# Buck Converter Design

PhysicsTest
TL;DR Summary
I need to design a Buck converter circuit with the following requirements 24-100V (2 - 10 Amps) to 12V (20 Amps) Output.
I need to design a Buck converter circuit with the following requirements 24-100V (2 - 10 Amps) to 12V (20 Amps) Output. From the net i could get the below circuit.

First major question is, do i need to interface a micro controller to achieve the functionality to drive the MOSFET? I am beginner in these kind of circuits. Few of the things i know is the PWM duty will be calculated as Voutput / Vinput = 12/24 = 50% duty and apply this duty to MOSFET driver circuit. But if the input is different from 12V, i am thinking that i shall read the input voltage using the micro controller and control the duty accordingly. Is my understanding correct?

I am beginner in these kind of circuits.
Then the best solution would be to check on manufacturer toolkits. TI, AD and practically any DCDC converter manufacturer has some kind of design toolkit: you give the parameters and get a design with calculated properties and a BOM.

i am thinking that i shall read the input voltage using the micro controller and control the duty accordingly. Is my understanding correct?
Not really. A uC usually just makes it difficult (you need a software).

Borek, DaveE, Buckethead and 1 other person
First major question is, do i need to interface a micro controller to achieve the functionality to drive the MOSFET?
You might only need a voltage comparator with hysteresis.
If the output voltage is low, then turn on MOSFET M1.
If the output voltage is high, then turn off MOSFET M1.
Simulate the circuit with SPICE to check output ripple, L1 and C1.

PhysicsTest
PhysicsTest
You might only need a voltage comparator with hysteresis.
If the output voltage is low, then turn on MOSFET M1.
If the output voltage is high, then turn off MOSFET M1.
Simulate the circuit with SPICE to check output ripple, L1 and C1.

Is the above block diagram correct for the voltage comparator?

Is the above block diagram correct for the voltage comparator?
Yes.
You might simplify the MOSFET driver by inverting the circuit and using the N-chan MOSFET as a low side switch.
You do not need to compare 12 V with 12 V. You could divide the output and compare that with a lower reference voltage.

Gold Member
Summary:: I need to design a Buck converter circuit with the following requirements 24-100V (2 - 10 Amps) to 12V (20 Amps) Output.

I need to design a Buck converter circuit with the following requirements 24-100V (2 - 10 Amps) to 12V (20 Amps) Output. From the net i could get the below circuit.
View attachment 298956

First major question is, do i need to interface a micro controller to achieve the functionality to drive the MOSFET? I am beginner in these kind of circuits. Few of the things i know is the PWM duty will be calculated as Voutput / Vinput = 12/24 = 50% duty and apply this duty to MOSFET driver circuit. But if the input is different from 12V, i am thinking that i shall read the input voltage using the micro controller and control the duty accordingly. Is my understanding correct?
In order for this to be a Switching regulator Rives suggestion in post #2 is your very best bet. There are specialty ICs that have all the essentials you (some including the FET) that give you excellent performance. With the Voltages and currents you are considering I would STRONGLY suggest going that route.

There are specialty ICs that have all the essentials you (some including the FET) that give you excellent performance.
Can you give an example part number ?

Gold Member
View attachment 298962

Is the above block diagram correct for the voltage comparator?
NO! This will not work. This is only a linear regulator and for the current output you need with the input voltage you want, you will likely burn up the FET. In order for this circuit to work your comparator needs to drive an oscillator (perhaps 30khz-100khz output) that has a pulse width variable output. The higher the load (the lower the output voltage) then the greater the duty cycle of the oscillator to keep the FET turned on longer.

If you are doing all this to learn about how switching regulators work, then cool, no better way to learn although you will be doing a lot of degugging. But if you need to get something running for your work or a product, then you definitely should do a search for Switching Regulator integrated ciruits from manufacturers web sites (Analog Devices, TI and so on), use their calculating tools to generate a BOM, and design the circuit around that chip. It will save you weeks of work.

NO! This will not work. This is only a linear regulator and for the current output you need with the input voltage you want, you will likely burn up the FET.
Rubbish.
It is not a linear regulator, it is a switching converter.

artis
Gold Member
Summary:: I need to design a Buck converter circuit with the following requirements 24-100V (2 - 10 Amps) to 12V (20 Amps) Output.

