[REWORKED] The Anatomy of a Boost Converter

Improved Boost Converter Circuit

⚠️ Important Safety Notice

This post is for educational demonstration only. While I have years of experience in electronics, this circuit involves high-voltage switching systems that can be dangerous if mismanaged. Please read the Full Safety Disclaimer at the bottom of this post before considering any practical work.

This post is a sequel to the previous post on boost converters. If you have not read the first post, I strongly recommend reading it first, as it explains the theory of operation covered in this post. You can read the first post here.

In the previous post, we constructed a simple boost converter on a breadboard. The results, however, were not particularly impressive. In this post, the circuit has been redesigned with improved components, and this time we will be driving an actual load rather than just a resistor. 

Updated Circuit Schematic

Since the theory of operation was already covered in the previous post, the focus here will be on the design enhancements made to the circuit. This version uses a genuine logic-level MOSFET with a dedicated gate driver, a proper power-conversion inductor, and a basic protection (regulation) circuit to power a 12V 65mA miniature filament bulb. Yes, the circuit looks much more complicated now, but rest assured, the underlying principle is the same. It's basically the same circuit but with better components, and a protection circuit at the output. That's all it is. 

IRLB8314 MOSFET

So, let’s go through the changes made to the circuit one by one. First, the counterfeit IRLZ44N MOSFET has been replaced with a genuine IRLB8314 MOSFET. This is also a logic-level device, but it offers exceptionally low drain-to-source on-state resistance (RDS(on)) and gate charge (Qg), making it well suited for this project. RDS(on) directly affects conduction losses in a MOSFET: When the device is fully turned on, it behaves like a small resistor with a resistance approximately equal to its RDS(on) value. A higher RDS(on) increases power dissipation during the MOSFET on-time (P=(I^2)*R), reducing overall efficiency, particularly at higher load currents. Lower RDS(on) therefore minimises voltage drop and conduction losses. Qg, on the other hand, mainly affects switching losses. It represents the amount of charge that must be supplied and removed by the gate driver to switch the MOSFET on and off. A MOSFET with a high Qg switches more slowly, increasing the duration during which both voltage and current are simultaneously present across the device. This leads to higher switching losses, especially at higher switching frequencies. On the contrary, a lower Qg allows faster switching transitions and therefore improves overall efficiency. 

Gate threshold voltage measured on the DCA55 component analyser

Using the DCA55 component analyser, we can see that the IRLB8314 turns on at an exceptionally low voltage of 1.76V. It's worth noting that while the IRLB8314 can theoretically operate directly from the 3.3 V PWM signal generated by the STM32, a dedicated MCP1407 MOSFET gate driver is used instead to provide a full 5V gate-drive swing for improved switching performance. 

MCP1407 High-Speed Power MOSFET Driver

Even though the IRLB8314 is a logic-level MOSFET, driving it directly from the STM32’s 3.3V PWM output is still not ideal. While the device achieves full enhancement at a gate-to-source voltage of 4.5V and features a very low RDS(on) of 3.2mΩ with a total gate charge of ≈40nC, the STM32 GPIO pin can typically source or sink only around 8-20 mA. As a result, the MOSFET gate capacitance charges and discharges slower, leading to longer rise and fall times during switching, which increase switching losses because the MOSFET spends more time operating in its linear region, where both voltage and current are simultaneously present across the device. This causes additional heat generation and reduces overall converter efficiency, especially at switching frequencies above several tens of kHz. With the MCP1407, which delivers up to 6A peak source and sink current, allowing the MOSFET gate to charge and discharge much more rapidly. This produces faster switching transitions, lower switching losses, improved thermal performance, and ensures the MOSFET operates closer to its specified RDS(on) value. Furthermore, the gate driver provides improved noise immunity and protects the STM32 GPIO pin from high transient currents. From these improvements, we can increase the 70kHz PWM signal in the last post to 128kHz, reducing the ripple current and for easier timer configuration on the STM32.

 470µH Ferrite Power Inductor (Left) and STPS1545 Schottky Diode (Right)

The old 4.7mH inductor used in the previous post has been replaced with a 470µH power inductor, designed for DC-DC converters. This inductor has a rated DC resistance of only 550mΩ (0.55Ω), a significant improvement over the old inductor, with a rated DC resistance of 8.33Ω. This reduces conduction losses in the inductor, which directly improves overall converter efficiency. In addition, the lower inductance value of 470µH is more appropriate for higher switching frequencies, allowing faster current ripple response and better transient performance. As for the Schottky diode, the SB560 used previously is no longer available (was used elsewhere), which is the sole reason for the switch to the STPS1545. It's still a Schottky diode, but it is rated for much higher current and power handling capability. Since both devices serve the same purpose with similar switching characteristics and forward voltage drop, this substitution should not noticeably affect the overall circuit performance.

