Posted in

How to design a MOSFET – based power supply?

by della•

Hey there, it’s Jake from the MOSFET supply team – been getting a ton of DMs lately from techs, design engineers, and even small startup founders asking the same thing: “How do I actually build a reliable, efficient power supply using MOSFETs?” Most of you aren’t just chugging along with off-the-shelf parts; you’re tweaking designs for IoT gadgets, portable tools, or solar setups, and your old power supplies are either wasting too much juice or frying when things get cold. I get it – I’ve been in the lab at 2 a.m. troubleshooting a power supply that kept dropping voltage mid-test, staring at a pile of MOSFETs like they’re magic black boxes. Spoiler: they are, but only if you design them right. Today, I’m breaking this down like I would for a new hire on our team – no jargon overload, no fancy equations that make your eyes cross, just real stuff we see every day at the supplier desk. Let’s dive in. MOSFETs

First, let’s get one thing straight: MOSFETs aren’t just switches for fun – they’re the backbone of every switching power supply (aka the ones that don’t get hot like linear supplies, aka the ones most of you actually need). Linear supplies are great for tiny, low-load apps, but if you’re running anything over a few hundred milliamps, switching supplies with MOSFETs will save you so much power. But here’s the catch: pick the wrong MOSFET, and your whole design will be garbage. I’ve seen a guy use a 100V N-channel MOSFET for a 12V supply just because it was cheap, and when the unit got kicked around, it shorted out. Oops. So step one: know exactly what you need from your MOSFET before you even start shopping.

Let’s start with the basics of power supply topologies – because the MOSFET’s job changes depending on what type of supply you’re building. The two big ones for low-to-medium power are buck converters and boost converters. Buck is when you need to take a higher input voltage and drop it to a lower, steady output (like 24V in to 5V for a Raspberry Pi). Boost is the opposite: lower input to higher output (like a 3.7V phone battery stepping up to 12V for a car phone charger). For buck converters, you use an N-channel MOSFET – it’s more efficient, has lower on-resistance (Rds(on)) than P-channel, so less heat. For boost, sometimes you’ll use a P-channel, but a lot of us prefer N-channel with a charge pump driver these days, since N-channel is easier to source and cheaper. Pro tip from the lab: if your design is going to be in a small enclosure with no airflow, stick to topologies that minimize switching losses – that’s where MOSFET selection hits hard.

Now, the part that trips everyone up: picking the right MOSFET specs. I see engineers online argue about this all the time, but at the end of the day, it’s about matching the part to your load, not chasing the highest voltage rating just because. Let’s break down the non-negotiables:

  1. Voltage rating (Vds): This has to be at least 20% higher than your maximum input voltage, and if there’s any transient spike (like from an inductive load or input ripple), add another 10%. Why? I’ve had a customer’s design fail because their MOSFET was rated for 25V, and their input spiked to 28V when they turned on a motor. RIP the prototype. So if you’re doing a 12V buck, don’t grab a 20V MOSFET – get at least 25V, but 30V is safer for most apps. For a boost converter going from 5V to 15V, your MOSFET’s Vds needs to be at least 20V plus the boost voltage, so like 35V minimum. No skimping here.

  2. On-resistance (Rds(on)): This is the big one for efficiency. Rds(on) is how much resistance the MOSFET has when it’s fully on (like, when it’s “closed” as a switch). Lower Rds(on) means less power lost as heat, which means your supply runs cooler and uses less battery. But here’s the tradeoff: MOSFETs with lower Rds(on) usually cost more and are a bit bigger. So if you’re designing a tiny portable gadget, you might accept a slightly higher Rds(on) to save space, just make sure it’s not so high that your MOSFET overheats at full load. For example, a 10A load: a MOSFET with 0.05Ω Rds(on) would lose 5W as heat (P=I²R = 10²*0.05), while one with 0.1Ω would lose 10W. That’s the difference between a warm case and a case you can’t touch. I always tell our customers: calculate your max power loss from Rds(on), then pick a MOSFET that can handle that without exceeding its maximum junction temperature.

