820-01700 A2141 GPU No Display Repair – CA601 Carbonised Short and UA600 Jumper Fix

Fault Description

This 16-inch MacBook Pro 2019, model A2141, logic board 820-01700, came from the Ashburton / Ashwood VIC 3147 area with a customer-reported fault:

Mac suddenly died
No power
No water damage

Initial USB-C meter reading:

5V / 0A

The MacBook showed no normal power-on activity.

Case Summary

Item	Details
Device	16-inch MacBook Pro 2019
Model	A2141
Logic board	820-01700
Customer area	Ashburton / Ashwood VIC 3147
Customer fault	Mac suddenly died, no power, no water damage
Initial USB-C reading	5V / 0A
First short found	PP5V_G3S short
Short location	CA601 hot during voltage injection
Cause	Bug drop / carbonised area around CA601
Board repair	Carbonised board removed, approx. 5 mm × 5 mm area
Trace repair	3 broken surface traces repaired
Chip replaced	UA600
After first repair	USB-C 20V with current changes
Remaining fault	No display
PMU_SYS_ALIVE	1.8V OK
DISP_GCON_RESET_L	1.8V OK
GPU rails	Pulsed briefly, then dropped
Final fault	UA600 pins 23 and 30 open
Final repair	Jumpers added from pins 23 and 30 to original devices
Final result	GPU rails stable, MacBook working

Initial Diagnosis – PP5V_G3S Short

We started by measuring the major inductors and power rails. The first confirmed fault was:

PP5V_G3S shorted

Voltage injection was used to locate the short. At around 3V injection, CA601 became hot.

There was a bug drop on CA601, which is a very common type of contamination we see in Australia. In this case, it had carbonised and burned the board.

The damaged area was not just a failed capacitor. The PCB itself had become conductive because of the carbonised material.

The repair required:

removing CA601;
cutting / digging out the carbonised shorted board area, approximately 5 mm × 5 mm;
repairing 3 broken surface-layer traces;
replacing UA600.

After this repair, the MacBook no longer stayed at 5V / 0A.

USB-C Reading After First Repair

After the PP5V_G3S short and carbonised area were repaired, the USB-C meter changed to:

20V, 0.23A → 0.45A → 1.2A → 0.45A → 0.35A

This showed the board was now negotiating 20V and trying to boot.

However, there was still:

No display on screen

At this stage, it was not a simple no-power fault anymore. The diagnosis moved to the GPU power sequence.

No GPU Voltages on Inductors

Multimeter measurements showed no voltage on the GPU-related inductors:

LA710 / LA720 / LA730 / LA740 / LA750 = no voltage
LA850 / LA860 / LA870 = no voltage
LA900 / LA901 / LA950 = no voltage

But the start conditions were present:

PMU_SYS_ALIVE = 1.8V OK
DISP_GCON_RESET_L = 1.8V OK

This was an important clue. The system was alive and the display/GPU control reset was released, but the GPU rails were not staying on.

Oscilloscope Measurements

A multimeter only showed the GPU rails as missing, but the oscilloscope showed the real behaviour.

Scope measurements showed:

EG_VR0 / EG_VR1 / EG_VR2 / EG_VR3 / EG_VR4 present for about 35 ms, then drop
GPU voltages present briefly, then drop in about 30 ms
PPGFX_S0_GPU reaches target voltage, then drops in about 1 ms

This means the GPU sequence was not missing completely.

The board was starting the GPU power sequence, the rails were trying to come up, but the sequence was aborted almost immediately.

That usually points to a missing PGOOD, missing sense path, open feedback / current-sense path, broken trace, or incomplete connection around the GPU regulator control circuit.

Why We Suspected Hidden Broken Connections

The earlier CA601 repair required removal of carbonised board material. That area was burned badly enough that we had to dig out approximately 5 mm × 5 mm of PCB material.

When a carbonised short is removed, the visible top-layer damage may not be the whole problem. The repair can also expose or interrupt:

surface traces;
thin copper around pads;
via connections;
hidden connections under nearby ICs;
control or sense lines used by GPU regulators.

Because all GPU enables appeared briefly, and all GPU rails tried to start then shut down, the logic pointed away from a totally dead U9801 sequence command. It pointed more toward an incomplete regulator feedback/control path around the repaired area.

So we removed UA600 again and diode-mode checked every pin.

The key finding:

UA600 pin 23 open
UA600 pin 30 open

We connected both pins back to their original devices using jumper wires.

