Design of SMPS

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi there guys,

First of all, forgive me if my English isn't so fluent. :p

I had 2 courses in the field of Power Electronics. One of the class was "Power Electronics" and the other was "Switching Power Systems". However I didn't had the part of the converter control theory as I'd like to.

Even with not much free time, I want to start build a SMPS to feel the problems that can arise from the practical point of view.

I have read a couple of books/papers/thesis and have read lot's of post in 2 or 3 forums. The sensation that I have is that there is lot of people that are building their SMPS without knowing well enough about the SMPS and switching power systems theory. Ok, they can read the Pressman's book or Marty's book and some applications note and with that they can build their SMPS. But I wonder if they do know the theory behind that. The idea that I have is that those guys doesn't even have an university degree. So I ask, do they know how to project a feedback loop, a controller project if they understand what is the steady-state equivalent circuit model, the AC equivalent circuit modelling, state-space averaging model, canonical model of the converter? Or just grab the book from Marty Brown or other one and build them the circuit. Grab some application note, follow the recipe and it's done?

The same goes for those who says that build their commercial SMPS.

I really don't know if that is relevant and important to build a SMPS. Of course that I can grab the app. note and the Marty's book and build the SMPS, but building a SMPS isn't a great responsibility to the point that maybe following the recipe it is not a negligent position?

My greatest problem is trying to understand and the model of the feedback loop design; get the transfer functions, see the phase margin, gain margin, etc.

On the other hand, I don't know at which point is important to know deeply how those things work in order to design a SMPS. We have several IC's that do the control part. We can just grab the datasheet and an app. note (as I told before) and understanding that we can design the feedback loop.

There anyone that works in this industry and could share his point of view and what think about what I said?

Thanks for reading. I'll wait for your answers
yes you are correct that many people don't know well and fully understand the operational principle of smps and to tell you frankly i can consider myself as one of them..but this doesn't prevent me from doing the experimentation about the spms which I have been doing in actual..I don't care if my background is not enough but as long as i fully understood the basics of electronics and the safety associated with it especially and particularly to smps..i won't let those very deep theories to bother me becoz i believe the start to fully understand the theory is to begin it in an actual..and i can tell to anyone that there are things in actual which sometimes not being described in theory..thanks..sorry if i may sound sarcastic..but it's not my intention..
More and more people choose the switching field, enticed by the obvious reduction in size and cost, many of them have almost no ideea of what is actually happening in an switching power supply, they just feel that they need to be in the trend, and that no SMPS can stay between them and theyr wrong they are...

I think there are 3 grups of people in this domain:

1.-Well experianced technicians that can always design anything they need ( and mostly they do ), they have a deep understanding of the theory involved in mostly anything thay work with, some of them feel confortable to share with others theyr knowledge and help the ones in need, some of them do not, eyther way it seams there are not enough of them.

2.-This is where the ones that try to reproduce copy/paste the work of others, belong to. Not that is anything wrong with it, they simply do not have the skills to handle the insides of switching, so they try to take it step by step mechanically, sometimes developing theyr own stile on handling stuff, sometimes things go ok and most of the times they do not, they find themselves stuck on something cus they did not pay enough attention to the theory involved in.

3.-The most numerous category is the one containing the ones that don't know much of anything but still they feel a SMPS is simple and easy to built, they keep come and open threads on electronics forums and ask for some schematic without havin any ideea what is involved, and most of the times they just get stuck at the first technicall response of someone else, not knowing what to say they just give up, many, too many of this situations...

I could write a boock with what i know of the feedback loop and EA compensation ( for example ), but i wonder how many people would realy understand it...
SMPS design can be highly complex.

I glibly just threw together a flyback SMPS and it didn't work it just blew the mosfet.
I was then told I needed a gapped transformer which moved things on a bit.

I then had a go at a half bridge by taking an app note design and trying to mod it for my own voltages and that fell apart too. A half bridge design is highly involved with maths to work out the transformer and the control components.

However I did get there eventually but it cost a lot in components and time and a lot of help from a forum member.
I had the same sensation. My firsts books were Marty Brown and Pressman, but a lot of datasheets and application notes I read at this time.