  3. Current rating (Id): This is the maximum continuous current the MOSFET can handle. Don’t just match your load current – add 50% headroom. Why? MOSFET current ratings drop at high temperatures, and if your app has inrush current (like when a motor starts up), that current is way higher than normal. For example, if your load is 5A, don’t get a 5A MOSFET – get a 7.5A or 10A part. I’ve seen this go wrong too: a drone battery supply that used a 6A MOSFET kept failing when the drone took off, because the inrush current hit 8A and the MOSFET overheated and died. Headroom is your friend.

Next up: driving the MOSFET. This is the part most first-timers skip, and it’s why their MOSFET switches slow and get hot. MOSFETs need a gate voltage to turn on – for N-channel MOSFETs, that’s usually 10V (most are rated for 5V or 10V gate drive, but 10V gives better performance). If your microcontroller or driver chip only outputs 3.3V, you can’t just hook it up directly to the MOSFET gate – it won’t fully turn on, so Rds(on) will be way higher, leading to heat loss. That’s where a gate driver comes in. For our customers, we usually recommend a simple low-side gate driver for most buck converters – it’s cheap, small, and does exactly what you need: amplifies the microcontroller’s 3.3V signal to the 10V the MOSFET needs. Pro tip: add a gate resistor (10Ω to 100Ω) between the driver and the MOSFET gate. Too small, and you get switching noise that messes up your circuit; too big, and switching is slow and causes losses. I’ve found 22Ω works for 90% of designs.

Now, the actual design steps, because specs mean nothing without putting it all together. Let’s walk through a real example: designing a 12V input to 5V, 2A buck converter for a portable sensor.

Step 1: Pick the topology – obvious, it’s a buck converter since we’re stepping down voltage.

Step 2: Calculate MOSFET specs. Input is 12V, so Vds needs to be at least 12V * 1.3 = 15.6V, so let’s grab a 20V MOSFET (room for transients). Load is 2A, so Id at least 2A * 1.5 = 3A. Now, Rds(on): we want low heat, so let’s look for a MOSFET with Rds(on) ≤ 0.1Ω at 10V gate drive. That gives us power loss of (2A)² * 0.1Ω = 0.4W – super low, no heat issues.

Step 3: Pick the rest of the circuit to match the MOSFET. Buck converters need an inductor and a capacitor too – but they work with your MOSFET. The inductor needs to handle the same current as your MOSFET plus a bit, and the capacitor smooths the output voltage. A rule of thumb: for 2A output, a 10µF capacitor works for basic apps, but 22µF is better for noise. We often recommend our customers pair our MOSFETs with off-the-shelf inductors from reputable suppliers, but if you need a part number, just ask our team – we have a list of compatible parts.

Step 4: Layout matters. Don’t route the power lines next to signal lines – switching MOSFETs make noise, and that noise can mess up your microcontroller’s readings. Keep the gate drive trace short (under 1cm if possible) to reduce noise, and use a ground plane to dissipate heat. I can’t stress layout enough – I had a customer who designed a perfect schematic, but their board layout had the power trace running over a signal trace, and the unit kept dropping voltage. We re-spun the layout, and it worked first try. 50% of power supply issues are bad layout, not bad parts.

Now, common mistakes I see all the time. Let’s get these out of the way so you don’t make them:

  • Using the wrong MOSFET type: N vs P channel. P-channel is easier to drive for low-side applications, but has higher Rds(on) so more loss. N-channel is the way to go for most medium-power apps, but needs a gate driver that can go above input voltage. Don’t mix them up – I’ve seen someone use a P-channel N-channel by accident in a boost converter, and it wouldn’t turn on at all.

  • Ignoring switching losses. When the MOSFET is switching on and off, there’s a short time where both voltage and current are high, leading to power loss. That’s why higher switching frequencies (like 100kHz vs 50kHz) mean smaller inductors and capacitors, but more switching loss in the MOSFET. If your app is battery-powered and needs to be efficient at low load, lower switching frequency is better. If you need a small board, higher frequency works – just pick a MOSFET with fast switching speed.

