MacBookPro 11.3 doesn't start unless power is pushed for ~8 secs,kernel_task high CPU, fresh install

[Camicia]

#1

Hi,

I hope that somebody has seen it before and hopefully they have some recommendations. What can I try next?

The problem:

In order to start my MacBookPro Retina( 11,3 - Late 2013 ) I need to keep the power button until the chime sound is completely finished. The release needs to be timely. If I release it before the computer shut down. If I keep it pressed too long it shut down like if you do a "hard reset" when the computer is running. It is also not able to recover from sleeping.

When it boots it is very slow. The CPU on kernel_task is 400% - 1200%. At boot time it is also very slow and the animations on the drive icons if you start with CMD+R are also super slow.

The fans are quite but I am sure they are working because I can turn them up with the temperature gadget. Other temperature sensors also seems Ok.

The problem started on High Sierra that I had been running for a little less than a year.

Mitigation:

Interestingly enough, if I open it, physically unplug the battery and I start with only the power adaptor, it works well.

I can push the power button for only 2 seconds, the kernel_task is less than 5% and the laptop is running fast as normal. If I wanted to transform it in a desktop that might work... but I obviously want it to move around.


What I tried so far:

• Reset SMC and clear NVRAM (multiple times).
• Start the Diagnostic with CMD+D. It returned "No issues found. ADP000".
• I wiped the HD (including EFI partition) and run an Internet restore that installed Maverick. The problem persisted. Interesting facts: it went from "About 1,056,366,442 hours and 38 minutes remaining" to "0" in a few seconds... Eventually it went back to 20 minutes but it took probably 4 hours to get slowly down to 0 and finish the installation. No improvement.
• I wiped again and installed Mojave. This time I did it with the battery unplugged. It took about 30 minutes. The trackpad was working but not recognized in "Settings - Trackpad" It was looking for bluetooth trackpads. When I plugged back the battery and rebooted, the trackpad started showing up but I started having again the 400% CPU on kernel_task and needing to press on power for longer than usual. And again it was not resuming from sleeping.
• I went to the Genius bar. They run tests and tried the "manual" SMC reset disconnecting the battery that I did already and it fixed nothing. They gave me the standard answer: it is probably a problem with the logic board, it is going to be $575+TAX (ouch!), it will took about 10 days and it has a 30 or 90 days warranty after the repair. I am not sure if it is completely guaranteed. They says that I will get back a working computer. Their tool did not report any "common" problems. The person there did not want me to take pictures. They run 3 tests (general, sensor and another one that I do not remeber). None of the temperatures in the report seemed off but there were:
  • a generic MLB F:050000 code - I think it was relative to the battery but I could not find anything on Google.
  • A failed test: TIM Test 75 percent Gardband. - Check that CPU heatsink is properly attached, using product spec theta dh + 75 percent guardband - on Google I did not see anything very actionable. It is a stress test and considering that the temperatures were normal, I would not expect any improvement changing the thermal paste or the heatsink. Am I wrong?
• The BIOS and SMC were supposedly updated to the latest version but I forced flashed the BIOS first and of all 3 parts of the SMC firmware (one by one) later. I though that some internal state may had been out of wack and reflashing them could reset it. I tried before with rEFInd but I was not able to run the SMCFlasher.efi so I used the OS commands: /usr/libexec/efiupdater --force-update and/usr/libexec/smcupdater
  I saw only the output of the commands in the OS setting up the flashing and the final bless command. On restart I could not see the EFI execution's output. By the way do you know how you can see the output? 
  The only significant difference was that when I restarted Mojave, the CPU was 400% but it dropped to 100% after a few minutes on the first reboot after SMC flash. I think it eventually went up again... for sure, it was back to ~400% after a reboot. I did not have the courage to try to downgrade the SMC and risk to brick the SMC that I am not sure how to flash if the computer stop booting.
• On Mojave, I tried to eliminate all .plist in ACPI_SMC_PlatformPlugin.kext as described in https://forums.macrumors.com/threads/how-to-solve-kernel_task-high-cpu-usage.1706948/ since my model was not there... no improvement.
• I read the BIOS from the MX25L6406E chip with a Raspberry Pi, replacing the ME region copying the whole ME Region from the SCAP BIOS file (the current version MBP112.0146.B00 ), and flash it back. After some googling that it is how I understood ME cleaning should be done. I am not completely positive that it is correct. But anyway... no changes.
• I run EtreCheckPro free version when Maverick was installed. It said that the test was running slow and some think else that did not help a lot. I can re run it or provide the output if necessary.

