Understanding Buck Converters

Thread Starter

Sphades

Joined Mar 7, 2020
14
Hello everyone!

I need a little help designing a buck converter. Basically I want to generate a 5V output that will stay the same regardless of load resistance and that will exist when the input varies between 12 and 48 volts.
I understand the basic principle of the buck converter circuit but I do not understand how to make the PWM generator work. This circuit is supposed to be used for a charging circuit of a battery. Those being said, how am I supposed to power up the op amps needed for the PWM generator when this circuit is used for the power supply?
I attached an image with the circuit ( values not calculated yet ).

Any help is welcome and thanks for every bit of help!
1627319597921.png

Update:

I found this IC online that is used with buck converters:
1627322292626.png
I want to do something like this but I don't really understand where do they take the voltage to power up the op amps, as it does not show.
 
Last edited:

Papabravo

Joined Feb 24, 2006
16,772
That would be from the Vin pin and the internal regulator.

There is a great deal of complexity in that block diagram. There are books and application notes that will reveal the details of what is inside the blocks. Are you sure you want to go there?
The best book I've seen so far is by Christophe Basso, Switch-Mode Power Supplies, (2014). It is not easy reading, but it is complete and deep.
The best application note is from Texas Instruments, by Brigitte Hauke, Basic Calculation of a Buck Converter's Power Stage, SLVA477B, December 2011, (Revised, August 2015)
 
Last edited:

Thread Starter

Sphades

Joined Mar 7, 2020
14
@Papabravo it is a requirement to design one, rather than to use an already existent module. I did some calculus regarding the Buck components, but I don't really understand how to design the PWM circuit without using an external power source as it would defy the purpose. @crutschow I don't think the op amps could handle such a wide range of input voltage. Or am I missing something? This buck should take power from some photovoltaic cells and deliver it to a charging circuit for a Li-Po battery, as well as some other additional circuitry in the system. Is it feasible? Can this be done or should I approach the problem differently?

Thank you all very much!
 

Ian0

Joined Aug 7, 2020
3,181
I found this IC online that is used with buck converters:
View attachment 244473
I want to do something like this but I don't really understand where do they take the voltage to power up the op amps, as it does not show.
From the output of the "Internal Regulator". The purpose of the internal regulator is to provide a stable voltage to run the internal circuitry.

It's one thing finding an IC to do the job
(Try
https://www.ti.com/lit/ds/symlink/lm5163.pdf?ts=1627283551149&ref_url=https%3A%2F%2Fwww.ti.com%2Fpower-management%2Fnon-isolated-dc-dc-switching-regulators%2Fstep-down-buck%2Fbuck-converter-integrated-switch%2Fproducts.html

and quite another understanding how it works and how to design one from scratch.

A basic buck regulator consists of these four components:
An error amplifier
A triangle-wave oscillator
A comparator
A power output stage

You can put one together easily if you only need an input of 15V, but easy-to-understand components tend to get scarce at higher voltages.
Designing it from the middle - outwards

Have a look at
https://www.electronics-tutorials.ws/waveforms/555_oscillator.html
and half-way down you'll find "50% Duty Cycle Astable Oscillator". The output squarewave is not what you need - you need the waveform at pins 2 and 6, which approximately a triangle-wave, going from 5V to 10V.
Connect that to one input of a comparator, and a variable voltage to the other input, and note that as you vary the "variable voltage" you get a variable width PWM on the output.
The PWM drives the output stage. You can use it to drive a P-channel MOSFET.

So where does the variable voltage come from?
From the error amplifier. The error amplifier measures the difference between the output voltage that you want, and the output voltage that you have, because the difference between the two is the error.
The error amplifier is an integrator - it moves the output gradually higher if the error is negative and gradually lower if the error is positive.
One input is connected to a fixed reference (which represents the value that you want), and the other input is connected to the final output of the power supply.

It's worth having a good look at how the UC3842 works (because it seems to be the grandfather of all PWM power supply ICs)*



* Yes - I know that there may be great-grandfather ICs out there.
 