But if the input is different from 12V, i am thinking that i shall read the input voltage using the micro controller and control the duty accordingly. Is my understanding correct?
What exactly were you planning on doing once you've read the input and determined it has changed?
-
Basically, what has been said by other posters here is correct. You sample the output, compare it against a reference and adjust the duty cycle accordingly. If the input voltage changes it will be reflected on the output and the feedback loop will take care of it.
-
Take the advice from @Rive in post #2. Study the documentation for several switching regulator ICs. Understand them inside and out.

Gold Member
NO! This will not work. This is only a linear regulator and for the current output you need with the input voltage you want, you will likely burn up the FET.
Since the "Gate Driver" is not specifically defined in the block diagram it is rather foolish of you to post what you did. The subject of the thread is buck convertor. It should be implied that the gate driver provides an on-off signal, not functioning as part of a linear regulator.

Gold Member
Since the "Gate Driver" is not specifically defined in the block diagram it is rather foolish of you to post what you did. The subject of the thread is buck convertor. It should be implied that the gate driver provides an on-off signal, not functioning as part of a linear regulator.
Perhaps, but I'm likely to think because of the detail of the block diagram (comparator block, etc.) that a block labeled simply "Gate Driver" is a simple circuit biasing the FET. A Voltage to Pulse Width Modulator would be a more appropriate name. Because of the simplicity of the original diagram I found it prudent to point out that the diagram with the Gate Driver was leading toward a linear regulator circuit and not a buck convertor.

Gold Member
Perhaps, but I'm likely to think because of the detail of the block diagram (comparator block, etc.) that a block labeled simply "Gate Driver" is a simple circuit biasing the FET. A Voltage to Pulse Width Modulator would be a more appropriate name. Because of the simplicity of the original diagram I found it prudent to point out that the diagram with the Gate Driver was leading toward a linear regulator circuit and not a buck convertor.
Doesn't the freewheeling diode and series inductor give it away?

Gold Member
Rubbish.
It is not a linear regulator, it is a switching converter.
It is not anything but a linear regulator without a pulse width modulator!

Gold Member
Doesn't the freewheeling diode and series inductor give it away?
Yes, of course it does. I didn't say the design wasn't moving in the right direction, I just pointed out that if you drive a FET gate with a comparator without a PWM circuit in there somewhere, even that circuit will work, but only as a linear regulator.

Mentor
I need to design a Buck converter circuit with the following requirements 24-100V (2 - 10 Amps) to 12V (20 Amps) Output. From the net i could get the below circuit.
View attachment 298956
Taking a couple steps back...

Why are you designing a moderately high power voltage regulator as your first project?

100V input is above SELV, so you really should be using an isolated switching power supply circuit for this, not a non-isolated buck DC-DC converter.

What is the application? What experience do you have with power components (the FET, the flywheel diode, etc.)? What experience do you have with safety certifications for circuits that involve voltages above SELV? Even if you are not going to submit the design to UL for testing and approvals, you still should be designing to their standards.

It is not anything but a linear regulator without a pulse width modulator!
You are forgetting the voltage comparator with hysteresis is in the feedback loop. That makes it a self oscillating switching regulator. It PWMs itself.

Only if the voltage comparator was replaced with a linear op-amp error-amplifier would it become a linear regulator.

Gold Member
You are forgetting the voltage comparator with hysteresis is in the feedback loop. That makes it a self oscillating switching regulator.

Only if the voltage comparator was replaced with a linear op-amp error-amplifier would it become a linear regulator.
OK, you are absolutely right. I stand corrected. I was looking at the comparator as operating in the linear region. Thank you for the correction. (tail between my legs).

Klystron, berkeman and Baluncore
Gold Member
One thing I learn as a working EE for many years is that it is better to avoid problems than fix them. If you aren't interested in the details of SMPS design then do this:

Go to TI.com and search for "simple switcher". Also look into their design tools, they will tell you how to do this right down to BOMs and PCB layouts. Analog.com also has similar stuff.

Just buy it, there are lots (maybe too many) of great integrated solutions out there. I can tell you as a one time expert in this field, this is what I often did for my simpler requirements.

PS: Oops! I just realized you need 240W with a wide range input. That's way less common, certainly not a "simple switcher". Still go to those sites they also make all of the integrated control IC's you'll need. Also look at power supply manufacturers. Buy it if you can. Also, step 1, what are you going to do with the 24W watts (approx.) of heat this thing will make?

