles or had different errors...

Mello_1993 said:
I agree with you. I already had several PS3 FAT consoles and I could see very high divergences where CPU is about 15 to 20 degrees above the RSX. I've had a FAT PS3 for 2 years with a thermal pads 3mm pressing on the CPU and it's perfect. CPU and RSX temperatures almost always work together.
Click to expand...

That's because the paste under IHS dried up/separated from the lid and you needed to delid it... But eraser mod and any other types of "pressure" mods can do the trick to press the die closer to the lid again.
 
I'm not saying the pressure mod is a pathway to repair, it's simply a good way to determine if you have a CPU BGA issue. The true fix is a CPU reball and a de-lid and this has been done many times since I pointed out that pressure on the CPU boots the console. The reason I highlighted the CPU pressure mod is because not everyone has the skills to do a CPU reball or the money to get it fixed, if that's the case they can apply the pressure mod because its inexpensive and requires little knowledge or skill to apply the mod. Now for my own research when I get a console that doesn't work, these are normally given to me, I apply the pressure mod and if it boots I use PS3 Advanced Tools to grab the codes. 3 of the consoles had the error code 3034 and all of them are still working. I do use dynamic fan control to keep the console below 68 degrees as I believe syscon fan control is a contributing factor to these consoles failing.

So let's be clear, this is not a competition between my mod and the Frankenstian, this is simply me highlighting that the RSX is blamed for the error when in actual fact its a CPU BGA issue. My point is that the CPU has gone unnoticed, even the syscon codes point to RSX when I have shown is highly likely to be a CPU BGA issue. I'm asking everyone to keep an open mind, it's possible that the CPU is the culprit more times than we care to realise.

(I only referbish backwards compatible PS3 consoles)
 
Last edited:
But may I suggest that you may be jumping to conclusions a bit prematurely? You say true fix is a cpu reball and this has been done many times. So have you been reballing CPUs too? Or only doing pressure mods ? How long have those consoles stayed working after each one? How much use have they been getting since? To disprove the tremendous research that Felix has compiled, you need to provide a lot more data than just saying "I applied pressure and it booted and worked for a few days,so it must be a Cell's bga issue". While I do see your point. I think the truth may be more complicated than that.

It seems the board will flex under a variety of conditions... So the moment you are adding higher pressure to the CPU, the RSX area may also be affected. Dynamic fan control will possibly extend the effect... But at the very least you could also compile your error codes, methods of refurbishments and time of use. Show some stats... For instance, what other errors have you been able to remedy by applying pressure ?
 
I'm not trying to disprove felix, there is absolutely no doubt that the RSX failing is a major factor to the death of these consoles. My point is that the CPU may be responsible for more system errors than we realise.

I have done 11 consoles with the pressure mod, 1 failed immediately the others are still working the oldest one is nearly 2 years old.

I don't currently perform reballs, it's simply to risky if done incorrectly can kill the MB. I know other people that have performed reflows on the CPU and its booted the console, this was done when pressure on the CPU booted the console.

My point is simple, we need to be open to the fact that the CPU BGA may be at fault, it's not just the RSX that causes these errors
 
Timbo9876 said:
I'm not trying to disprove felix, there is absolutely no doubt that the RSX failing is a major factor to the death of these consoles. My point is that the CPU may be responsible for more system errors than we realise.

I have done 11 consoles with the pressure mod, 1 failed immediately the others are still working the oldest one is nearly 2 years old.

I don't currently perform reballs, it's simply to risky if done incorrectly can kill the MB. I know other people that have performed reflows on the CPU and its booted the console, this was done when pressure on the CPU booted the console.

My point is simple, we need to be open to the fact that the CPU BGA may be at fault, it's not just the RSX that causes these errors
Click to expand...

What were your syscon logs for the 11 consoles, i'm just curious.
 
Timbo9876 said:
In other words the heat generated to remove the old RSX and then reball the new RSX could actually be flexing the board to reconnect BGA defects on the CPU.
Click to expand...
This shows a rather serious lack of understanding about the mechanics of BGA flip chips.

IF what you are saying was true, as soon as your board cools down off your BGA machine, it would no longer be flexed and thus 0 revival would occur. Simply heating does not and cannot revive a chip with BGA damage. It must be reballed.

