LG OLED EAY65689411 / 423 / 422 / 425: CARLI 820 nF, MOSFET Failure and a Faulty MD6752 – DVCC Short as the Key to Diagnosis

Starting Point

The LG OLED power board family EAY65689411, EAY65689423, EAY65689422 and EAY65689425 uses, among other components, the Sanken controller MD6752. In a recent repair case the fault picture was initially a bit murky: the primary auxiliary supply was present, but the controller didn't seem to start properly and the power board stayed essentially dead.

The decisive fault eventually turned out to be an internal defect in the MD6752. A simple resistance measurement between DVCC and GND was especially helpful here — but before the controller became the prime suspect, this case already had a backstory that has repeated across several repairs of this board family (more on that further below).

The board showed no normal start sequence. Voltage was present at the MD6752's VCC pin. Initial readings were around 12.7 V, rising slowly to about 13.2 V after some time. This first suggested that the VCC supply might be too low for the MD6752 to reach its start threshold.

The measurement across the diode in the VCC path pointed the same way at first: about 14.2 V before the diode, about 13.2 V after it (at VCC) — a drop of roughly 0.98 V. Relatively high, but not clearly faulty on its own.

Comparing with a working board delivered the key insight: there too, VCC at the MD6752 measured only about 12.3 V during operation. That made it clear the 12–13 V at VCC was not the actual cause. The MD6752 needs a higher voltage to start, but thanks to its UVLO hysteresis it can keep running on a noticeably lower VCC once started.

The Internal Supply Rails Looked Bad

The more interesting readings came from the controller's internal supply pins. These values made it fairly clear the controller had not powered up cleanly. AVCC and DVCC should normally sit around 3.3 V, VCORE around 1.8 V. Instead, practically all internal supply rails had collapsed.

At this point an external fault could still have been suspected — a failed transistor, capacitor, or faulty supporting circuitry around the MD6752. Then came the decisive measurement.

The Key: Measuring DVCC Against GND

The board was fully powered down and the capacitors discharged. The resistance between MD6752 Pin 10 (DVCC) and MD6752 Pin 8 (GND) was then measured.

The result on the faulty board: only about 7 Ω. Highly suspicious. The same measurement on a working board, for comparison, showed a resistance in the megaohm range. That made it clear: the DVCC rail on the faulty board had a massive low-resistance path to GND.

Could It Just Be the DVCC Capacitor?

A small decoupling capacitor sits directly on DVCC. MLCC ceramic capacitors in particular can short internally and pull an entire supply rail down to ground. So the capacitor was desoldered and the DVCC → GND measurement was repeated.

The short remained. That effectively ruled out the external capacitor as the cause. The low-resistance path had to originate either from another component on the DVCC rail, or internally from the MD6752 itself. Given how clearly the comparison with a working board diverged, suspicion fell heavily on the controller.

MD6752 Replaced – Fault Resolved

The Sanken MD6752 was replaced. Afterwards, the short between DVCC and GND was gone, and the power board worked again.

That confirmed the diagnosis: the MD6752 had an internal defect on its DVCC supply, pulling down the 3.3 V digital rail. The abnormally low readings at AVCC, DVCC and VCORE were therefore not separate individual problems, but consequences of the same controller defect.

Why This Fault Easily Points the Wrong Way

What makes this case particularly interesting is that the VCC voltage looked very suspicious at first. 12.7 to 13.2 V on the MD6752 can initially look like the external VCC supply simply isn't reaching a high enough level. Comparing with a working board, though, showed that even there, only about 12.3 V was present during normal operation.

One could easily have spent a lot of time on the VCC diode, the auxiliary supply, resistors, or the VSEN path. Only the resistance measurement of the internal supply rails brought clarity.

Recurring Failure Pattern: CARLI 820 nF → MOSFET Damage → MD6752

Across several of these LG OLED power boards, the same conspicuous pattern showed up before the actual controller fault: a CARLI film capacitor rated 820 nF had noticeably weakened. Afterwards, the associated power MOSFETs failed or physically ruptured. Only after replacing the obviously destroyed power components did a further consequential fault sometimes appear: the Sanken MD6752 no longer started correctly — exactly the picture described in detail above.

The wording "possible consequential damage" matters here: the repair cases show a very clear recurring correlation. Without capturing the signals at the exact moment of failure, it can't be proven beyond doubt whether the weakened 820 nF capacitor alone causes the MOSFET failure, or whether further aging effects are also involved. For practical repair work, though, this pattern is still very valuable.

• CARLI 820 nF loses capacitance
• resonant/power stage operates outside its intended operating point
• increased stress on the power MOSFETs
• MOSFET failure
• possible electrical kickback into driver/controller circuitry
• MD6752 sustains collateral damage
• DVCC becomes internally low-resistance
• controller no longer starts

Don't Just Replace the Ruptured MOSFETs

Anyone repairing an EAY65689411 / EAY65689423 / EAY65689422 / EAY65689425 by replacing only the failed MOSFETs should definitely also measure the CARLI 820 nF film capacitors. A capacitor can look completely unremarkable from the outside while having already lost significant capacitance.

If new MOSFETs are fitted while a noticeably weakened capacitor remains in the power/resonant path, there's a real risk the actual root cause of the failure was never addressed. The final DVCC check matters here too, since a damaged MD6752 doesn't have to show any visible signs.

• measure CARLI 820 nF for actual capacitance
• check the power MOSFETs
• check gate resistors and driver circuitry
• then measure DVCC → GND on the MD6752

Fast Diagnostic Path for MD6752 Power Boards

For an EAY656894xx power board with an MD6752 that doesn't start at all, or produces no gate drive, this is the sequence worth recommending by now:

• power down the board and safely discharge the electrolytic capacitors
• check the CARLI 820 nF capacitor and the power MOSFETs — a weak capacitor and MOSFET damage frequently occur together in this board family
• measure DVCC (Pin 10) → GND (Pin 8) and compare against a working board — a few ohms is highly suspicious, a good board sits in the megaohm range
• likewise compare AVCC → GND and VCORE → GND
• if DVCC is low-resistance, first remove or isolate the decoupling capacitor on that rail
• if the short remains afterward, the MD6752 itself becomes a strong suspect
• only after that does it make sense to dig deeper into VSEN, feedback, PFC, or LLC regulation

Conclusion

On this LG OLED power board, the seemingly low VCC voltage was not the actual root cause — it was an internal short inside the Sanken MD6752, most often a consequential fault at the end of a chain that starts with a weakened CARLI 820 nF capacitor and MOSFET failure.

The single most important measurement of this entire repair case was remarkably simple: DVCC against GND — a few ohms on a faulty board, megaohms on a working one. The short persisted after removing the DVCC decoupling capacitor; replacing the MD6752 fully resolved the fault.

Especially with the EAY65689411, EAY65689423, EAY65689422 and EAY65689425 power boards, it pays off to check the internal supply rails for a short very early when an MD6752 looks "dead" — and, after a prior MOSFET replacement, not to overlook the CARLI 820 nF capacitors. A resistance measurement at Pin 10 can get you to the answer far faster than a long search through the actual LLC or feedback circuitry.

Safety Note

All measurements described here are, in part, on the primary side of a mains-powered switch-mode power supply. Perform resistance and diode measurements only with the mains plug removed and the high-voltage electrolytic capacitors safely discharged. For voltage measurements, Pin 8 of the MD6752 is Primary GND — it must not be confused with Secondary GND or mainboard GND.