boardrepair.me › Boards › 820-01700
Intel MacBook Pro
Board Number 820-01700
A-Number(s) A1990
Model MacBook Pro 15" 2018–2019 (Four Thunderbolt 3)
Year 2018–2019
```html

Board 820-01700 — MacBook Pro 15" 2018–2019 (Four Thunderbolt 3)

The 820-01700 is the primary logic board for MacBook Pro 15-inch models from 2018–2019 (A1990), featuring Intel sixth-generation Core i7/i9 processors paired with AMD Radeon Pro 555X or 560X discrete GPUs and Apple's T2 security coprocessor. This generation introduced higher power delivery demands (96W USB-C), tighter thermal integration, and critical T2 firmware dependencies that make no-power faults particularly challenging to diagnose without proper SMC recovery techniques.

Key Test Points & Voltages

Net / Test Point Power State Expected Voltage Notes
PPVRTC_G3H Sleep / Off 3.0–3.15 V Always-on rail from battery. Supplies RTC oscillator and initial T2 wake logic. Measure at C7220 bulk cap.
PP3V42_G3H Sleep / Off 3.35–3.50 V Gate drive and initial bias rail. Powered by ISL9239 buck converter. Critical for T2 boot sequence.
PPBUS_G3H Sleep / Off 11.5–12.5 V Main unregulated rail from ISL9239. Absence indicates charger/SMC issue or input protection fault.
PP5V_S5 S5 standby 4.9–5.1 V Secondary always-on rail for USB device power. Fault here prevents USB wake and charging detection.
PP3V3_S5 S5 standby 3.2–3.4 V Critical for T2 function. Powered by CD3215. Most common missing rail in no-power faults. Measure at L7501 inductor.
PP1V8_S3 S3 sleep 1.75–1.85 V DRAM and chipset I/O supply. Absence prevents OS resume. Buck converter U6601.
PP3V3_S3 S3 sleep 3.2–3.4 V Chipset and sensor rail. U6701 buck converter. Fault indicates SoC PMIC issue.
PPVCORE_S0 S0 active 0.8–1.0 V (dynamic) CPU core voltage. Delivered by U5500 + U5501 VRM. Dynamic rail—measure under load for accuracy.
PPVDDIO_GPU_S0 S0 active 0.9–1.0 V Radeon Pro 555X/560X VddIO supply. Shorts on this rail cause GPU failure and PP5V collapse.
PP12V_S0 S0 active 11.8–12.2 V Unregulated 12 V for GPU power distribution. Low rail indicates input regulation problem or dGPU short.
PPVDDQ_S0 S0 active 0.6–0.65 V (LPDDR4) DRAM supply. Absence prevents memory detection and boot. Measure at SODIMM area.
PP1V05_S0 S0 active 1.02–1.08 V Chipset analog supply. Low voltage here causes random freeze or graphical corruption under load.

Diode Mode Reference

Use a quality multimeter in diode mode. Place red probe to component cathode; black probe to ground plane. Healthy diodes typically read 0.45–0.65 V depending on semiconductor type and meter. Shorted devices read ≈0 V; open devices read OL (no conduction).

Test Point / Net Red (−) to GND Expected Range Common Bad Reading Notes
D7200 (PPVRTC FET gate) Gate to GND 0.50–0.60 V OL or 0.0 V Schottky gate driver. Open gate prevents RTC rail startup. Measure before L7200.
D7201 (PP3V42 Schottky) Anode to GND 0.48–0.58 V 0.0 V (shorted) Common failure point on 2018 boards. Short here collapses PP3V42 under load.
U7200 ISL9239 VSW pin VSW node to GND 0.52–0.65 V 0.0 V or floating Switch node. Diode reading checks for bootstrap capacitor and control FET integrity.
D7300 (PP5V_S5 rectifier) Cathode to GND 0.50–0.62 V OL MBR3045 Schottky. Open diode prevents 5 V standby rail. Measure at C7302.
D7400 (PP3V3_S5 sync FET body diode) Body diode (Source to GND) 0.48–0.56 V 0.0 V or OL Indicates FET body diode health. Bad reading = FET must be replaced with rail.
D7401 (CD3215 catch diode) Cathode to GND 0.51–0.61 V 0.0 V Synchronous boost catch diode. Shorted unit collapses PP3V3_S5 rail and prevents SMC firmware load.
U6601 PP1V8_S3 controller output FET switching node to GND 0.49–0.59 V OL (open control loop) Onsemi buck. OL reading indicates PWM fault or bootstrap cap failure.
D6500 (GPU VddIO input protection) Anode to GND 0.54–0.64 V 0.0 V (shorted dGPU) Common failure: dGPU shorts this diode, collapsing entire PPVDDIO_GPU rail and forcing hard shutdown.