However, heat CAN revive a chip with an internal flip chip problem. The reason for this is that when you heat the chip above the Tg of the underfill, it softens and then the slight flexing due to being heated can cause connections to reconnect. When the chip cools, the underfill hardens again which "holds" the connections together. However, this effect does not last even if the chip remains below the Tg.

As a result, it is impossible for the systems to behave the way they do when exposed to heat of an RSX reball or just a heatgun, if what you are saying is true.

The theory about the defective RSX chips was made by research and science. If you wish to challenge it, you need to provide research and science proving your point. You can't just say that the scientifically achieved consensus is wrong without evidence. You need to prove it.

Kind Regards,
Josh

PS: The reason that the whole NEC/TOKIN debacle happened was due to the heat applied to the board while removing the NEC/TOKINs. This does pretty much exactly what I describe above with a temporary "revival" of the RSX. The data has clearly shown that most of those consoles died again very quickly. So that is a moot point and irrelevant to this conversation.
 
Most of the consoles I did where GLOD or ones with Graphical Artifacts. However I did have 3x 3034 and 2X 3010, the others simply had 1001 or 1002 errors, no errors where generated as GLOD doesn't seem to generate an error in Syscon.

I'm no expert and I don't pretend to be, if your saying the heat required to remove the RSX and replace a new one can not generate a false positive then I bow to your superior knowledge.

If the statistics are accurate that felix has put up then I'm guessing it must be a coincidence that 70 percent of faulty consoles that have been given to me boot after pressure is applied to the CPU.

Like I say I'm no expert I'm simply putting my findings on the table so we can consider them. I will continue to make recordings of faulty consoles I have been given and I will apply pressure on the CPU, if it boots I will see if it generated a serious error code using PS3 Advanced Tools. I will continue to test the mod and see how long it lasts.

I'm only interested in getting to the bottom of why these consoles fail we don't have access to all the consoles in felixs findings so it's possible some CPU BGA errors slipped through the net.

Peace ✌️
 
Timbo9876 said:
Most of the consoles I did where GLOD or ones with Graphical Artifacts. However I did have 3x 3034 and 2X 3010, the others simply had 1001 or 1002 errors, no errors where generated as GLOD doesn't seem to generate an error in Syscon.

I'm no expert and I don't pretend to be, if your saying the heat required to remove the RSX and replace a new one can not generate a false positive then I bow to your superior knowledge.

If the statistics are accurate that felix has put up then I'm guessing it must be a coincidence that 70 percent of faulty consoles that have been given to me boot after pressure is applied to the CPU.

Like I say I'm no expert I'm simply putting my findings on the table so we can consider them. I will continue to make recordings of faulty consoles I have been given and I will apply pressure on the CPU, if it boots I will see if it generated a serious error code using PS3 Advanced Tools. I will continue to test the mod and see how long it lasts.

I'm only interested in getting to the bottom of why these consoles fail we don't have access to all the consoles in felixs findings so it's possible some CPU BGA errors slipped through the net.

Peace ✌️
Click to expand...

GLODs should be checked with a different method, but oftentimes it still points to RSX issues as well... Although somewhat ambigously. At least in some of the GLOD consoles I had. You can use "hdmi chstat 0", "hdmi ports", "hdmi redid 0" commands. Usually what happens during a glod is that RSX is not setting the resolution, even though HDMI chip is detected and the cable is also detected to be plugged in. Some consoles will output to analog only, others will not output anything. Here's an example of the latest GLOD console I had that showed black screens on all outputs.

GLOD_RSX_CHSTAT_LOG.jpg


Another thing to check is the south bridge log, where you can sometimes see at which point the boot stops or if there are any errors.
As far as Cell is concerned, here it has been detected just fine and the system has booted. Ofc, RSX has also been detected, but that just shows that it has passed the bittraining and flexio interface is working. However RSX likes to fail in different ways so go figure...

GLOD_SB_log_A00.jpg
 
That's really interesting, like you say both the RSX and CPU have been detected but the console fails to display. So why is it then the console then displays and works fine after pressure has been applied to the back of the CELL processor chip? This for me would indicate a CPU BGA issue but you seem to be saying that can't be the case as the CPU has passed the boot sequence? However so has the RSX So maybe syscon can't quite determine what's going on and all it does know is the system is failing to output hence the resolution settings are not set.