Did you see this problem before? How was it fixed?


Description of the system:

It has an NVIDIA GeForce GT 750M. In the "secret" NRAM variables

Some general info:
<key>SMC_version_system</key>
<string>2.19f12</string>
<key>_name</key>
<string>hardware_overview</string>
<key>boot_rom_version</key>
<string>MBP112.0146.B00</string>
<key>cpu_type</key>
<string>Intel Core i7</string>
<key>current_processor_speed</key>
<string>2.3 GHz</string>
<key>l2_cache_core</key>
<string>256 KB</string>
<key>l3_cache</key>
<string>6 MB</string>
<key>machine_model</key>
<string>MacBookPro11,3</string>
<key>machine_name</key>
<string>MacBook Pro</string>
<key>number_processors</key>
<integer>4</integer>
<key>packages</key>
<integer>1</integer>
<key>physical_memory</key>
<string>16 GB</string>

"BoardID" : "Mac-2BD1B31983FE1663"
The logic board is a 820-3787-A that is the version of the 820-3662 with the extra NVIDIA GPU.

Battery details (since it works without the battery I wonder if this may help):
<array>
<dict>
<key>AdapterInfo</key>
<integer>0</integer>
<key>Amperage</key>
<integer>-904</integer>
<key>AppleRawCurrentCapacity</key>
<integer>5032</integer>
<key>AppleRawMaxCapacity</key>
<integer>7279</integer>
<key>AvgTimeToEmpty</key>
<integer>334</integer>
<key>AvgTimeToFull</key>
<integer>65535</integer>
<key>BatteryData</key>
<dict>
<key>CycleCount</key>
<integer>741</integer>
<key>DesignCapacity</key>
<integer>8440</integer>
<key>QmaxCell0</key>
<integer>20767</integer>
<key>QmaxCell1</key>
<integer>18975</integer>
<key>QmaxCell2</key>
<integer>29471</integer>
<key>ResScale</key>
<integer>0</integer>
<key>StateOfCharge</key>
<integer>17920</integer>
<key>Voltage</key>
<integer>11562</integer>
</dict>
<key>BatteryFCCData</key>
<dict>
<key>DOD0</key>
<integer>0</integer>
<key>DOD1</key>
<integer>0</integer>
<key>DOD2</key>
<integer>0</integer>
<key>PassedCharge</key>
<integer>0</integer>
<key>ResScale</key>
<integer>0</integer>
</dict>
<key>BatteryInstalled</key>
<true/>
<key>BatteryInvalidWakeSeconds</key>
<integer>30</integer>
<key>BootPathUpdated</key>
<integer>1514809090</integer>
<key>CellVoltage</key>
<array>
<integer>3854</integer>
<integer>3866</integer>
<integer>3842</integer>
<integer>0</integer>
</array>
<key>ChargerData</key>
<dict>
<key>ChargingCurrent</key>
<integer>0</integer>
<key>ChargingVoltage</key>
<integer>0</integer>
<key>NotChargingReason</key>
<integer>1</integer>
</dict>
<key>CurrentCapacity</key>
<integer>5032</integer>
<key>CycleCount</key>
<integer>741</integer>
<key>DesignCapacity</key>
<integer>8440</integer>
<key>DesignCycleCount70</key>
<integer>65535</integer>
<key>DesignCycleCount9C</key>
<integer>1000</integer>
<key>DeviceName</key>
<string>bq20z451</string>
<key>ExternalChargeCapable</key>
<false/>
<key>ExternalConnected</key>
<false/>
<key>FirmwareSerialNumber</key>
<integer>1</integer>
