PS3 Fault finding YLOD with the SYSCON - First steps and Error reporting
RIP-Felix
Senior Member
What are you using for the boom arm to hold the IR6000? That's the one thing preventing me from doing this project!
RIP-Felix
Senior Member
I must be a complete idiot! You would be shocked how long it took me to figure this out...
Cardboard and avoiding pads under the PSU was my previous method...lol! Now I just need to extend the 5V connector and I'll have room to work.
Sigh...we can't all be geniuses.
vyktormvmpay25
Senior Member
Sem001 board have that 5 pins cable double length,just a tip.
I've extended both half meter, it is hard to assembly every time for testing, so I test it on jig after reball.
I've extended both half meter, it is hard to assembly every time for testing, so I test it on jig after reball.
I did the same when I realised I need live probing to find the real issue. But the length of cable and tilty mainboard bothered me. Not to mention a powerlet that dangling outside the PSU. So I 3d printed an adaptor for PSU and made some extension wires. Now I feel much faster every time I setup a probing session
It's just syscon uart output. Before bringup, type "hdmi vbs ffffffff", that will turn on all the verbose ouput for hdmi. Then bringup will show those hdmi debug messages.
SkaziChris
Member
Hello Felix,
sorry for long time to reply.
I finally got my hands on RIGOL oscilloscope.
I measured the CPU VDDC during the bringup sequence.
The amplitude is 170mV... which is way above 50mV.
So.. I guess this definitely confirms a NEC/Tokin defect?
RIP-Felix
Senior Member
Any chance you can upload the STL/OBJ file so I can print one? I don't think the forum will allow it, but maybe you can put it up on thingiverse?
RIP-Felix
Senior Member
Yes that's correct.
Could you do some test before you replace the tokins? I'd like a good screen cap of the VDDC for both CPU and RSX.
First capture: Startup Behavior
- CH1 (yellow probe) = CPU tokin output side
- CH2 (Blue probe) = RSX tokin output side
- Set trigger level at about 0.8V on channel 1 (CPU).
- DC Coupling
- Bandwidth Limit = 20MHz
- Horizontal = 50ms.
- Vertical = 50mV & 20mV
- Acquire --> HiRes Mode
- Measure --> Measure All.
- Measure --> measure all --> measure all source --> check boxes for CH1 & CH2. This way it'll display measurements for both channels (CPU & RSX).
- After triggering, move the voltages down (DC offset) so they both fit under the measurement box. A DC offset of around 1.1v -1.3v is what I found was needed to bring the voltage readings into view. The more you're zoomed in the more turns it takes of the rotary knob. Just pay attention to the DC offset, slow down when it gets close to -1v and the voltage should come into view. It's easier if you start zoomed out and move your way in close one step at a time. It will probably take a few triggers to get it dialed in. Play around until you get the hang of it.
White line is the plateau you want to zoom in on for the voltage ripple/noise measurment. see below...
- CH1 (yellow probe) = CPU tokin output side
- CH2 (Blue probe) = RSX tokin output side
- Move the horizontal position over to 300ms, or about the middle of the second to last voltage plateau. It's the point in the startup sequence I marked with a white line in the last picture above. Basically it's the voltage plateau before the console finishes the power on sequence. After that the GPU will start a 60Hz pattern, which I suspect is the frame buffer. We want to see the buck converter ripple before any real load is placed on the console, and most YLOD consoles make it this far into the power on sequence, so that's why I chose it for this measurement. The ripple here should be under 50mV MAX, but is typically 20mV with good tokins. But before you can measure it accurately, you need to zoom way in on the ripple...
- So set Horizontal = 1us/div and Vertical = 50mV/div.
- The measure all can be useful here, but the cursor is most accurate for the Vpp ripple. Play with the cursor function and figure out how to measure the peak-peak voltage using it. The Measure all window is convenient, but measures transient voltage spikes (noise) in the Vpp measurement. We want ripple voltage which is the P-P of the larger waveform, not the noise spikes.
The examples above are from PS3#7, a console with bad RSX tokins.
Last edited:
Here it is:https://www.thingiverse.com/thing:5176630
I'm not an expert at designing this, especially the reinforce part at the bottom of the powerlet housing. But it's usable.
RIP-Felix
Senior Member
You're about to surpass me. I stopped buying new systems at #10.
I still fight the urge to check e-bay. Everytime I do I find a deal too good to pass up. I past up on one the other day that looked like a working A model for $40. That sick part of me said, "but what if?" NO...NO...NO! I had to stop. I told myself "they probably reflowed it!" Luckily they didn't show the bottom, where the warranty sticker was. If they had and it was intact, I'd have PS3#11 and no room for it!
