Look at the area around a laptop's processor and you will see a row of identical square inductors, a power stage next to each, and banks of capacitors spread around and under the chip. That is the processor's core regulator, built from several parallel phases. Each phase is a buck converter of its own, and the controller runs them in turn so they share the load. Understanding how the phases, inductors and capacitors work together turns that crowded area into a set of repeatable measurements — and makes faults stand out by comparison.
Why several phases
A processor core can draw tens of amps at a voltage around one volt. A single converter delivering all of that would need a huge inductor and MOSFETs, run hot and produce large ripple. Splitting the load between phases:
- spreads the heat across several power stages and inductors;
- reduces the ripple at the output, because the phases switch at different moments and their ripples partly cancel;
- improves response to sudden load changes.
The concept is described under multiphase vrm, and the regulator as a whole in The Laptop CPU VCore Circuit Explained.
How phases share the work
The controller starts each phase's switching cycle at a different moment — interleaving. With two phases, they are half a cycle apart; with four, a quarter. The controller also measures each phase's current and adjusts its timing so that every phase carries a fair share. A phase that carries much more or much less than its siblings is a sign of a fault.
At light load, many controllers switch off some phases — phase shedding — and run with only one or two. That saves power at idle and means not every phase switches all the time.
Power stages
Each phase has its switches. On modern boards these are often integrated power stages: one package containing the high-side and low-side MOSFETs, the gate driver and often a current-sense output. Typical pins:
| Pin function | What it does |
|---|---|
| Input (VIN) | The system rail |
| Switch node (SW, PHASE) | Output to the inductor |
| PWM input | Switching command from the controller |
| Driver supply (VCC) | Power for the internal driver, often 5 V |
| Boot | Bootstrap for the high-side driver |
| Current sense / monitor | Reports the phase current to the controller |
| Enable or tri-state | Lets the controller switch the phase off |
On older boards, separate dual MOSFETs and a driver chip do the same job. How a single phase works is explained in Buck Converters on Laptop Boards: How a Rail Is Made.
Phase inductors
Each phase has an inductor between its switch node and the common output. Core inductors are chosen for:
- low resistance, to avoid wasting power at high current;
- high saturation current, so the inductance does not collapse at peak current;
- compact size.
Some designs use coupled inductors — two windings in one body — to improve response. Inductors are described under power inductor. Their output pads are excellent measurement points for the core rail; their switch-node pads show each phase separately.
Output capacitors
The core rail needs a lot of capacitance close to the processor:
- polymer bulk capacitors near the inductors hold the rail through load steps, described under polymer capacitor;
- many ceramic capacitors around and under the processor — often on the bottom side of the board — handle the fastest changes.
These capacitors are numerous and sit under the processor's heatsink and the processor itself, which is why a shorted ceramic capacitor on the core rail can be hard to find by eye.
Measuring phase by phase
With the laptop running:
- switch nodes — with an oscilloscope, each phase shows a square wave, offset in time from its neighbours; with a meter, each reads roughly the output voltage as an average;
- current balance — a thermal camera under load shows whether all phases run at similar temperatures.
With power removed:
- switch node to ground and to the input for each phase, compared with its siblings;
- power stage pins in diode mode, compared with the same pins on the other phases.
Comparison is the key technique: phases are identical by design, so a phase that measures differently is the suspect. Phase timing and ripple for a given inductor can be estimated with Buck Converter Calculator (Duty Cycle and Inductor Ripple).
Common failures
| Failure | How it shows |
|---|---|
| Shorted power stage (switch node to ground) | Core rail or switch node shorted; controller in protection |
| Shorted power stage (input to switch node) | Switch node at system rail voltage; can damage the processor |
| Power stage not switching | One phase cold under load; others overloaded |
| Shorted ceramic capacitor | Core rail shorted; found by injection |
| Degraded polymer capacitor | Instability or shutdowns under sudden load |
| Cracked inductor | Noise, intermittent faults, sometimes a hot phase |
Isolating a short
When the core rail is shorted:
- Lift each phase inductor in turn, or all of them, to separate the power stages from the rail.
- A switch node that stays shorted on the power-stage side points at that stage.
- If the rail side stays shorted with all inductors lifted, the short is in the capacitors or the processor.
- Inject a low voltage — at or below the core's normal level — and look for the warm part.
Current sensing and balance
The controller needs to know how much current each phase carries. Smart power stages report it directly through a current-monitor pin; older designs sense it across the inductor's own resistance with a small resistor and capacitor network, or across the low-side MOSFET. If a phase's current-sense signal is wrong — a cracked sense resistor, a damaged network capacitor — the controller miscalculates that phase's share. It may load the phase too heavily or too lightly, or trip its overcurrent protection under load. When a phase misbehaves but its power stage tests fine, check its current-sense parts.
Replacing parts
- Power stages — the same part, or a documented equivalent with the same pinout and ratings.
- Inductors — the same value, footprint and current ratings; never a smaller saturation current.
- Capacitors — the same value, voltage rating and size; for polymer parts, similar resistance (ESR) as well.
Example: three cool phases and one warm one
A laptop shuts down under heavy load but runs fine at idle. Under a stress test on the bench, a thermal camera shows three phase inductors at similar temperatures and the fourth much cooler. Its power stage receives a PWM signal, but its switch node shows no switching on the oscilloscope: the stage's driver supply pin reads zero, because a small resistor feeding it has cracked. The remaining three phases were carrying the whole load and tripping the controller's current limit. A new resistor brings the fourth phase back, all four run at similar temperatures, and the shutdowns stop.
FAQ
How many phases does a laptop core regulator have?
It varies — often two to four for the processor core on thin laptops, more on high-power gaming designs.
Why do some phases not switch at idle?
The controller sheds phases at light load to save power.
Can one bad phase stop the laptop?
It can cause shutdowns under load, or stop start-up if the controller detects the fault.
Where are the core rail's ceramic capacitors?
Many are around and under the processor, often on the bottom side of the board.
Is a cooler phase a problem?
Under heavy load, yes — it may not be switching, leaving the others overloaded.
Bottom line
A core regulator is several buck phases sharing one output. Compare phases with each other — switch nodes, power stage pins, temperatures — and the faulty one stands out. Isolate shorts by lifting inductors and injecting low voltage, and replace power stages, inductors and capacitors with parts that match the originals' ratings.