I want to try repairing a TV. I have a soldering i...

The 2–3 day recovery is the strongest clue. It suggests a retained electrical state, abnormal rail discharge, reset/enable problem, or a component whose behavior changes with temperature or stored charge. Your steady-state voltage readings do not yet identify the fault.

Also, the numbers printed on the inductors—1R0, 2R2, 4R7, 6R8—are inductance values, not rail names. A measurement on the switching side of an inductor can be misleading with a multimeter. Identify each regulator’s output from the output capacitors, controller IC, and PCB routing.

What your current results tell us

• The LCD panel and T-con are probably not the primary cause. The operating system is apparently not reaching the point where Wi-Fi, Bluetooth, graphics, or menus initialize.
• CPU warmth only proves that at least one processor power rail is present and current is flowing. A processor held in reset, looping in boot ROM, or repeatedly crashing can still reach 60°C.
• The backlight being on proves that the backlight-enable signal was asserted at some point. It does not prove that Linux/Android or the display pipeline booted.
• Approximately 30 Ω to ground on a 0.96 V processor rail is not automatically suspicious. Modern SoCs commonly measure surprisingly low resistance.
• Slowly increasing resistance is usually the meter charging capacitors and semiconductor junctions. It is not evidence of a bad capacitor.
• The missing internal-speaker audio may be unrelated, but it could also be important if the audio amplifier is holding an I²C bus low or disturbing a shared rail.

First: make the recovery mechanism measurable

Waiting three days is inefficient. Determine what must discharge before the TV can boot again.

1. Measure rail decay after unplugging

Put the TV into its failed state, unplug it, and measure these points over time:

Time after unplugging	Mainboard input	Standby rail	3.3 V	1.8 V	DDR rail	Core rail	eMMC VCC	eMMC VCCQ
Immediately
10 seconds
1 minute
10 minutes
1 hour

You are looking for:

• A rail that remains at several hundred millivolts for an unusually long time.
• A rail that initially falls, then stabilizes at a residual voltage.
• A standby, RTC, reset, or memory rail that behaves differently from the others.
• A rail fed backward through another IC after its regulator shuts down.

Pay particular attention to:

• SoC RTC/always-on supply.
• Reset supervisor supply.
• eMMC VCC and VCCQ.
• 1.8 V standby or always-on rail.
• Any supercapacitor, battery, or large capacitor near the processor.

2. Deliberately discharge rails

After unplugging and verifying that you are working only on the isolated low-voltage secondary side, discharge suspected rails through a resistor—not a screwdriver or direct short.

A reasonable starting value is 470 Ω to 1 kΩ. Connect it from the rail to secondary ground until the rail is below roughly 50 mV.

Test one domain at a time:

1. Discharge the mainboard’s 12 V input.
2. Reconnect and test.
3. If it still fails, repeat while discharging the 3.3 V rail.
4. Then 1.8 V.
5. Then eMMC supply rails.
6. Then any always-on/RTC rail.

If discharging one particular rail makes the TV boot immediately instead of waiting days, you have localized the fault dramatically. Investigate:

• Missing/open bleeder resistor.
• Leaky or back-feeding IC.
• Reset supervisor that is not seeing a true power-on reset.
• Regulator enable pin that remains partially biased.
• eMMC or SoC I/O rail violating power-down sequencing.

Do not discharge an unknown rail until you have verified that it is not backed by a battery or supercapacitor.

Compare a successful boot and a failed boot

The most valuable diagnostic technique is not measuring random components. It is measuring the same signals during one known-good boot and one failed boot.

Create two columns:

Signal	Successful cold boot	Failed boot
Standby supply
PS_ON / main power enable
Main 12 V
5 V
3.3 V
1.8 V
DDR supply
SoC core supply
eMMC VCC
eMMC VCCQ
SoC RESET_N
PMIC PGOOD
Backlight enable
Audio-amplifier enable
I²C SDA/SCL idle state

The exact rail voltages depend on the SoC and memory parts. Typical classes include approximately:

• Core: 0.8–1.1 V.
• DDR: commonly 1.2, 1.35, or 1.5 V.
• I/O: 1.8 or 3.3 V.
• eMMC VCC: often around 2.8–3.3 V.
• eMMC VCCQ: often 1.8 or 3.3 V.

Do not rely on those typical values as specifications. Read the regulator, SoC, DDR, and eMMC part numbers and use their datasheets.