U5500 CPU VRM upper FET G Gate to GND (at rest) 0.0 V 0.5+ V (bootstrap leakage) Gate should be discharged at power-off. High reading = bootstrap cap leakage or PWM IC failure.
C7501 PP3V3_S5 bulk cap positive lead Positive to GND 0.52–0.66 V 0.0 V or OL Large 47 µF cap. Low reading indicates internal cap short or T2 GND short at FCBGA pins.
L7501 PP3V3_S5 inductor node Output node to GND 0.53–0.65 V 0.0 V or floating Presence of voltage in diode mode confirms capacitor and rail integrity before load testing.
D5600 (PPVCORE_S0 bootstrap diode) Anode to GND 0.50–0.62 V OL Bootstrap circuit for CPU VRM. Open diode causes CPU core voltage collapse under load, thermal shutdown cycles.

Common Faults & Symptoms

1. No Power — T2 Chip Fault, PP3V3_S5 Rail Absent

Symptom: Board draws zero current on initial plug-in. No fans spin, no LED, no chime, no SMC response. All power button presses inert.

Likely Cause: CD3215 boost converter shorted or disabled by T2 via GPIO; corrupted T2 firmware; D7401 catch diode failed short; input protection circuit malfunction preventing PPBUS_G3H rise.

Diagnostic Step: Measure PPBUS_G3H at C7220. If absent (0.0 V), charger or input stage is dead—bypass ISL9239 with lab bench supply at 12 V to verify SMC responsiveness. If PPBUS_G3H present but PP3V3_S5 missing, scope U7200 boost output; check D7401 in diode mode (0.0 V = shorted). If both rails present, force SMC reset via iFixit T2 SSD removal recovery.

Fix: Replace CD3215 IC; replace D7401 if shorted; replace T2 module if unrecoverable via DFU recovery. 2018 boards frequently fail D7401 due to manufacturing defect in catch diode batches.

2. GPU Failure — PPVDDIO_GPU_S0 Short, Radeon Pro 555X/560X Defective

Symptom: Board boots fine on integrated graphics but hangs on login when dGPU initialization begins. Instant shutdown under any GPU load (web browser, video playback). Thermal camera shows dGPU region ice-cold (no power dissipation).

Likely Cause: BGA short on Radeon Pro 555X die; VddIO supply FET shorted under load; decoupling capacitor internal short on dGPU power plane.

Diagnostic Step: Boot into Recovery Mode (Cmd+R); disable dGPU via NVRAM preference to confirm integrated GPU works. Measure PPVDDIO_GPU_S0 at rest (expected 0.9–1.0 V). Under GPU load, voltage collapses to 0.0 V and PP5V sags below 4.5 V. Diode-mode probe of D6500 reads 0.0 V (shorted). Thermally image GPU area—no heat dissipation = no current flow = shorted supply or dead BGA.

Fix: Reflow dGPU BGA at 250 °C peak; if symptoms persist, GPU is internally shorted and requires AMD Radeon Pro 555X BGA replacement (micro-soldering required). Inspect U6600 dGPU power controller for cracks or missing/cracked capacitors nearby.

3. Backlight Flicker — eDP Cable Chafing on Chassis Edge

Symptom: Backlight brightness flickers 2–5 Hz, worsens at low brightness levels. Occasionally goes completely dark for 1–2 seconds then returns. Opening/closing lid can trigger flicker cessation temporarily.

Likely Cause: eDP signal cable from LCD to board has micro-cracks in internal conductors due to sharp chassis edge chafing at upper-left hinge area. High-speed eDP clock and data lines are noise-sensitive.

Diagnostic Step: Boot into safe mode (Cmd+Shift); observe flicker frequency. Does it correlate with laptop thermal expansion or hinge movement? Visually inspect eDP cable routing at chassis corners—look for white stress marks or abrasion. Under backlight boost clock scope (captured at eDP connector), look for runt pulses or clock duty-cycle distortion at flicker frequency.

Fix: Route eDP cable away from sharp edges by inserting foam shim between cable and aluminum frame. If cable sheath is visibly damaged, replace eDP cable assembly (part number A1990-EDLC-EDP-02 or equivalent). 2018–2019 batch has notoriously poor cable strain relief at hinge.

4. Thermal Throttle Under Load — Dried Thermal Paste on CPU and dGPU

Symptom: Machine runs fine for 5–10 minutes, then suddenly drops CPU clock from 3.9 GHz to 1.2 GHz under sustained load (video encoding, gaming). sysctl hw.thermal.level reports level 94 (highest). CPU and dGPU die are hot to touch but not scorching. Machine remains stable at throttled speed.

Likely Cause: Thermal interface material (TIM) between CPU/dGPU dies and heatspreader has dried, cracked, or settled unevenly after 4–5 years of thermal cycling. Thermal resistance increases from nominal 0.05 K/W to 0.15+ K/W.