Last edited:
Gold Member
comparator as operating in the linear region
That's an odd turn of phrase. You don't see that too much. I call those amplifiers, myself.

Last edited:
Gold Member
BTW, you can read my favorite SMPS textbook for free. Even though I'm biased by learning this material as he did from our prof R.D. Middlebrook, I really do think it's the best for what it covers. Don't be scared by some of the complexity, you don't need all of it.

Pay attention to continuous vs. discontinuous conduction modes, that is a common issue with your wide input range. The alternative is CCM with a huge inductor.

Last edited:
berkeman and Joshy
PhysicsTest
Thank you all for the suggestions, I understand it is a tough job to design the circuit on my own, i will try to learn in parallel, but i need to close this activity, i have not gone in details of why the requirements are given in such a way, as i am not much aware of the standards.
I have checked the Analog Devices website
https://www.analog.com/en/parametri...nmin|24_vinmax|100_vout|12_iout|20&p5362=Buck

I could get the following devices
LTC7871
LT8228
LTC3871
LTC3703

Gold Member
Rule #1 of component selection: Derate. I would not choose a part with a 100V absolute maximum spec for a 100V application. It is likely that you will have to create a lower voltage supply for the control IC.

Also, the controller isn't the first part of the design to worry about. Assume that you can design or buy the necessary controls later. First you need to focus on the selection of the inductor(s), the MOSFET(s) & diodes and their heat sinks, and the capacitors at the input and output. Figure out the high power part first. Then select an appropriate control scheme. When you have completed this design and it's working well, you will realize that the inductor(s), the heatsink/transistors and the capacitors (esp. cap ripple current) were the most important, largest, and most expensive parts.

Also consider that you may not have the necessary background to do a relatively difficult buck converter design quickly (yet). What about asking for help or more time? Is that better than throwing together something that isn't reliable?

I don't know your system; specifically why you need to operate from 24 -100V. But, in my former life of doing this stuff, I would put some effort into reconsidering those numbers or talking to the systems designers that chose them. You, or they, may not understand the ultimate cost of those decisions.

Finally, just to get some perspective on what this thing will look like when you're done. Look online at some other 240W 12V power supplies, don't worry about the exact specs, this is just to get an idea of how big they are, what are the significant component sizes and mechanical design elements. Hint, it's not the ICs.

Honestly, I don't think that apart from the LTC3703 you would be able to manage the others.

PhysicsTest
PhysicsTest
So, i will select the components as per the application note provided in the data sheet.

Understand and select the pin configuration as per the requirement, and design the PCB. Is my design job over and then perform the testing?

The components from the AppNote 'typical application' won't do: that design does not match with your initial requirements.

Maybe you should try Webench instead.

Mentor
Thank you all for the suggestions, I understand it is a tough job to design the circuit on my own, i will try to learn in parallel, but i need to close this activity, i have not gone in details of why the requirements are given in such a way, as i am not much aware of the standards.
Did you understand my comments about SELV? (If not, you should not be doing this project yet)

Do you understand that if somebody gets hurt because of your design (how you ignored the safety regulations surrounding SELV), that you could be liable for those damages? Why are you not saying what this strange project is for? I don't think we will be willing to help you much more, given the safety issues involved.

PhysicsTest
Ok I will check and update what this project is for, but some safety features like short circuit planned to implement. But I study in detail as well.

artis
Ok I will check and update what this project is for, but some safety features like short circuit planned to implement. But I study in detail as well.
I suggest search google for "buck converter" , find a schematic that is based on an IC as there are tons of them for these purposes. Then come here post it and ask questions. Taking a working schematic that is simple is easier for you as well as for us commenting.

What will be the source of your input voltage/current?
I ask this because a buck converter is a NON-isolated device/topology which means it doesn't incorporate a transformer with electrically separate windings as does most higher end SMPS designs. That is one of the reasons why it's called "buck" converter because you omit various parts and make it cheap...

The way it achieves voltage step down without a transformer is simply using an inductor which is just a coil of wire often with a magnetic core, whether ferrite or steel lamination stack depending on frequency.
An inductor is also called a choke. The converter has a switch and this switch switches the current through the inductor periodically.
Inductors resist current change just as capacitors resist voltage change. Therefore constantly switching an inductor on/off means only a limited current can go through , this current is limited by the inductance of the inductor as well as the waveform and frequency of the switched current.