The fact remains when you push down on the CPU the console boots, what's going on if its nothing to do with the CPU BGA?

https://youtube.com/shorts/8lE-KlIieLs?feature=share

 
Timbo9876 said:
That's really interesting, like you say both the RSX and CPU have been detected but the console fails to display. So why is it then the console then displays and works fine after pressure has been applied to the back of the CELL processor chip? This for me would indicate a CPU BGA issue but you seem to be saying that can't be the case as the CPU has passed the boot sequence? However so has the RSX So maybe syscon can't quite determine what's going on and all it does know is the system is failing to output hence the resolution settings are not set.

The fact remains when you push down on the CPU the console boots, what's going on if its nothing to do with the CPU BGA?

https://youtube.com/shorts/8lE-KlIieLs?feature=share

Click to expand...

Well... I don't have an explanation for your case, other than pressuring Cell flexes the board and may be causing RSX to get pressured harder too. Especially at the flexio side, which is the closest side to RSX. I haven't tried the Cell pressure trick on my board. I think some of them also get revived from baking at 100 C...And yes, Cell has passed the boot sequence so I highly doubt it's involved in GLOD in any way in my case. Also syscon is not the one that's supposed to set the resolution, it's RSX.
 
I get that, I'm all about logical conclusions which is why you have to ask yourself "why is it that pressure works on a 3032 error" which we know is a CPU issue and then also works when it's a GLOD issue?"

When it's a known CPU error we conclude that the pressure reconnects the BGA but when it's a GLOD we then say that the pressure must do something to the RSX. Surely it's more likely that on both occasions its pushing down on the CPU BGA and its this that is revolving the isuue.

Graphical Artifacts did indeed return on a few of the consoles, however this was when it was extremely hot here in the UK and the console was constantly running at 68 degrees using webmans dynamic fan control. So for these few consoles this was enough for the Artifacts to return, however once the console cooled down it only needed a video reset and it worked fine again. To prevent this from happening again I advised my friends and family to set webman to keep the console under 60 degrees whilst the weather was hot. This was enough to keep the consoles working whilst the weather was so hot, when it got cooler we simply put the settings back to 68 degrees and all has been fine since.

Like I say I'm just putting this out there, it's through people on this site highlighting there findings that I have learnt so much, if we could get to the bottom of this it may help others.
 
Last edited:
Timbo9876 said:
I get that, I'm all about logical conclusions which is why you have to ask yourself "why is it that pressure works on a 3032 error" which we know is a CPU issue and then also works when it's a GLOD issue?"

When it's a known CPU error we conclude that the pressure reconnects the BGA but when it's a GLOD we then say that the pressure must do something to the RSX. Surely it's more likely that on both occasions its pushing down on the CPU BGA and its this that is revolving the isuue.

Graphical Artifacts did indeed return on a few of the consoles, however this was when it was extremely hot here in the UK and the console was constantly running at 68 degrees using webmans dynamic fan control. So for these few consoles this was enough for the Artifacts to return, however once the console cooled down it only needed a video reset and it worked fine again. To prevent this from happening again I advised my friends and family to set webman to keep the console under 60 degrees whilst the weather was hot. This was enough to keep the consoles working whilst the weather was so hot, when it got cooler we simply put the settings back to 68 degrees and all has been fine since.

Like I say I'm just putting this out there, it's through people on this site highlighting there findings that I have learnt so much, if we could get to the bottom of this it may help others.
Click to expand...

When did the pressure work on a 3032 error ? Did you mean 3034 ? Because in the case of the original poster, I don't think he ever tried the pressure trick. But his case is something rare, not a typical one. It's a board I worked on and at some point I went a little wild on reballing and badly tuned thermal profile. I was struggling with boards popping because of high moisture (they came from japan by sea for 6 months!!!). So Cell or some contact on the board probably just popped (either inside substrate/die or or copper interlayers on the board) and it's a pain to track down now. But like he said, I left it as a practice board and I just sent it over to him for spares. He decided to try and diagnose it but this is a rare case with a rare 3032 error. And also it's unlikely that balls are merged anywhere according to xray. We measured hell of a lot of resistances, voltages etc, and there's no shorts in any of the lines. The only reason may be that the balls are not all touching the pads, which I doubt... Anyhow, I don't think he was trying to prove or disprove anything, it's just interesting to see xray image.

