"It doesn't turn on" is the end of the chain, not the start

A laptop comes in dead. Press the power button, nothing — no fan, no backlight, no LED.

The symptom-led approach is to guess and substitute. Try another charger, try another battery, reseat the RAM, try a known-good screen. Sometimes it works. Mostly it burns an hour and ends with the board swapped out, which turns one failed component into a whole-unit write-off.

The problem with "it doesn't turn on" is that it is not a symptom of anything specific. It is the final state of a sequence that stopped somewhere. What you want to know is where.

The board boots in a defined order

Nothing about power-on is simultaneous. There is a sequence, and it is documented in the schematic:

1. Input voltage arrives and is validated — the charger's ID is read, the adapter is accepted.
2. Always-on rails come up. These live whenever the board has power at all, including with the lid shut.
3. The embedded controller wakes on those rails and watches the power button.
4. The button is pressed. The EC begins the sequence.
5. Enable signals assert in order, each bringing up its rail.
6. Each regulator returns a power-good signal.
7. Only once the chain of power-good signals is complete does the CPU come out of reset and start executing.

Every stage gates the next. A rail that never comes up means its enable never asserted, or the regulator failed, or something downstream is pulling it to ground. The board is not being mysterious. It stopped at a specific step, and it will show you which one.

Working forward instead of backward

So the question stops being "why won't it turn on" and becomes "how far did it get".

Measure the always-on rails first. Present? Then input and the first stage are fine, and the entire charger-and-adapter branch of guesses is eliminated — without swapping anything.

Then press the button and watch the enables. Either the EC asserted them or it did not. If it did not, the problem is upstream: the EC, its own supply, its clock, or the button circuit itself. If it did assert and the rail still did not come up, the problem is that regulator or its load.

Each measurement halves the board. It usually takes surprisingly few to land on one node.

Bench power supply as an instrument

Connecting the board to a bench supply set to the correct voltage with a current limit turns the current reading into a live diagnostic, before anything has been measured with a probe.

A board drawing nothing is not starting the sequence at all. A board drawing a few hundred milliamps and sitting there is powered but not sequencing. A board that slams straight into the current limit has a short somewhere, and there is no point examining the sequence until that is found.

That last case has its own method. Injecting a low voltage into the shorted rail at a modest current pushes real power into whatever is faulty, and the faulty part is the thing that gets warm. A thermal camera, or in a pinch isopropyl alcohol and watching where it flashes off first, localises it physically. Often it is a ceramic capacitor that has cracked and gone short — a component worth pence, on a rail that had taken the entire machine down.

Why this matters commercially

I do this inside an IT asset disposition operation, and the diagnostic method feeds a decision that is not really about repair at all.

Every device that arrives gets a grade, and the grade determines whether it is refurbished and resold, broken for parts, or scrapped. That call is bounded by what can actually be fixed. A workshop that can only replace boards has to grade a dead board as scrap, because for them it is. A workshop that can find and replace one shorted capacitor grades the same device as refurbishable.

The capability and the grading standard are the same decision. Knowing what is repairable is what makes the grade mean anything — which is why the person setting the grading standard should be the person who can do the repair.

The habit worth building

Resist starting at the symptom. "It doesn't turn on" invites you to think about the end state, and the end state is the one place on the board where there is no useful information.

Start at the beginning of the sequence and walk forward until it stops. The board will tell you where. It has been telling you all along.