Mainboard power delivery configurations
Mainboard power delivery configurations
Hello everyone, I'm curious about different ways PWM signals are handled in MOSFETs. Some motherboards use highside and lowside configurations like my B450 Mortar Max. How does that work? For example, would it be: pwm control → highside MOSFET → two lowside ones? This seems to involve duplication or something similar. I'm trying to understand this better. ## Edit ## Understanding MOSFET and VRM setup
They divide the PWM signal into two phases and monitor current flow. This approach balances performance and efficiency. Some designs use current monitoring and adjust the PWM to one phase for better balance. (like ISL6617) High-side and low-side MOSFETs regulate voltage, with the low side managing most of the work. For instance, using SM4337 and SM3503 without doublers is common. Edit: For more on VRM, check Zoid's PCB analyses. Updated August 24, 2020 by TofuHaroto
Doublers utilize one PWM signal from the VRM controller—typically one per phase—and amplify it by doubling the output. This creates two opposing signals, usually switching in opposite directions (for example, when one indicates a turn-on, the other indicates a turn-off). When a VRM signal like 1011 is active (enabling phases 1, 3, and 4 while deactivating phase 2) for a brief interval (around 1/300,000 to 1/600,000 seconds), the doubler performs specific actions: during half the cycle it outputs 10 00 10 10 and in the other half 01 00 01 01. Overall, across the entire period, the pattern alternates, resulting in a sequence like 11 00 11 11 but not simultaneously. This method allows using less expensive or mass-produced VRMs that support only four phases, while doubling them with doublers to achieve eight phases. The current is spread across more phases and components, reducing heat per group and smoothing the output voltage. Each phase requires a specific type of MOSFET: a high-side one for lower current at 12V, and a low-side one for higher current between 0.5v and 1.5v. To cut costs or manage heat better, manufacturers often pair two low-side MOSFETs (and sometimes two high-side ones), sharing the load so each handles half the current. This distributes heat more evenly and allows easier dissipation via heatsinks. Premium VRMs incorporate power stages—integrated chips housing all drivers, MOSFETs, and advanced features like current sensing, overcurrent protection, thermal management, and efficiency optimizations during low-power operation.
Visualize the hi-side and lo-side MOSFETs as two valves or faucets, controlling flow. The driver activates the hi-side for a short burst, allowing 12V to enter and boost the phase before shutting it off rapidly. Think of a car engine spraying fuel into the cylinder—similar to how energy moves through the system. Next, the driver opens the lo-side MOSFET, letting some energy pass through inductors and capacitors. Once the voltage reaches the target level (like 1.3V for the processor), the driver closes the lo-side switch and restarts the cycle. The inductor stores energy while the hi-side is off, ensuring a steady supply. These buildzoid videos provide a clear explanation of the process.
Mosfets combine high and low sides with the driver IC, generating less heat and simplifying cooling compared to discrete units (though this varies by model). There are DRmos and SPS options; SPS models include current and temperature monitoring plus safety features, while DR mosfets lack these. For instance, the SIC639 or 634 series are DRmos examples, and ISL99370/60/90 or 21472/21462 are SPS variants. Keep checking Buildzoid for clear explanations.
Various configurations exist, such as high-fidelity and low-fidelity setups, or specialized chips that combine multiple MOSFETs on a single component, often paired with driver circuits. Some VRM controller chips include integrated drivers, while others rely solely on PWM signals sent to separate MOSFET drivers, which then manage the actual switching devices. Certain VRM controllers feature just one or two built-in drivers, whereas older motherboards used six-phase controllers arranged in four phases for CPU cores, which could connect to double channels, MOSFET drivers, or power stages containing their own drivers. The remaining two phases might serve the SoC or integrated graphics, allowing direct connections of high- and low-side MOSFETs to the built-in drivers.