Last edited:

ronsimpson

Joined Oct 7, 2019
1,450
What output current. This will take 12 t o60V down to 5V@1A.
1627333525225.png
This is a really good place to start. Look at he opamps and voltage comparators inside the IC. The UC3842 often runs from 160 or 320 volts but the IC can not work at that voltage. The UC384X if often used in boot strap mode. If you must make your own "PWM" then copy the 3842 and or some other and power it from a external source until you get the bugs worked out then use boot strap to keep it working with out an external supply.
UCx84x Current-Mode PWM Controllers datasheet (Rev. F) (ti.com)
You will need to read up on boot strap.
1627334209820.png
 

Papabravo

Joined Feb 24, 2006
16,772
@Papabravo it is a requirement to design one, rather than to use an already existent module. I did some calculus regarding the Buck components, but I don't really understand how to design the PWM circuit without using an external power source as it would defy the purpose. @crutschow I don't think the op amps could handle such a wide range of input voltage. Or am I missing something? This buck should take power from some photovoltaic cells and deliver it to a charging circuit for a Li-Po battery, as well as some other additional circuitry in the system. Is it feasible? Can this be done or should I approach the problem differently?

Thank you all very much!
I was not questioning any requirements regarding the design of such a regulator. Regardless of how you go about it, you can't just slap one together, as that is the road to perdition. You need a deep understanding of what is going on inside or you WILL be "at sea" when things go wrong as they inevitably will. A vanishingly small fraction of the people who have ever designed a switch-mode power supply get it even close to right the first dozen or so times and are pretty much guaranteed to let the magic smoke out on multiple occasions. On top of all that you might as well get used to the idea that you're going to end up throwing the first one away. If all this does not intimidate you - it should. You might be better off punting and hiring someone who knows what the flip they are doing.
 

crutschow

Joined Mar 14, 2008
27,691
it is a requirement to design one
Who made this "requirement".
Is this some sort of homework or school project?

As Pb noted, designing and building you own buck regulator from scratch is not for the inexperienced unless you are somewhat masochistic. :eek:
 

Papabravo

Joined Feb 24, 2006
16,772
Who made this "requirement".
Is this some sort of homework or school project?

As Pb noted, designing and building you own buck regulator from scratch is not for the inexperienced unless you are somewhat masochistic. :eek:
More likely unfamiliar with the complexities involved. Even the single chip implementations can be daunting if you don't understand what is going on with the various blocks.
 

DickCappels

Joined Aug 21, 2008
7,687
On the other hand, a friend of mine who is more of a tinkerer than an engineer did a pretty nice job of getting a National Semiconductor buck converter chip working with a minimum amount of coaching using the available application notes and datasheet, and he didn't even use an oscilloscope!

Back in the times before monolithic switching regulators the engineer had a lot more variables to deal with, starting with choosing a topology. Today it can be much easier. Getting the finer points like noise, transient response and regulation right might take more insight and work.
 

crutschow

Joined Mar 14, 2008
27,691
If you really need to do this from scratch, and the output is to charge a battery, then I would look at a hysteretic (bang-bang) buck regulator.
They tend to have a higher output ripple voltage, but that shouldn't be a problem if you are charging a battery.

They require no error amplifier or triangle-wave oscillator, and are inherently stable so need no feedback compensation.

You could use a linear regulator from the input supply to reduce the voltage to an acceptable level for the control electronics.
 

Thread Starter

Sphades

Joined Mar 7, 2020
14
Hello everyone, and thank you for your responses!

I am required to design this myself as mean of learning the thinking process and design rules of such a circuit. Is more about learning than making something very efficient and I needed a point in the right direction and for such, I thank you all.

As for what @crutschow said, I will take a look into the subject you mentioned. I didn't really want to use a regulator due to it's inefficiency and high power dissipation.
 

Papabravo

Joined Feb 24, 2006
16,772
Hello everyone, and thank you for your responses!

I am required to design this myself as mean of learning the thinking process and design rules of such a circuit. Is more about learning than making something very efficient and I needed a point in the right direction and for such, I thank you all.

As for what @crutschow said, I will take a look into the subject you mentioned. I didn't really want to use a regulator due to it's inefficiency and high power dissipation.
One of the topologies you might want to consider is the flyback converter. It is basically a buck regulator with a transformer. The transformer can efficiently reduce the input voltage to a manageable range. If you're a visual learner then the series by Robert Bolanos is for you. This will get you started. It is quite extensive but delivered in bite sized chunks.

 

crutschow

Joined Mar 14, 2008
27,691
I didn't really want to use a regulator due to it's inefficiency and high power dissipation.
Since the current required by the control circuitry is usually small, the dissipation should be acceptable.

Some circuits use a linear bootstrap circuit to initially power the circuitry from the input, and then switch to using the output voltage when it reaches its normal value.
 

Thread Starter

Sphades

Joined Mar 7, 2020
14
I ended up using the IC i found for the switching part of the buck. At least for now. I might look up into more details later. Thanks everyone!
 
Top