As for your logic, I'm not here to argue but I'm not sure if you get my logic either . You keep saying it works when it's a glod issue. Well, I never denied that it could "work" , by clamping the board harder under CPU, but adding pressure under CPU causes the larger area of the board to be squeezed harder as well, which is not just CPU. You say "I only did it for cpu" , but you're causing a stronger bend and the bend will affect RSX area as well. They are close enough to each other and it's just physics that squeezing one side will spread the effect to a larger area. But it's not even just RSX bga, it might even be causing microscopic bends on the RSX's substrate and subsequently its bumpgate under the die is affected. Which is most likely the most common true cause for all the graphics related problems, GLODs and 3034 errors.

Once again, once Cell has passed the booting sequence I'm convinced there is nothing wrong with it. Otherwise it wouldn't even boot. So perhaps you can do the test yourself on the next GLOD board by checking SB log before doing anything else. SB log activation needs a few steps though. I could share a mini guide for it if you want.

The whole fan manipulation thing is really quite a desperate measure. You can't seriously blame hot weather and say 68 Degrees is too much. It's not even that high in the end. The stock fan settings let it run into 70s. I myself use stock fan setting and Cell on my frankies also runs at 75 C in heavy games and it's all fine. For cell especially, the normal conditions were stated to be 90 C in the official documents... And there should not be ANY artifacts or video resets needed for a healthy console. And it definitely doesn't sound like anything related to Cell, because RSX is responsible for most of the graphical stuff. IMHO it's all because rsx is on its last breath and you're just trying to delay the microscopic bumpgate failure. It will react to the temps and nearby pressure due to mechanical stress/bending moment (it's an area of mechanics, how forces interact), but it will eventually fail entirely.
 
Okay, what your saying makes perfect sense it's just strange that if you apply pressure to the RSX directly it solves nothing, but applying it to the CPU boots the console.

I did find that resting the board on a hot plate at 160 degrees boots about 65 percent of YLOD consoles I have come across. This I belive to be RSX bump failure and these consoles get used for spares, because I paint the console white even the case gets used for spares. This means I have about 25 COK 002 and 5 COK 001 boards I have out of action. I have considered buying a rework station however I'm not a repair expert and the process looks incredibly difficult, maybe sometime in the future I may consider it.

Now the consoles that wouldn't boot with heat where always the GLOD consoles, this method wouldn't boot the console, but the pressure mod did.(I tried to gently heat the RSX and got no boot, I then tried the pressure mod and it booted) I'm currently having my house done up so have mostly put the PS3 console refurbishments on hold till after Christmas, it does seem if you keep the console below 65 degrees you are at a much lower risk of bump gate. Now the consoles that did boot with heat would last from a few minutes to a few weeks however the consoles that boot with pressure last a lot longer with only one failing out of 11. I started to get only get GLOD consoles givern to me or ones with Artifacts and that's when I discovered the pressure mod would boot the console and lasted months, which turned into years. I then also discovered that some YLOD consoles would boot using the same mod, I will now need to revisit my spares pile to see if the pressure mod works on any of these, I have tried a few boards and it didn't but I really need to go through the lot to be sure. From the moment I discovered the mod it seems nearly every broken console would boot under pressure with only one failing, I then started thinking that the ones I used heat to get a temporary boot may have damaged the console in some why which is why the pressure now won't work on these consoles, I'm no expert so this is just a hunch.

So my intuitive thinking tells me that GLOD consoles have a much different fault than your regular YLOD console, I'm not a repair expert and I only have basic soldering skills, but like I say if feels like a different fault, it could still all point straight back to the RSX in the end.

The only way I will know for sure is to take a GLOD console that boots under pressure, remove the mod and perform a frankenstian mod on the console, if the mod works then that would prove that the CPU pressure is somehow applying enough stress on the board to effect the RSX. Unfortunately I would say that I won't be doing any frankenstian mods anytime soon but when I start I will.share my findings.

This is all a massive mind field which is why i only referbish working consoles as prevention is better than a fix. Along the way I have either been given non working consoles or I have bought them for spares, the BD drive is a spare I constantly need. I then started to experiment on these consoles and that's when i made these discovery's

PS, miss type on the error code
And yes please for the guide
 
Last edited:
You must log in or register to reply here.

Similar threads

Back
Top