A multimeter may miss the actual failure

A DMM can show 1.00 V in both states while hiding:

• A 200-millisecond startup delay.
• Repeated voltage collapse.
• Excessive ripple.
• A reset pulse.
• Regulator hiccup mode.
• An enable line oscillating.
• A rail coming up in the wrong order.

For this fault, the next genuinely useful tool is an oscilloscope. Even a modest scope is enough for power sequencing, reset, UART, and low-speed buses. Use a ×10 probe.

Capture power-up from the moment AC or the mainboard supply is applied. Trigger from a main enable signal or a rising power rail.

Check reset before assuming bad storage

Find the SoC reset signal, usually labelled something like:

• RESET_N
• SYS_RST
• SOC_RST
• POR_N
• CPU_RESET

Trace it from the SoC to a PMIC, supervisor IC, transistor, or RC network.

On a normal boot, it should generally:

1. Remain low while supplies stabilize.
2. Transition cleanly high.
3. Remain high.

Failure patterns:

• Stays low: supervisor, PMIC power-good, rail threshold, or reset capacitor problem.
• Goes high too early: faulty supervisor or incorrect discharge of its timing capacitor.
• Pulses repeatedly: watchdog reset or collapsing rail.
• Goes high and later falls: processor crash, watchdog, or PMIC fault.

A DMM can detect a reset line that is permanently low, but not short pulses.

Also check the reset IC’s:

• Supply voltage.
• Reset output.
• Manual-reset input, if present.
• Timing capacitor.
• Power-good input.
• Any pull-up resistor.

A tiny supervisor IC or leaky ceramic capacitor can cause exactly the sort of “works only after complete discharge” behavior you describe.

Check the clock

A powered processor cannot boot without its reference clock.

Locate the main crystal or oscillator, often 24 MHz, 25 MHz, or 27 MHz. Compare successful and failed boots.

Be careful probing directly on a crystal; probe capacitance can stop oscillation. Prefer:

• A buffered clock output.
• A test pad.
• A clock input/output with a ×10 low-capacitance probe.

Interpretation:

• Clock present on successful boot but absent on failed boot: oscillator, enable circuit, load capacitors, SoC clock domain, or power sequencing.
• Clock present in both states: continue to reset and boot-storage activity.

Use the boot UART

A UART boot log is often the fastest route to the actual cause.

Look for unpopulated headers or test pads labelled:

• TX
• RX
• GND
• UART
• DEBUG
• CONSOLE

First measure the idle voltage. It may be 3.3 V or 1.8 V. Do not attach a 5 V serial adapter.

Initially connect only:

• Mainboard ground.
• Adapter RX to TV TX.

Common rates include 115200 baud, but the board may use another rate.

Capture both:

1. A successful cold boot.
2. A failed boot.

Possible outcomes:

• No UART activity and reset is low: power/reset fault.
• No UART activity, reset high, and clock absent: oscillator/clock fault.
• No UART activity with reset and clock valid: SoC, boot-mode straps, PMIC sequencing, or very early ROM failure.
• Repeated eMMC initialization errors: storage, storage supply, eMMC signal integrity, or eMMC degradation.
• DDR initialization error: DDR supply, DDR reference voltage, clocking, termination, or BGA connection.
• Kernel begins booting and then crashes: software, peripheral bus, storage corruption, or later rail failure.
• Log repeats from the beginning: watchdog or brownout reboot loop.

Do not write to the UART pins until you know what they are.

Check eMMC activity rather than guessing “bad eMMC”

Before condemning the eMMC, prove that the SoC is attempting to access it.

Measure:

• eMMC VCC.
• eMMC VCCQ.
• Clock.
• Command.
• At least data bit D0.

With a scope or logic analyzer:

• A normal boot should produce bursts of clock and command/data activity.
• No eMMC clock usually means the processor has not reached storage initialization.
• Repeated short attempts may indicate initialization failure.
• Activity that proceeds for a while and then stops may indicate a read error or later boot stage.
• A supply dip synchronized with access suggests a regulator, capacitor, or loading problem.

Do not use reflow as a diagnostic test. Reflow can temporarily change cracked joints, internal package connections, or even leakage, while also creating new faults. It destroys useful evidence.

Check whether the audio fault is connected

Because internal audio was already dead, test the audio amplifier early.

Identify its part number and locate:

• Power supply, often 5 V, 12 V, or 24 V.
• Logic supply, often 3.3 V.
• Enable or shutdown pin.
• Mute pin.
• I²C SDA and SCL.
• I²S clocks/data.