So your output gets limited current and limited current results in a voltage drop. This is how the inductor with the help of a periodically switching current can result in a lowered voltage.

The reason I ask about your source is because if the switch fails and transistors tend to fail short circuited it means your output voltage will rise to the input voltage and depending on what is attached it will either fry , blow apart or hurt someone if someone is touching it.

PS. do you know the function of the diode in the circuit? It plays a simple but crucial role and is implemented in many switching designs...
https://components101.com/articles/buck-converter-basics-working-design-and-operation

And do come back , experimenting is fun but you need to learn the basics to make it even more fun otherwise you will be like me trying to figure out complex stuff with lacking knowledge and that results in more mistakes and more burnt out parts.
I'd say the difference between knowing theory and not knowing it is the weight of dead parts after a finished project or sometimes the weight of a dead experimenter... be careful high voltage and current depending on circumstance can really give your heart a "beating"

PhysicsTest
Gold Member
but some safety features like short circuit planned to implement.
How about implementing ALL of the required safety features? Especially, exposure to hazardous voltages.

berkeman
Ok I will check and update what this project is for, but some safety features like short circuit planned to implement. But I study in detail as well.
You need an input voltage below 50 volts to be low voltage.
You will not receive help so long as the input voltage is above 50 V.
You must respecify the project.

berkeman
Gold Member
You need an input voltage below 50 volts to be low voltage.
You will not receive help so long as the input voltage is above 50 V.
You must respecify the project.
Yes.

The limits I'm used to (from EN60950-1 et.al.) are 60Vdc or 42.4Vac (which is 60V peak).

Granted this seems like pointless nit picking. It is, but that's not my point. My point is that designers need to know the relevant safety standards for their application. If you don't, you need to stop and find out. Medical? Patient contact? Industrial machines? Autos? Household appliances? IT equipment? Lab equipment? Military equipment? Mining?... They all have safety standards.

We don't know yours, nor is it our responsibility. I used to bill at \$100/hour for that sort of advice (it'd be more now, a decade later).

My big gripe is that this is never mentioned in schools. I don't expect that they actually teach the standards, that's a waste of effort, they are all a little different and they'll change during your career anyway. Maybe teach some general concepts of design for safety. But, for god's sake, engineers should know they exist from day one and that they'll have to figure out and read the ones that matter to them.

PS: My other big gripe are the MBA types that will hire anybody to do the work as quickly as possible. "OK, maybe you're really a chemist, or a microcontroller guy, but you can also do our power supplies, because I don't want to hire specialists". This is why you'll be better off buying a Toyota than a Yugo. This is why we have lawyers. Don't feed the lawyers.

\EndOfRant

Last edited:
PhysicsTest
I saw the data sheet of the LTC3703 and there was a design example on page 27, i followed the same procedure.
LTC3703 data sheet
Specifications:
VIN = 24V to 100V, VOUT = 12V, IOUT(Max) = 10A, f = 250kHz

1. RSET
As per the formula in the data sheet
RSET(kOhm) = 7100/(250 - 25) = 31.5KOhm -> eq1

2. Inductor value for about 40% ripple current at maximum VIN
L = 12/250K*0.4*10(1 - 12/100) = 10.56uH. -> eq2

3. Minimum On-time
ton(min) = 12/(100*250kHz) = 480nS -> eq3
which is above LTC3703's 200ns Minimum on-time.

4. Top and Bottom MOSFET switch

Since the drain of each MOSFET will see the full supply voltage 100V (max) plus any ringing, choose a 125V MOSFET to provide a margin of safety.

PMAIN = 1.32W -> eq4

= 96.4DegC.

=1.84 -> eq5
TJ = 70 + 1.84*20 = 106.8

5. Current Limit Resistor

RDS(ON) = (25mOhm)/2 [1 + 0.009(105 - 25)] = 21.5mOhm
IMAX pin voltage = 10*0.0215 = 0.215V
RMAX = 0.215/12uA = 18k -> eq6

6. CIN is chosen for an RMS current rating of 5A (IMAX/2) at 85DegC.
7. For the output capacitor 2 low ESR OS-CON capacitors (18mF) are used.

This is how i am approaching, i need to select the MOSFET as per the specifications and started drawing the design in LTSPICE.