<key>FullyCharged</key>
<false/>
<key>IOGeneralInterest</key>
<string>IOCommand is not serializable</string>
<key>IOObjectClass</key>
<string>AppleSmartBattery</string>
<key>IOObjectRetainCount</key>
<integer>6</integer>
<key>IORegistryEntryName</key>
<string>AppleSmartBattery</string>
<key>IOReportLegend</key>
<array>
<dict>
<key>IOReportChannelInfo</key>
<dict>
<key>IOReportChannelUnit</key>
<integer>0</integer>
</dict>
<key>IOReportGroupName</key>
<string>Battery</string>
</dict>
</array>
<key>IOReportLegendPublic</key>
<true/>
<key>IOServiceBusyState</key>
<integer>0</integer>
<key>IOServiceBusyTime</key>
<integer>239684</integer>
<key>IOServiceState</key>
<integer>30</integer>
<key>InstantAmperage</key>
<integer>-1043</integer>
<key>InstantTimeToEmpty</key>
<integer>233</integer>
<key>IsCharging</key>
<false/>
<key>LegacyBatteryInfo</key>
<dict>
<key>Amperage</key>
<integer>-904</integer>
<key>Capacity</key>
<integer>7279</integer>
<key>Current</key>
<integer>5032</integer>
<key>Cycle Count</key>
<integer>741</integer>
<key>Flags</key>
<integer>4</integer>
<key>Voltage</key>
<integer>11562</integer>
</dict>
<key>Location</key>
<integer>0</integer>
<key>ManufactureDate</key>
<integer>17253</integer>
<key>Manufacturer</key>
<string>SMP</string>
<key>MaxCapacity</key>
<integer>7279</integer>
<key>MaxErr</key>
<integer>1</integer>
<key>OperationStatus</key>
<integer>58435</integer>
<key>PackReserve</key>
<integer>200</integer>
<key>PermanentFailureStatus</key>
<integer>0</integer>
<key>PostChargeWaitSeconds</key>
<integer>120</integer>
<key>PostDischargeWaitSeconds</key>
<integer>120</integer>
<key>Temperature</key>
<integer>3020</integer>
<key>TimeRemaining</key>
<integer>334</integer>
<key>Voltage</key>
<integer>11562</integer>
</dict>
</array>
<key>IORegistryEntryName</key>
<string>AppleSmartBatteryManager</string>
<key>IOServiceBusyState</key>
<integer>0</integer>
<key>IOServiceBusyTime</key>
<integer>401772</integer>
<key>IOServiceState</key>
<integer>30</integer>
<key>IOUserClientClass</key>
<string>AppleSmartBatteryManagerUserClient</string>
</dict>
</array>
<key>IORegistryEntryName</key>
<string>AppleECSMBusController</string>
<key>IOSMBusSmartBatteryManager</key>
<true/>
<key>IOServiceBusyState</key>
<integer>0</integer>
<key>IOServiceBusyTime</key>
<integer>5523372</integer>
<key>IOServiceState</key>
<integer>30</integer>
<key>_SBS</key>
<integer>1</integer>
</dict>
</array>

[Camicia]

Update:The Charger BMON sensor (ampere drawn from the battery) is always reporting 18A ... If the SMC thinks that the battery is constantly sending 18A it thinks that something is wrong.

 

How can I fix it? I was looking at the schematic 820-3787-A and the SMC is driven by the ISL6359 chip. I checked the resistors between SMC and the ISL6359, so the problem must be around the ISL6359. Did anybody fixed this problem before?