Check the I²C bus

With power on, SDA and SCL should normally idle high. Compare good and failed states.

• If SDA or SCL is permanently low, disconnect power and measure resistance to ground.
• Determine which peripherals share the bus.
• The audio amplifier, tuner, EEPROM, or another peripheral could be holding it low.

If the audio amplifier has a separate supply feed through a ferrite bead, small inductor, fuse, or zero-ohm resistor, you can isolate its supply and retest booting. Alternatively, isolate its I²C series resistors if they are clearly identified.

This is more meaningful than merely disconnecting the speakers. A defective amplifier remains electrically attached when the speakers are unplugged.

If disabling or isolating the audio amplifier changes the boot behavior, investigate:

• Shorted or leaky amplifier supply.
• Failed amplifier logic interface.
• SDA/SCL clamp or short.
• Fault output connected to the SoC.
• Shared reset or power-good line.

When measuring a class-D speaker output, remember that it is often bridged and switching. Neither speaker terminal should automatically be treated as ground.

Use controlled heating and cooling

First determine whether recovery really requires days or merely complete cooling.

After producing the failed state:

1. Unplug the set.
2. Let the whole board cool to room temperature.
3. Try again after 30 minutes, 1 hour, and several hours.
4. Record the result.

If it remains failed when fully cool, temperature alone is unlikely to explain it.

Then use localized cooling:

• eMMC.
• PMIC.
• Reset supervisor.
• Crystal/oscillator area.
• Audio amplifier.
• Individual regulators.
• SoC last, because its large thermal mass affects nearby parts.

Cool one area, reconnect power, and test immediately. A repeatable response is evidence. Confirm it in the opposite direction with controlled gentle heating.

Avoid indiscriminate heating with a large heat gun. It affects too many components and can warp the PCB.

How to analyze a suspicious regulator

Once you find a rail that differs between successful and failed boots, measure the regulator’s:

• Input.
• Enable.
• Output.
• Feedback.
• Power-good.
• Switch node with a scope.
• Resistance from output to ground while unpowered.

Interpretation:

Observation	Likely direction
Enable is low	Upstream sequencing/control fault
Enable is high, output is zero	Regulator failure, overload, or short
Output rises and repeatedly collapses	Protection/hiccup, excessive load, bad output capacitor
Output is correct but PGOOD remains low	Ripple, threshold, supervisor, or PGOOD circuit
Output is correct in both states	Continue downstream to reset/clock/data activity
Regulator input falls during startup	Main supply, connector, protection, or upstream load problem

Measure ripple directly across an output capacitor with the probe ground connection kept very short. A long probe ground lead can produce misleading switching noise.

Do not overlook shutdown behavior

A remote-power command often puts a TV into standby rather than completing an immediate hard shutdown. For testing, do not unplug it the instant the screen goes dark.

Let it remain connected for at least a minute or two after the remote shutdown, and watch:

• Current draw.
• Main supply enable.
• eMMC activity.
• Standby rail transitions.

If there is a menu option for full shutdown, use that. Repeatedly removing power while the operating system or eMMC is still active can complicate the fault and potentially cause filesystem damage.

A practical diagnostic order

1. Record the exact LED blink pattern with slow-motion video.
2. Obtain the television model, mainboard number, power-board number, and IC markings.
3. Map and label every regulator output.
4. Measure power-off rail decay.
5. Use resistor discharge to identify the rail/domain responsible for the multi-day recovery.
6. Compare successful and failed startup sequencing.
7. Check SoC reset and reference clock.
8. Capture UART output.
9. Check eMMC supply and bus activity.
10. Check SDA/SCL, especially around the failed audio amplifier.
11. Use selective cooling/heating only after the electrical comparisons.
12. Replace a component only after its inputs, outputs, and behavior prove it is responsible.

The most informative next experiment

The single best test you can perform with your existing multimeter is:

Put the television into the failed state, unplug it, record how every low-voltage rail decays, then discharge each persistent rail individually through approximately 470 Ω and see which discharge restores immediate booting.

That directly tests the unusual two-to-three-day dependency. The next most informative additions are a scope capture of RESET_N and the power rails, followed by a boot UART log.

The exact TV model and the board numbers printed on the mainboard would make it possible to map your 1R0, 2R2, 4R7, and 6R8 circuits to specific SoC, DDR, eMMC, audio, and standby rails.