Diagnostic Step: Measure CPU die temperature via sysctl hw.sensors or HWINFO. If Tj max (105 °C on i7-8750H) is reached within 10 minutes of sustained 100% load at stock clock, TIM failure is primary suspect. Open machine; visually inspect CPU and dGPU TIM surface (lift heatsink assembly carefully). Dried/cracked paste = confirmed. Apply thermal camera to heatspreader underside—localized hot spots indicate TIM gaps.

Fix: Remove heatsink; gently clean CPU/dGPU dies and heatspreader contact surfaces with isopropyl alcohol and lint-free wipes. Apply fresh Thermal Grizzly Kryonaut or Arctic MX-6 (0.5–1.0 mm pea-sized dots). Reassemble and verify Tj drop by 15–25 °C under identical load. Test sustained video encoding to confirm throttling cessation.

Diagnostic Workflow (No Power)

  1. Confirm zero current draw: Plug board into 12 V lab supply at current-limiting mode (max 0.5 A). Multimeter should read 0.0–0.02 A. If current jumps above 0.1 A immediately, input stage or T2 power domain is shorted—stop and probe PPBUS_G3H and PP3V3_S5 for short locations.
  2. Measure PPBUS_G3H at C7220 bulk cap positive: Expected 11.5–12.5 V at battery or external 12 V supply. If 0.0 V, ISL9239 or input protection circuit is dead. Trace back to charger input and voltage selector FETs (U7100, U7101). Measure gate voltage on both FETs—should toggle or remain biased ON.
  3. If PPBUS_G3H present, measure PP3V3_S5 at L7501 inductor: Expected 3.2–3.4 V. If 0.0 V, CD3215 boost converter has failed or T2 has disabled it via BOOST_EN GPIO. Measure boost converter FB (feedback) pin voltage at U7200 pin 5—should be near 0.8 V. If GND-referenced, IC output is shorted or feedback divider is open.
  4. Scope CD3215 input and output: Set oscilloscope to 200 mV/div, 100 ns/div. Input (switch node at L7200) should show 50–100 kHz square wave with 40–60% duty cycle. Output at D7401 cathode should be steady DC with minimal ripple (<50 mV pk-pk). If no switching, IC is in shutdown mode or has failed.
  5. Check T2 power domains in sequence: With PPBUS_G3H and PP3V3_S5 confirmed, measure PP3V42_G3H, PP5V_S5, PP1V8_S3. If any rail is missing, that buck converter's feedback network or switch FET has failed. Diode-mode probe the associated Schottky catch diode.
  6. Force T2 reset via SSD removal: If all rails present but no SMC response, T2 firmware is corrupted. Power off, remove battery and SSD, wait 30 seconds, reinstall. Plug battery in; listen for SMC chime. If still silent, try connecting external 12 V supply to PPBUS_G3H and repeat. If no chime after three attempts, T2 EEPROM or main SoC is defective.
  7. Verify SMC responsiveness: With board powered on lab supply, use multimeter to monitor PP3V3_S3 rail. If this rail oscillates or ramps up slowly (0–3.3 V over 5 seconds), SMC is booting. Wait 15 seconds for full initialization. If rail remains at 0.0 V, SMC initialization is hung—check T2 power domains again and consider T2 JTAG recovery or replacement.
  8. Check CPU/chipset supplies: If SMC boots successfully, measure PP1V8_S0, PPVCORE_S0, PPVDDQ_S0 under CPU load (run Geekbench). Expected values: core voltage 0.8–1.0 V dynamic, DRAM 0.6–0.65 V steady. Missing core voltage prevents CPU boot; missing DRAM voltage prevents memory POST.
  9. Final test — press power button: If all rails present and SMC initialized, depress power button and listen for fans ramping. MacOS boot chime should sound within 5 seconds. If board remains silent, SMC is initialized but power sequencer is hung—likely PPVCORE VRM issue or PP1V8_S0 stuck low. Reflow CPU VRM controller U5500 or replace if cracked.
  10. Document rail presence: Before concluding repair, create a voltage snapshot table of all rails at full power. Store for customer handoff and future diagnostics.

Related Boards

The 820-01700 shares significant schematic and power delivery architecture with its revision 820-01814 (MacBook Pro 15" 2019 refresh with improved thermal solution and updated dGPU options), but 820-01814 features a larger heatsink assembly and revised PP3V3_S5 boost converter layout to address early 820-01700 catch-diode failures. The 820-00928 (MacBook Pro 13" 2018 Two Thunderbolt 3) is electrically similar but omits the discrete GPU entirely and uses a simpler single-rail power tree; CPU VRM topology is identical, allowing cross-reference of PWM controller test points. Charger compatibility is identical across all three (96 W USB-C USBC1–USBC4 ports).

See Also

```

← All board references · A-number lookup · Print