820-00923 - no SSD - missing clock signal

realmacmods

Member
Working on a 820-00923. Works perfectly other than not seeing the internal SSD. Ive verified every power and enable signal related to SSD i can think of, and everything seems normal. Seem to be getting PWRGOODs on everything SSD related as well. That said, happy to provide specific measurements if i happened to miss something.

The only thing that is anomalous is the lack of the 100MHZ SSD clock signals. the clock request is low (.090v which is a not zero, but i think would still count as a logic low).
J9600 cover is present ( and replaced for testing sake). I have only been able to look for those clocks with the j9600 cover removed as i dont see any other place to measure it. Will SSD clock be present with the cover removed? Is there anywhere to measure for it on the board otherwise?



No evidence of liquid damage. It did have a Drivesavers sticker on it, so it is possible they cleaned it, but it looks very clean everywhere.
Assuming i am on the right path of the missing clock signal, is there anything other than the CPU that could be causing this? It looks like the request goes low, and the cpu should start spitting out 100MHZ, with only traces in between the CPU and j9600.

Continuity on j9600 seems good

Wifi works fine, so the cpu must be putting out the AP clock (if i understand that correctly anyway). Camera did not work in Linux, but that doesnt say much. Doing a Mac OS USB install presently to test that. I am leaning towards a bad connection from CPU ball to board, but it seems unlikely to have those connections be bad while absolutely everything else is working.

SSD not present in linux, SSD not present in Mac OS Disk Utility. Mac OS online diagnostics come back as no issues found.
 

2informaticos

Administrator
Staff member
You can't get SSD clock with J9600 disconnected.
PCH expects some confirmation that SSD is present.

Compare all SSD signals with good board.

Not sure yet if SSD must have low, or high level on clock request line.
Some people reported high level in working state; opposite to airport situation.
SSD clock must be available for check on the other side, near U9100.
 

realmacmods

Member
Ok, i found the test points. Kind of hidden under one of the NAND shields. Running a USB install presently, but when that finishes im going to check for clock there.

My thought on the clock request needing to be low is that it should normally be held high through r2080 (from 3v3_S0) until it can be taken low by a logic signal from the CPU through r2084.
 

realmacmods

Member
Ok, can confirm there is no clock signal on either of those test points near u9100. PCIE_CLK100M_SSD_LBp/n

Also my boot after the OS install on an external USB drive failed with the message of not being able to install a needed firmware update. This happened both on Sierra (OEM OS) and Ventura (Latest OS) from Internet recovery.
 

2informaticos

Administrator
Staff member
The touchbar requires internal SSD to install its firmware.
Did you try to boot in safe mode, from external macOS?

Compare the SSD control signals (reset, enable, clock request, etc) with good board.
 

realmacmods

Member
Tested every pin on u9300 just to be sure. Everything checks out as normal. i think I can almost certainly rule out a power issue with the SSD at this point

I dont have a known good board to test with at this time. What else is critical to the SSD that i can point my efforts to? U8600 looks important but is going to be difficult to accurately test given the pin count.
 

realmacmods

Member
i get elevated steady voltages, but no discernible clock signal. Solid 1.8v on XTALi and solid 0.9V on XTALO
I also noticed that SSD_BOOT_L is at a constant 1.8V. _L suggests to me it should be logic low, but you never know with these things.
 

realmacmods

Member
I just swapped y8600 with a donor board. No change. r8600, and r8602 measure correctly. I assume that this should be a simple crystal oscillator at 24mhz?

Booting Linux with y8600 removed from the board yields no change positive or negative
 
Last edited:

2informaticos

Administrator
Staff member
An o-scope should catch clear 24MHz oscillation there.
If not available, compare DC levels with working board.
 

realmacmods

Member
this clock situation may not be relevant. I just looked it up and the scope im using maxes out below 1MHZ, so there is a very good chance my measurements are irrelevant.
 

realmacmods

Member
yeah, the little scope i was using is indeed hopeless for the measurement task i was giving it. I used a better one and the 24mhz is present at the crystal. 100mhz is beyond this scope's ability but im going to order a 200mhz model. That said, i have a strong feeling the elevated DC at the 100mhz test points is my clock signal.

Any chance at all that this is a bios or firmware issue? Feels extremely unlikely, but ive yet to find anything at all wrong elsewhere.

Is there any good way to tell if u8600 is alive and functioning?
 
Top