Hi. I've been watch this thread for a while and thanks for all the work! I've got error codes from 3 of my boards(2 SEM-001, 1 DIA-001) successfully, which is great.
The DIA-001 board I've got is having 2031 and 2131 errors, according to error code guide it's a RSX thermal sensor(IC2101) error. The error log is:
Code:>$ errlog errlog ofst[108]:err_code:0xffffffff, clock:0x28c08b6e 2021/08/31 06:40:46 ofst[112]:err_code:0xa0082131, clock:0x28c08b7a 2021/08/31 06:40:58 ofst[116]:err_code:0xa0902131, clock:0x28c08b7b 2021/08/31 06:40:59 ofst[120]:err_code:0xa0082131, clock:0x28c08b88 2021/08/31 06:41:12 ofst[124]:err_code:0xa0082131, clock:0x28c08b95 2021/08/31 06:41:25 ofst[ 0]:err_code:0xa0902131, clock:0x28c08b95 2021/08/31 06:41:25 ofst[ 4]:err_code:0xa0a02131, clock:0x28c08ba7 2021/08/31 06:41:43 ofst[ 8]:err_code:0xa0a02031, clock:0x28c08ba7 2021/08/31 06:41:43 ofst[ 12]:err_code:0xa0a02031, clock:0x28c08ba7 2021/08/31 06:41:43 ofst[ 16]:err_code:0xa0a02031, clock:0x28c08ba7 2021/08/31 06:41:43 ofst[ 20]:err_code:0xa0a02031, clock:0x28c08ba7 2021/08/31 06:41:43 ofst[ 24]:err_code:0xa0a02031, clock:0x28c08ba7 2021/08/31 06:41:43 ofst[ 28]:err_code:0xa0a02031, clock:0x28c08ba7 2021/08/31 06:41:43 ofst[ 32]:err_code:0xa0082131, clock:0x28c08baa 2021/08/31 06:41:46 ofst[ 36]:err_code:0xa0902131, clock:0x28c08baa 2021/08/31 06:41:46 ofst[ 40]:err_code:0xa0082131, clock:0x28c08bb2 2021/08/31 06:41:54 ofst[ 44]:err_code:0xa0902131, clock:0x28c08bb2 2021/08/31 06:41:54 ofst[ 48]:err_code:0xa0082131, clock:0x28c08bc5 2021/08/31 06:42:13 ofst[ 52]:err_code:0xa0902131, clock:0x28c08bc5 2021/08/31 06:42:13 ofst[ 56]:err_code:0xa0082131, clock:0x28c08bd1 2021/08/31 06:42:25 ofst[ 60]:err_code:0xa0902131, clock:0x28c08bd1 2021/08/31 06:42:25 ofst[ 64]:err_code:0xa0a02131, clock:0xffffffff ofst[ 68]:err_code:0xa0a02031, clock:0xffffffff ofst[ 72]:err_code:0xa0a02131, clock:0xffffffff ofst[ 76]:err_code:0xa0a02131, clock:0xffffffff ofst[ 80]:err_code:0xa0a02131, clock:0xffffffff ofst[ 84]:err_code:0xa0a02131, clock:0xffffffff ofst[ 88]:err_code:0xa0082131, clock:0xffffffff ofst[ 92]:err_code:0xa0a02131, clock:0xffffffff ofst[ 96]:err_code:0xa0a02031, clock:0xffffffff ofst[100]:err_code:0xa0a02031, clock:0xffffffff ofst[104]:err_code:0xa0a02031, clock:0xffffffff [mullion]$
And the bringup sequence isn't very interesting. It stops early:
Code:>$ auth Auth successful >$ bringup bringup [SSM] state: 0000 -> 0101 Bringup Mode #0 (0xFF) [SSM] ssmCb_OnStartingBePowOn() called. [SSM] First Boot. [SSM] Bringup mode : syspm_stat=00000000/00000000 [POWSEQ] PowerSeq_Setup called. [SSM] state: 0101 -> 0301 [SSM] PowSeq Fail : Detected ! [SSM] state: 0301 -> 0700 [POWSEQ] AV Backend Letup [SSM] Shutdown mode : syspm_stat=00000000/00000000 [ERROR]: 0xa0082131 [ERROR]: 0xa0902131 >$
Luckily on this DIA-001 I can find something wrong with two capacitors seems to be on the power rail of RSX
https://imgur.com/a/mHeb2x9
And according to the service manual, though not DIA-001, the thermal sensor for RSX should be IC2101. On https://www.psdevwiki.com/ps3/Thermal#RSX it says could be a OnSemi or Ti sensor. Later I found this little chip sitting next to RSX:
https://imgur.com/gallery/MThP4cT
My question is, do I need to do any multi-meter for this thermal sensor checking before I order some replacement for the two leaky capcitors? If so, how? After all this is a thermal sensor error but could be well related to power issue. Thanks in advance!
