820-01700 A2141 AMD GPU Disable Repair – PPBUS_G3H Short and Data Recovery
Fault Description
This 16-inch MacBook Pro 2019, model A2141, logic board 820-01700, came from the Berwick / Harkaway VIC 3806 area.
The customer reported:
Mac suddenly died
No power
No water damage
Already checked by another repair shop
Customer was told it was not repairable
Customer was also told all data was goneInitial USB-C reading:
5V / 0AThis is a typical “dead board” symptom, but in this case the main goal was not just repair. The customer needed the data recovered.
Case Summary
| Item | Details |
|---|---|
| Device | 16-inch MacBook Pro 2019 |
| Model | A2141 |
| Logic board | 820-01700 |
| Customer area | Berwick / Harkaway VIC 3806 |
| Fault | No power, no water damage |
| Previous diagnosis | Another shop said not repairable and data was gone |
| Initial USB-C reading | 5V / 0A |
| Main rail fault | PPBUS_G3H shorted |
| Shorted GPU rails | LA850, LA860, LA870, LA950, LA750 short to ground |
| Conclusion | AMD GPU and VRAM section failed |
| Repair option accepted | Dedicated AMD GPU disable repair |
| Display path used after repair | Intel graphics / Intel eDP |
| Backlight method | Hardware backlight restored, software brightness dimming |
| Final result | MacBook working, data accessible |
Initial Diagnosis – PPBUS_G3H Short
The first important finding was:
PPBUS_G3H shortedFurther checks showed multiple AMD GPU-related rails were also shorted to ground:
LA850 short to ground
LA860 short to ground
LA870 short to ground
LA950 short to ground
LA750 short to groundThese rails are related to the AMD GPU and its video RAM power section.
After removing:
LA850
LA860
LA870
LA950
LA750the PPBUS_G3H short was gone.
This confirmed the short was not in the charger circuit or the main PPBUS generation path. The load was on the AMD GPU / VRAM side.
Why GPU Disable Was Chosen
At this point, the AMD GPU and video RAM section was not healthy. A full repair would mean dealing with the failed GPU / VRAM side, which may not be practical or cost-effective.
The customer accepted the dedicated GPU disable repair option.
This repair does not restore the MacBook to original factory GPU function. Instead, it disables the faulty AMD graphics section and forces the MacBook to use the Intel graphics path.
For this case, it was a good option because:
- the customer mainly needed data recovery;
- the original logic board could be made to boot again;
- the AMD GPU/VRAM short was removed from the main power path;
- the system runs cooler without the AMD GPU;
- battery runtime is usually longer;
- full hardware backlight can still be restored.
Pros and Cons of AMD GPU Disable Repair
Pros
- Can recover data from a MacBook that was previously considered dead.
- Avoids full logic board replacement.
- Lower heat because the AMD GPU is no longer running.
- Longer battery runtime in many cases.
- Good for normal use: documents, web, email, office work, browsing and basic tasks.
- Can keep the original board and data intact.
Cons
- Not ideal for gaming.
- Some external-display workflows may be limited.
- Brightness control is handled by software dimming after full hardware backlight is restored.
- It is a practical repair option, not a factory-original repair.
GPU Disable Procedure
The working logic was:
AMD GPU power rails disabled
↓
Fake VR PGOOD sequence retained
↓
U9850 forced permanently to Intel eDP
↓
Intel panel-power request directly controls panel power
↓
Intel backlight-enable directly controls U8400 enable
↓
BKLT_PWM_MLB2TCON pulled high through 1 kΩ
↓
Full hardware backlight restored
↓
Brightness Slider provides software dimmingThe important idea is that the system must still “see” a valid display/GPU power-good condition, even though the faulty AMD GPU power rails are no longer used.
If the fake PGOOD / routing logic is not done correctly, the Mac may power on but still have no image, no backlight, or unstable display behaviour.
Repair Timeline
| Step | Result |
|---|---|
| Customer reported no power | Confirmed |
| USB-C meter | 5V / 0A |
| PPBUS_G3H checked | Shorted |
| GPU / VRAM rail check | LA850, LA860, LA870, LA950 and LA750 short to ground |
| Removed shorted GPU rail inductors | PPBUS_G3H returned |
| Diagnosis | AMD GPU and VRAM section failed |
| Customer option | Accepted GPU disable repair |
| AMD GPU rails | Disabled |
| GPU PGOOD logic | Fake VR PGOOD sequence retained |
| Display route | U9850 forced to Intel eDP |
| Panel power | Intel panel-power request controls panel power |
| Backlight enable | Intel backlight-enable controls U8400 enable |
| Brightness PWM | BKLT_PWM_MLB2TCON pulled high through 1 kΩ |
| Brightness control | Brightness Slider software dimming |
| Final result | MacBook boots, display works, data recovered |
Key Lesson
This case shows that data recovery is not always lost just because the dedicated GPU has failed.
If the AMD GPU / VRAM section is shorting PPBUS_G3H, the MacBook may look completely dead. But once the failed GPU power rails are isolated, the Intel graphics path can sometimes be used to restore boot and display output.
The key is understanding the relationship between:
PPBUS_G3H
GPU power rails
U9801 GMUX / PGOOD logic
U9850 display path selection
Intel eDP output
Backlight enable and PWMBlindly replacing parts is not enough. The GPU disable repair depends on controlling the power-good logic and display routing correctly.
Final Fix
The final repair included:
- confirmed PPBUS_G3H short;
- confirmed multiple AMD GPU / VRAM rails shorted to ground;
- removed LA850, LA860, LA870, LA950 and LA750 to isolate the failed GPU section;
- retained fake GPU VR PGOOD logic;
- forced U9850 permanently to Intel eDP;
- routed Intel panel-power request to panel power;
- routed Intel backlight-enable to U8400 enable;
- pulled BKLT_PWM_MLB2TCON high through 1 kΩ;
- restored full hardware backlight;
- used Brightness Slider for software dimming;
- confirmed MacBook working and data accessible.
Final result:
MacBook working
Data recovered
System running cooler
Battery runtime improvedIf you are interested in the deeper diagnostic side of Mac repair, we have more real logic board fault cases documented here:
Mac Logic Board Repair Case Studies
Need Board-Level Mac Repair?
We specialise in MacBook logic board repair, USB-C charging faults, CD3215/CD3217 controller replacement, water damage repair and component-level diagnostics.