I have running actually a couple of them. One is a simple FlyBack using UC3842, a MOSFET, and a recycled core from a PC monitor. This gives me +-14V to power a couple of TDA2006 and 2003´s, and +12 and +5 for 2 old PC cd readers, and it is working from about a decade all weekends when I´m at home, almost 15 hours per day. In some circumstances of power line failure, I could us it once started, from 220V (Nominal line voltage in Argentina) down to 36Vac (you obviously cannot believe me, but I swear that it is true). (the aluminium box in the first photo).

Other is a solar cell battery charger for my ham radio station. It has benn constructed almost all using standard discrete devices, for example, a TS556 (CMOS version of 556) is used as PWM'er. It consists of a 4047 clack generator driving them in antiphase, and amplifying the current, it finally get the gate of a two buck converters working 180deg out of phase. (The black box with analog instruments).


  • Audio.JPG
    132.2 KB · Views: 423
  • Solar Cells 01.JPG
    Solar Cells 01.JPG
    140.4 KB · Views: 414
  • Reguleta.JPG
    119.8 KB · Views: 400
This is your problem, not mine. I saw working with only 36VAc and is sufficient for me. The design parameters was standard 90-240VAC, but once started at 100V, is collapses about 30VAC with light load as a class AB audio amplifier with low volume, but not zero.

Respect to closing the loop, the AN "slup068" from TI is very clear and easy to understand. It was of help for me.
Last edited:
Not all flyback controllers need a minimum load, the ones designed for capacitor charging turn off when the capacitor is charged and wait till the voltage falls before firing a few shots to keep the capacitor charged. It is quite possible for flyback converters to work over a wide input voltage range provided the turn off speed of the switching device is fast and the stray capacitance is small. Naturally the output current will fall with falling input voltage Energy stored in an inductor is 1/2LI^2 and the the charge rate of an inductor dI/dt= V/L so at low voltage there is insufficient time to store much power in the flyback inductance.
Reducing input voltage to 40% of normal for a constant frequency discontinuous mode flyback which was reaching a maximum duty cycle limit will decrease the output power to 16%. If there is no duty cycle limit the output power can be higher than that.

You might consider that 'real fast', but it might be more than enough to power the nearly idling amplifier. :)
It's one thing to say it could work with verry low Vin in some strict conditions, but it's a whole other thing to claim it is workking no matter what, and that is my point, you cannot hope to get good voltage regulation and power with souch low Vin, it could start up, or keep working but only in some conditions that you already have mentioned verry well.

Anyway, this does not help much anyone, i feel we are getting stuck on one topik.
It is all about your life. You can read and read it. Make SPICE models, struggle with close loop control, stability, than build it just to see than it doesnt work in way you hoped it would. Than, at the end of the day you just place capacitor 0.1 uF in the FEEDBACK amplifier control ( make it slow) and that is it. It works, no more pain with oscillations and unstable operation =)

From the other hand, you can download PWM controller data sheet, follow supplier recommendations, spend minimum time and have a very good result.

So what do we choice? Depends on what we what to gain from it. That is what happened to me.
The Flyback topology is said to have the largest dynamic range, I could't believe when before a lightning bolt that gets the mains power supply in Barrio Garay, and a fire artwork occurred in the power lines and distribution transformers. The voltage at line slowly from 180 down to 0V and vice versa. The SMPS at my audio set starts alone at about 85-90V and gets working up to a moment that the system become unstable and about 35V then collapses and no starts again till voltage goes to 85VAC again. Sincerely, I couldn't see what I was viewing.

Also, are the easiest to design.
If by easy you mean a smaller part number then yes, i agree, but if you mean an easyer theory and calculations then it is clearly you haven't design anything woth mentioning... i could fill more than a few pages with calculations just for the power transformer...

I designed two: the one for the audio equipment (80W), and another one for powering a wireless phone (7W) from my solar power circuit. As both of them has been conceived with materials recycled or from what I have in the junk, I had 1/2 the problem resolved. Both are currently working at home. Certainly, never design for commercial or industrial usage.
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.