  • Not testing for transients. As I mentioned earlier, input spikes or output shorts can kill a MOSFET. Add a zener diode across the MOSFET’s drain and source to clamp any voltage spikes. It’s a cheap part, saves your whole circuit. I’ve seen a prototype get hit by a short, and the zener absorbed the spike – the MOSFET was fine, the unit worked. No zener, and we’d have had to redo the whole board.

Wait, what about if you’re doing high-power supplies, like for a robotic arm or an EV charger? The rules change a bit, right? For high power (over 100W), you might use parallel MOSFETs to split the current. If you have a 20A load, two 10A MOSFETs in parallel will each carry 10A, so their combined Rds(on) is half of one, way less loss. But you have to make sure they’re matched – same Rds(on), same gate capacitance – otherwise one MOSFET will carry most of the current and overheat. Our team helps a lot of high-power customers with parallel setups, we can test MOSFET matching for you if you need it.

Another thing: thermal management. MOSFETs get hot, especially when Rds(on) is high or current is high. If your MOSFET is going to run over 60°C, add a heat sink. Even a small, cheap heat sink will drop the junction temperature by 20°C or more, extending the MOSFET’s life. For high-power apps, you might use a heatsink with a fan, or even liquid cooling, but that’s overkill for most hobbyist or small business designs.

Now, let’s talk about when to switch things up. If you’re building a linear power supply, MOSFETs still play a role – they’re used as pass elements instead of the old BJTs. Linear MOSFET supplies are more efficient than BJT ones, and have lower dropout voltage. But again, same rules apply: pick the right Vds, Rds(on), current rating. We have a lot of customers who use our low-voltage MOSFETs for linear supplies on battery-powered medical devices, and they love how reliable they are.

I know this sounds like a lot, but here’s the thing: you don’t have to figure it out alone. Our team gets tons of requests for design help – we’ve done everything from tiny IoT buck converters to high-power solar charge controllers. If you’re stuck on specs, or layout, or just want a second opinion on your schematic, hit us up. We’re MOSFET suppliers, but we don’t just send boxes and disappear – we work with our customers to make their designs work. A lot of our regulars tell us that once they start using our MOSFETs, their power supply failure rates drop from 15% to less than 2%, and that’s the best part for us – knowing we’re helping you make better products.

Wait, one last tip: don’t buy cheap knockoff MOSFETs. I know they’re tempting on Amazon or Alibaba, but 9 times out of 10, they’re counterfeit – the specs are wrong, Rds(on) is higher than advertised, or the Vds rating is faked. We’ve seen customers who bought a “100A MOSFET” that actually only handles 15A, and it burned up their whole circuit. Stick to reputable suppliers, or at least check the datasheet from the original manufacturer. Every MOSFET we sell comes with the full, official datasheet, no fakes – that’s our promise.

So to wrap this up: designing a MOSFET-based power supply is all about matching the right MOSFET specs to your load, picking the right topology, driving the MOSFET correctly, and getting the layout right. Skip the cheap parts, add headroom, test for transients, and if you get stuck, our team is here to help. At the end of the day, a good power supply doesn’t have to be complicated – it just has to be reliable, efficient, and fit your design.

If you’re ready to source MOSFETs for your next power supply project, or need help refining your design, reach out to our team to connect on procurement and design support. We’re here to make sure your next power supply works the first time, no troubleshooting at 2 a.m. required.

DIODE References

  1. “MOSFET Selection Guidelines for Switching Power Supplies”, Infineon Technologies Application Note
  2. “Designing Efficient Buck Converters with N-Channel MOSFETs”, Texas Instruments Technical Brief
  3. “Power MOSFET Basics: Understanding Rds(on) and Vds Rating”, ON Semiconductor Application Note
  4. “PCB Layout Best Practices for Switching Power Supplies”, Analog Devices Design Guide

Tongke Electronic Co., Ltd
Tongke Electronic Co., Ltd. is one of the most experienced mosfets manufacturers and suppliers in China, featured by quality products and low price. Please rest assured to wholesale advanced mosfets made in China here from our factory. Contact us for pricelist.
Address: No.3,Chayuan Rd, Street 3, AilingKan, Dalingshan, Dongguan, Guangdong, China.
E-mail: jack@ctk-elec.com
WebSite: https://www.ctkchip.com/