Protection Circuitry

Moving on to the protection circuitry, I've constructed a basic voltage clamp and regulator with a 12V 1N5349B Zener diode and a transistor buffer with a TTC004B Bipolar Junction Transistor (BJT). If the input voltage rises above 12V for whatever reason, the Zener clamps down the excess voltage (dissipates it as heat), the bulb therefore remains clamped at the stable regulated voltage. The transistor acts as a buffer and helps with the load current by providing high current gain, allowing the Zener diode to operate at very low power while the BJT handles the heavier current demanded by the bulb. As the BJT is configured as an emitter follower, its output (emitter) closely follows the stable base voltage set by the Zener (but with a 0.7V drop), maintaining a clean and protected ~11.3V across the bulb even when input voltage surges, transients, or fluctuations. I'm not gonna go too deep into the details of this circuit. But if you're interested and want to learn more, feel free drop a comment and I'm more than happy to do a post about it. 

Unregulated output at 58% Duty Cycle

Now, let's talk results. As discussed in the previous post, the output voltage of a boost converter can be approximated by the input voltage multiplied by 1/(1 − Duty Cycle). Based on this relationship, achieving a 12 V output from a 5 V input requires a duty cycle of approximately 58.33%, which was rounded to 58% for implementation. The STM32 was therefore programmed to generate a 128 kHz PWM signal with a 58% duty cycle. The multimeter measured an output voltage of 11.08V (hence a input voltage sag from 5V to 4.65V). As in the previous test, the input supply is loaded and does not maintain a perfect 5 V under operating conditions. But regardless, the measured output voltage is reasonably close to the theoretical 12 V target, demonstrating that the converter is operating as expected.

Regulated output at 58% Duty Cycle

As mentioned before, the BJT will introduce a voltage drop of around 0.7V. We can observe this by measuring the voltage across the bulb, which is 10.36V. Therefore, the BJT has dropped 0.72V in this case. Although the bulb lights up, it's pretty dim at just 10.36V. Let's see how we can compensate for that so that the bulb can work closer to its rated 12V.

Unregulated output at 63% Duty Cycle

The easiest way is to simply increase the duty cycle. I reprogrammed the STM32 to increase the PWM's duty cycle from 58% to 63%, yielding a theoretical input of 13.5V. Accounting for the voltage sag of the input voltage, the output measured a 12.12V at the unregulated output, which does hit the 12V mark we're aiming for. 

Regulated output at 63% Duty Cycle

Past the 0.7V drop from the BJT, the bulb receives 11.4V, as measured on the multimeter. If you zoom in to the image above and compare it to the image with 58%, you'd notice the bulb is noticeably brighter, as it works closer to its intended 12V working voltage. 

Now, you've rightly noticed that the boost converter is far from perfect. Although with all these improvements and optimisations on the hardware side, practicality often shows how idealistic theories can be. To constantly compensate for the voltage sag and load variations, a feedback network is needed, where the microcontroller monitors the input and output voltage and dynamically adjust the duty cycle to the optimum percentage to match the desired voltage. I'm not going to get into that in detail but if you're interested, please do look it up. 

And here we are, we have successfully improved the boost converter circuit with a load protection mechanism, and successfully powered a 12V miniature filament bulb, with simply a 5V input. There are, of course, many other ways and techniques that can boost voltages. Boost converters still remain one of the most efficient and robust DC-DC conversion topologies. As always, feel free to share your thoughts in the comments, and I'll see you in the next post. 

⚠️ FULL SAFETY DISCLAIMER

The project described in this post is for educational demonstration only. I have years of experience in electronics and have followed strict safety protocols in building and testing it.  If you do not have a similar level of experience in electronics or electrical engineering, do not attempt to recreate this circuit on your own. High-voltage switching circuits can generate potentially lethal voltages capable of causing severe electric shock, burns, fire, or death. Breadboarded prototypes lack the thermal shutdowns, proper creepage/clearance distances, fusing, and safety certifications found in commercial products.  

Therefore:

• Beginners, early-stage students, or hobbyists without comparable experience should never build high-voltage switching circuits without the direct, in-person supervision of a qualified professional (university lab supervisor, licensed electrician, or experienced engineer).  
• For everyday testing and project use, it is cheaper, smaller, and significantly safer to purchase a pre-made IC module (like the MT3608). These are designed to be "plug-and-play" with safety standards already in place.

All information is provided “as is” without any warranties of any kind. This is not professional, licensed, or safety advice. By reading or attempting to replicate any part of this project, you do so entirely at your own risk. I am not responsible for any damage to equipment, personal injury (including death), fire, or any other loss resulting from the use or misuse of this information. Safety is your responsibility. Always comply with your local electrical codes, university lab rules, and regulations. To the fullest extent permitted by law, I disclaim all liability.

Click here to read the blog's full terms and disclaimers.