I recently picked up a few PS3 dev kits for pretty cheap, and one doesn't seem to want to boot. After the status light becomes solid green and the power button is pressed, the HDDs/fans spin up for about one to two seconds and the GPOs flash, and then the device goes into the "system failure" LED pattern. Resetting the device through the reset button attempts seems to get it back to the initial boot state but it still enters system failure mode when the power button is pressed. Does anyone have any ideas or quick fixes for this, or do I need to start tearing it apart? Thanks
Looks like your CPU/GPU hardware damaged due to extremely high temperature on the motherboard. The only safely solution is reballing. Maybe i'm wrong but this is what i think.
Thanks for the info, I was hoping it wouldn't come to that but probably wishful thinking on my part. Interestingly, I was able to get it to boot (or hang, not sure as I didn't have any monitors hooked up) by flipping the GPI switches randomly after pressing the power button. This only happened one time though I have not been able to reproduce it with a monitor attached.
The DECR-1000A have the same issues as the retail PS3 and debugging units. I agree with SILENT_pavel that maybe the RSX/CELL has some problems. Only the DECR-1400A is safe for any ylod or system failure. The main problem is that the MB is on pressure on the gpu/cpu mounting points. Its like on the 360. There was the xclamp mod to reduce it, but that was only death by time. And as many people think its not a problem of high temp. The MB works on that mounting points on any startup. Also on temp of 30°... So any time you boot from cold, your mounting points work. Result = ylod by time. Some consoles fast, some unit after years.. Thats the point you will get a ylod on any rev of the PS3 sooner or later. Only the decr-1400 have a complete redesign for that problem. When you see a TOOL MB opend, you know what i mean. And this redesign fix any ylod problem. You maybe ask now why sony do this not on retail units? ...
I won't have any time to use them for dev work for a few months, so they are boxed up at the moment. The LED pattern is "fast flashing red" as described in the Reference Tool Instruction Manual (page 13) however. It goes into this state immediately after the full-speed fan spool up that lasts for about a second when you turn on the device. Thanks for the info, I got the unit for $100 so it is no big loss if it is toast. Once things settle down for me a bit I will dig in and try to fix it, hopefully it is something obvious. I at least got two spare HDDs out of it so I consider it a wash event if I can't fix it. I looked at getting a DECR-1400 but I couldn't find any for sale, and I am mostly interested in OS-level and hypervisor hacking so a 1000a seemed like the best choice. If you ever have one you want to sell let me know though (or a DTL-H500, or DTL-H1000, or DTL-H2000...)
If you think about selling some spare parts of the unit, im looking for a working GPO board and a vertical stand. For the decr-1400A, there was some units sold on ebay an here, maybe start a WTB thread. regards Chris
AFAIK it's the same problem that affected 8400/8600 in laptops. But it's essentially caused by a combination of heat cycling, which puts pressure on the mounting because everything expands and contracts at different rates and bad soldering after the lead was removed from solder which caused cold joints. When people were wrapping their 360's in towels and making them overheat it would slightly melt the solder again which caused it to work for a while, but it was too weak to make a permanent fix. Clamps just put everything under more pressure to try to fix it, but it's not a permanent fix either and can do more damage in the long term. My slim PS3 GPU was playing up after I left it running and the game crashed, the fans were going crazy. It wouldn't come on for a while then it booted up but 1080p was corrupted. A little while later when it was warm 1080p started acting up again (lower resolutions were generally fine). It repeated this a couple of times, but has been ok since.
You sure it comes from the CPU/GPU issue? The hardware is complete different to prevent it. And i never heard or see a ylod TOOL (decr-1400). A drive error ok, but..?
? the 1400 design is good, but ~ it near 0.1~2mm gap between heatsink with GPU and CPU, this issue still happen on PS4 .
I've had a similar experience with my DECR-1000A unit. A couple of things to try. 1. Connect a network cable the debug port and use the admin web page to run a diagnostic test. It's a basic test but it does check for things such as memory, fans, cpu boot and so forth. 2. Open the unit up and remove/reseat the CP daughterboard cards. These two cards are in the back of the unit and have a set of diagnostic LED's that light up even when the unit is in standby. In particular there was a metal frame between the CP and motherboard that I found was keeping the CP from making a firm enough connection with the motherboard. I ended up bending it just slightly the allow for a better connection. The working theory is the CP was able to communicate with the PS3 motherboard enough to power up the unit, but a broken connection to part of the connector meant the CP wasn't getting the results it wanted to so it initiated a shutdown.
The same reason Sony wouldn't put Nichicon high quality capacitors on retail boards, they didn't sell these units for several thousands of dollars, which meant they couldn't give retail/dex the same amount of quality control and components given to DECR/DEH units.