Alright. I finally managed to fix this DIA-001. With the right tool to replace the thermal sensor IC, I can now confirm if you have 08 2131 accompanying with 90 2131, or 08 2031 accompanying with 90 2031, it's likely to be the RSX thermal sensor(IC2101) itself.
To be honest I wasn't sure at the beginning so the board has been shelved for a while. Later I did some live probing around the IC2101 and probably shorted some of the pins together. Before the live probing, the 2031 error I got was pretty late so it used to be a long YLOD. After the suspecting short, the YLOD became a instant one but the error code remains the same. In addition, I found the 3.3V thermal voltage is almost always on, that is, it's on when syscon started working but probably not generating any useful data because the RSX/CELL weren't turned on yet. So a half dead thermal IC might only generate errors after bringup a few seconds, but a totally dead thermal IC could generate errors instantly when bringup because the syscon probably already detected something wrong with it. All of these make me think the 2031/2131 error was indeed generated by the thermal sensor IC itself. To my surprise, I was right.
SkaziChris
Member
Yes that's correct.
Could you do some test before you replace the tokins? I'd like a good screen cap of the VDDC for both CPU and RSX.
You don't have to assemble the console and use RF shielding like I did. The point here is just to show how I propped up the probes during the test. Also it's really important to use ground springs!!! They cut down on the noise ALOT. Should have come with your scope. They really are needed to get good capture of the ripple.
First capture: Startup Behavior
- CH1 (yellow probe) = CPU tokin output side
- CH2 (Blue probe) = RSX tokin output side
- Set trigger level at about 0.8V on channel 1 (CPU).
- DC Coupling
- Bandwidth Limit = 20MHz
- Horizontal = 50ms.
- Vertical = 50mV & 20mV
- Acquire --> HiRes Mode
- Measure --> Measure All.
- Measure --> measure all --> measure all source --> check boxes for CH1 & CH2. This way it'll display measurements for both channels (CPU & RSX).
- After triggering, move the voltages down (DC offset) so they both fit under the measurement box. A DC offset of around 1.1v -1.3v is what I found was needed to bring the voltage readings into view. The more you're zoomed in the more turns it takes of the rotary knob. Just pay attention to the DC offset, slow down when it gets close to -1v and the voltage should come into view. It's easier if you start zoomed out and move your way in close one step at a time. It will probably take a few triggers to get it dialed in. Play around until you get the hang of it.
Second capture: CPU/GPU Core Voltage Ripple & Noise
View attachment 35614
White line is the plateau you want to zoom in on for the voltage ripple/noise measurment. see below...
- CH1 (yellow probe) = CPU tokin output side
- CH2 (Blue probe) = RSX tokin output side
- Move the horizontal position over to 300ms, or about the middle of the second to last voltage plateau. It's the point in the startup sequence I marked with a white line in the last picture above. Basically it's the voltage plateau before the console finishes the power on sequence. After that the GPU will start a 60Hz pattern, which I suspect is the frame buffer. We want to see the buck converter ripple before any real load is placed on the console, and most YLOD consoles make it this far into the power on sequence, so that's why I chose it for this measurement. The ripple here should be under 50mV MAX, but is typically 20mV with good tokins. But before you can measure it accurately, you need to zoom way in on the ripple...
- So set Horizontal = 1us/div and Vertical = 50mV/div.
- The measure all can be useful here, but the cursor is most accurate for the Vpp ripple. Play with the cursor function and figure out how to measure the peak-peak voltage using it. The Measure all window is convenient, but measures transient voltage spikes (noise) in the Vpp measurement. We want ripple voltage which is the P-P of the larger waveform, not the noise spikes.
The examples above are from PS3#7, a console with bad RSX tokins.
Hello Felix,
my measures from the "First capture: Startup Behavior" are different than those on your example.
See below.
@RIP-Felix Well I have surpassed you then as I have foun that I have 12 PS3 systems from my friend (although one system was missing it's motherboard so I am not sure it counts.) Also have 4 other systems that I had previously before I got the extra systems from my friend ( 2 that work and 2 that don't).
RIP-Felix
Senior Member
Okay, from what I can see your console is staying on for like 3 seconds before the YLOD. But you're only getting up to ~110mV on both CPU/RSX VDDC. That's confusing to me, because in every case I've seen the VDDC fail the console YLOD in less than 1.5 seconds.