After that:

All GPU voltages became stable
The board worked normally

Circuit Logic

The repair logic was:

Mac suddenly died
↓
USB-C = 5V / 0A
↓
Measure major inductors
↓
PP5V_G3S short found
↓
Inject 3V
↓
CA601 hot
↓
Bug drop / carbonised board area found
↓
Remove carbonised area, repair 3 traces, replace UA600
↓
USB-C now negotiates 20V and current rises
↓
No display
↓
PMU_SYS_ALIVE = 1.8V OK
DISP_GCON_RESET_L = 1.8V OK
↓
No stable GPU voltages on LA710/720/730/740/750, LA850/860/870, LA900/901/950
↓
Scope shows EG_VR0–EG_VR4 enables appear for about 35 ms
↓
GPU rails appear briefly, then drop
↓
PPGFX_S0_GPU reaches target voltage then drops in about 1 ms
↓
GPU sequence starts but aborts
↓
Suspect broken control / sense / feedback path after carbonised board repair
↓
Remove UA600 and diode-mode all pins
↓
Pins 23 and 30 open
↓
Jumper pins 23 and 30 back to original devices
↓
GPU rails stable
↓
MacBook works

GPU Power Sequence Lesson

The key to this repair was understanding the GPU power-on sequence.

The important simplified sequence was:

Stage	Signal / Rail	Meaning
Precondition	PP3V3_S5 / PP1V8_S5	Supplies GPU control logic
Start request	PMU_SYS_ALIVE	System is alive and sequence can begin
Reset release	DISP_GCON_RESET_L	GPU/display controller logic released
VR0	EG_VR0_EN → PP3V3_S0_GPU	GPU 3.3V rail starts first
VR1	EG_VR1_EN → PP1V8_S0_GPU	GPU 1.8V rail starts
VR2	EG_VR2_EN → PP0V75_S0_GPU	GPU 0.75V rail starts
VR3	EG_VR3_EN → PPGFX_S0_GPU / PPSOC_S0_GPU	GPU graphics and SOC rails start
VR4	EG_VR4_EN → memory rails	GPU memory rails start
Final	GPU_RESET_L	GPU released after rails are stable

In this case, the sequence started but aborted. That was the clue. If the start request or reset path were missing, the enables would not appear. But because the enables appeared briefly, we knew the command sequence had begun.

The failure happened after the rail startup attempt.

Repair Timeline

Step	Result
Customer reported sudden no power	Confirmed
USB-C meter	5V / 0A
Measured inductors	PP5V_G3S short found
Voltage injection	CA601 hot
Board inspection	Bug drop / carbonised burned area found
Carbonised area removed	Approx. 5 mm × 5 mm board area cleaned out
Surface trace repair	3 broken traces repaired
UA600	Replaced
USB-C after first repair	20V, current pattern appears
New symptom	No display
PMU_SYS_ALIVE	1.8V OK
DISP_GCON_RESET_L	1.8V OK
GPU inductors	No stable voltages
Scope test	EG_VR0–EG_VR4 appear for about 35 ms
GPU rail behaviour	Rails appear briefly then drop
PPGFX_S0_GPU	Reaches target then drops in about 1 ms
UA600 removed	Pin-by-pin diode-mode test
Fault found	UA600 pins 23 and 30 open
Final repair	Jumpers added to original devices
Final test	GPU rails stable, MacBook working

Key Lesson

When a GPU rail is completely missing, it is tempting to look only for a failed regulator IC.

But in this case, the oscilloscope showed something more useful:

The GPU rails were not absent.
They were starting, then aborting.

That changed the diagnosis.

The board already had the start request and reset conditions. The GPU sequence was being commanded. The failure was in the rail completion path.

Because the board had been burned and repaired around the carbonised CA601 area, we suspected hidden open traces or missing connections. Checking UA600 pins in diode mode found the real fault: pins 23 and 30 were open.

The key lesson:

Know the GPU power sequence before replacing parts.

Final Fix

The final repair included:

PP5V_G3S short diagnosis;
CA601 hot-spot location by voltage injection;
removal of carbonised burned board area;
repair of 3 broken surface traces;
UA600 replacement;
GPU power-sequence testing by oscilloscope;
identification of short-lived EG_VR0–EG_VR4 enable pulses;
confirmation that GPU rails were starting then aborting;
UA600 pin diode-mode testing;
jumper repair for pins 23 and 30;
final GPU voltage stability test.

Final result:

16-inch MacBook Pro A2141 fully functional

If 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, data recovery, USB-C charging faults, water damage repair and component-level diagnostics.