This is a general PSA, not directed at any specific user. I would like to clarify
@DeadEnd's point. He's not saying you should clear the errorlog first thing!
The first thing you should do when receiving a YLOD console (bought, gifted, or the thing up and YLOD'd on you) is dump the ENTIRE errorlog. I don't care how you do it as long as you save that record of events. Internal access is better, but if you can't be bothered to mess with code then external is better than nothing. Once you have that record saved...
The second thing you should do is clear the errorlog and test the console. This way you know what the current issue is and there will be no confusion about the error it triggered.
Most people test the console first thing when they get it, erasing the oldest errors from the log. Those errors might have contained a useful progression of events that can aid in diagnosing the problem.
For example...
The errorlog
@NO_ob just posted shows a CPU overheat (A0801200) in the last slot before it would have been erased by the next code. Since then the console has had a series of unexpected shutdowns (1001). They can be caused by many things, but the other associated codes reveal their cause. In this case an overheating CPU caused the console shutdown with 1200/2203. 2203 is a SB error often associated with CPU overheats and/or BGA failures affecting the CPU side of the FlexIO interface. It can be a sign the CPU needs reballed. But it's also been seen when the CPU thermal monitor needs replaced. In this case it disappeared after the delid, so I wouldn't worry about it. It was just a transient error caused by the overheat.
Error 1001 is a CPU Core voltage instability. In other consoles we've seen 1001's lead up to bad RSX tokins (1002). Tokins tend to wear out at the same rate, so either the CPU or RSX tokins can go bad. You can get a bunch of 1001's if the voltage ripple/noise is teetering on the edge, but not bad enough to cause an instant YLOD. They instead occur randomly or during intense games. They will get progressively worse until it causes an instant YLOD.
We've seen 1001's lead up to RSX BGA/bump failures (3034) too. In some cases we've even seen them lead up to 20 2120 / 21 3013. This is a combination we've seen in dead CPU's. And we've seen 1001's caused by nothing more than powering the console off at the back rocker switch instead of shutting down properly. The point is that 1001's are commonly associated with other errors, and are only meaningful when considered in context.
The context of
@NO_ob's 1001's are the HDMI errors occurring upon shutdown (
90 2120). These are very different than when the CPU is first initialized (
20 2120). If they were occurring that early, it means the CPU is not able to initialize (missing a critical voltage, shorting out, or dead). Instead, a
90 2120 means the CPU is fine. It passed all the power related checks and made it to the power on state (80). However, there's is an issue communicating with the HDMI transmitter when attempting to shutdown. That usually implicated the VDDIO lines the RSX uses to communicate to the HDMI transmitter. They can be broken (BGA defect, broken trace, bumps failure). Or it could be the HDMI transmitter and/or associated SMDs (Thermistor, resistors, filter, the HDMI chip itself).
Te easiest potential cause is ripple/noise interference from bad tokins, which can prevent adequate signal integrity. There is evidence this can cause the system to hang on shutdown giving this error. We've seen 1002's associated with 90 2120 disappears when the tokins were replaced. The issue with this is that the heat from installing the tokins can hide a BGA defect which can also cause the 2120. So it looks like the console is fixed when it isn't, and the errors come back. Usually with a full blown BGA defect at that point (3034/4XXX). But if replacing tokins with tantalum doesn't fix the 2120 error, there is definitely something more serious going on.
The competing hypothesis's I'm currently forming are...
- As solder balls deform and cracks propagate the resistance increases. This increased resistance is registered as a VRM failure (1001 and/or 1002) at some point. First randomly, then more often, until the crack finally fully breaks the connection (3034/4XXX). Therefore 1001/1002 errors can be warning signs of a BGA defect before the break occurs.
- The competing hypothesis is that when people delid and do not glue the IHS back on afterwards, they are putting a great deal more stress on the BGA with every thermal cycle. The increased flexing causes the BGA to break soon after replacing tokins and "fixing" their console. 1001/1002 errors were fixed by tokin replacement, but the BGA subsequently failed from the delid. It wasn't that 1001/1002's led up to the BGA defect, it was another mod performed at the same time as the fix, and that led to an entirely different problem that just looks like the two are related (association bias).
This is just an attempt to explain why some consoles have 1001/1002 errors leading up to a BGA failure. They do not always appear to be linked. Most PS3's with BGA failuers do not have any 1001/1002 errors in the errorlog. And many consoles who's 1002's were fixed by replacing tokins have not been reported to fail with a BGA defect soon after. But we rely on unreliable data for this kind of thing, so who knows.