I have a thought. Double check that your probes are set to 10x and the scope is set to 10x probes. If I multiply your voltage levels by ten it seems about right.
You are measuring at the tokins, yes? And do you have any of the tokins removed? Maybe a picture would be useful here.
Last edited:
SkaziChris
Member
Yes that's correct.
Could you do some test before you replace the tokins? I'd like a good screen cap of the VDDC for both CPU and RSX.
You don't have to assemble the console and use RF shielding like I did. The point here is just to show how I propped up the probes during the test. Also it's really important to use ground springs!!! They cut down on the noise ALOT. Should have come with your scope. They really are needed to get good capture of the ripple.
First capture: Startup Behavior
- CH1 (yellow probe) = CPU tokin output side
- CH2 (Blue probe) = RSX tokin output side
- Set trigger level at about 0.8V on channel 1 (CPU).
- DC Coupling
- Bandwidth Limit = 20MHz
- Horizontal = 50ms.
- Vertical = 50mV & 20mV
- Acquire --> HiRes Mode
- Measure --> Measure All.
- Measure --> measure all --> measure all source --> check boxes for CH1 & CH2. This way it'll display measurements for both channels (CPU & RSX).
- After triggering, move the voltages down (DC offset) so they both fit under the measurement box. A DC offset of around 1.1v -1.3v is what I found was needed to bring the voltage readings into view. The more you're zoomed in the more turns it takes of the rotary knob. Just pay attention to the DC offset, slow down when it gets close to -1v and the voltage should come into view. It's easier if you start zoomed out and move your way in close one step at a time. It will probably take a few triggers to get it dialed in. Play around until you get the hang of it.
Second capture: CPU/GPU Core Voltage Ripple & Noise
View attachment 35614
White line is the plateau you want to zoom in on for the voltage ripple/noise measurment. see below...
- CH1 (yellow probe) = CPU tokin output side
- CH2 (Blue probe) = RSX tokin output side
- Move the horizontal position over to 300ms, or about the middle of the second to last voltage plateau. It's the point in the startup sequence I marked with a white line in the last picture above. Basically it's the voltage plateau before the console finishes the power on sequence. After that the GPU will start a 60Hz pattern, which I suspect is the frame buffer. We want to see the buck converter ripple before any real load is placed on the console, and most YLOD consoles make it this far into the power on sequence, so that's why I chose it for this measurement. The ripple here should be under 50mV MAX, but is typically 20mV with good tokins. But before you can measure it accurately, you need to zoom way in on the ripple...
- So set Horizontal = 1us/div and Vertical = 50mV/div.
- The measure all can be useful here, but the cursor is most accurate for the Vpp ripple. Play with the cursor function and figure out how to measure the peak-peak voltage using it. The Measure all window is convenient, but measures transient voltage spikes (noise) in the Vpp measurement. We want ripple voltage which is the P-P of the larger waveform, not the noise spikes.
The examples above are from PS3#7, a console with bad RSX tokins.
About the second measurement "Second capture: CPU/GPU Core Voltage Ripple & Noise"
Here is where it gets interesting...
Notice the yellow probe (CPU tokins).
They look entirely different from standard capacitor charge-discarge cycle.
So perhaps... the CPU tokins are bad... not the RSX ones...
RIP-Felix
Senior Member
Dang you're fast. I edited my previous post hoping you wouldn't read it before I did so. I realized my brain fart after posting.
Anyway double check that your probes and scope are set to 10x. The voltage level seems like a mismatch.
Nice scope BTW!
SkaziChris
Member
Yeah, I had one probe set to 1X instead of 10X.
Here are the results when everything is set to 10X (probes and a scope).
PS. These tapes just hold the pins and provide some pressure on them.
RIP-Felix
Senior Member
RSX tokins certainly have the bad waveform.
One thing I do know is that bad CPU tokins can affect RSX tokin ripple, but as bad as that looks and with your errors, it's clear all the tokins need to go!
Before I get concerned about your voltage levels, what model PS3 is that again? They are low for 90nm Cell/RSX. Perhaps it's normal for 65/40nm models. IDK, I don't work on them.
The CPU voltage reading being that flat does concern me. I think we should do some probing (carefully not to short any pins). Fill out the following voltage test worksheet using the jumper locations guide. It's a lot of work but will help rule out other possible issues.
Also, what is the resistance of the CPU and RSX tokins to GND?
Similar threads
-
-
PS3 CECHA00 with several SYSCON errors (3004, 1001,1002, 2120,3011)
- Started by LSL
- Replies: 7
-
PS3 Interesting PS3 Errors (1802, 1701, 1601) and Capacitor Replacement- Started by Cheshire UK
- Replies: 2
-