Power Supply Design FAQ
Frequently Asked Questions

Answers to frequently asked questions about power supply design.

On my switching-mode power supply design home page I give my policy for answering email which includes the restriction that I have to answer it in less than ten minutes from my memory. There are exceptions, sometimes I will spent one to four hours researching an answer and have spent over 40 hour researching a question if it has potential in ending up as one of the problem/solution pages on my website.

The problem is that I often get requests that would take me months to answer. For example, it typically takes me three months to design a power supply, not counting layout and testing of the prototype. (My personal best is 40 hours.) There are now computer tools available that were not available when I did design that can shorten this time, but it still takes more than ten minutes to design a power supply. Yet, the most common question I get, is to do this. My answer is in the first question/request discussed below, "Send me a power supply design for ..."

Send me a power supply (transformer, snubber, etc.) design for...

This question comes from both students and engineers. The students may have never designed a power supply and think it is simpler than it is. Since design takes more than the ten minutes I can allocate to answering a question, my answer to all is to find a reference design or application note close to the design that they want and then modify it for their purposes.

Which brings up the next question. Where do you get such a reference design or application note? My old answer was to do a search of the websites of those that make parts used in power supplies. The vendors who make power controller integrated circuits, power semiconductors, and sometimes magnetics. The starting point is to go to my vendor page Electronic Parts for Power Supply Design and use the "find in page" command on your browser to search on the keyword "semiconductors" (without the quotes) or any of the keywords. The signal to noise ration is below 50% but still manageable. This is going to take a lot of time, four hours or more, and some luck. This method still works, but there is a better way I would recommend first. IEEE Xplore has teamed with GlobalSpec in a beta program to search application notes as well as the IEEE literature. This is available to non-members. Go to IEEE.org and in the Find an article in the IEEE Xplore digital library search box (not the main search box), type in what you are interested in. For example, say you want design information on the SEPIC converter. Type in SEPIC. I did this (02/15/08) and got 204 papers on the subject. Notice that there is a tab in a navigation bar near the top that says Application Notes (Beta). I did this on the same date and got five application notes on SEPIC converters. Clicking on the GlobalSpec link at the bottom of the list got me a total of 187 application notes related to the SEPIC converter. Things keep getting better on the Internet! But you still have to review the abstracts and then download and skim or read those that you think will be useful. This takes time, and it has to be your time.

Which leads to the next question. Can you help me search your vendor page Electronic Parts for Power Supply Design more efficiently? My usual answer is to start with the Texas Instruments website, www.ti.com, work your way into power management products and then look for application notes and training that will help. Make sure you look at all the Unitrode seminars under training. You can easily spent a day on this website and still not find what you are looking for. But it has to be your time, not mine. Then you can expand your search to other websites. I will add to these and expands my comments, but try National Semiconductor, www.national.com, Maxim, www.maxim-ic.com, On Semiconductor, www.onsemi.com, Fairchild Semiconductor, www.fairchildsemi.com, Analog Devices, www.analog.com/en/, International Rectifier, www.irf.com, Intersil, www.intersil.com, IXYS, www.ixys.com, Linear Technology, www.linear.com, Microsemi, www.microsemi.com, Power Integrations, www.powerint.com, STMicroelectronics, www.st.com/stonline, Torex Semiconductor, www.torex.co.jp/english, and Vishay, www.vishay.com. But again, check out IEEE Xplore as described in the paragraph above.

What amazes me is the incredible wealth of design information in this haystack of websites when you browse them looking for a the needle, a reference design for you application. Like looking for a needle in a haystack, it takes time and luck. It has to be your time, not mine. I wish you luck.

Why does my power supply fail when I increase the load?

My power supply is working fine up to 25% load, but it fails when I try to go to full load. (Usually by failure of the power transistors, power diodes, or losing control.)

The failure of the power switches is usually due to leaving the safe operating area of the power switch when the load is increased. But it can also be loss of control due to electromagnetic interference (EMI) from the switches getting into the control electronics causing the power supply to lose control.

The solution to both problems is usually the proper use of snubbers.

The rapid switching of power switches in switching-mode power supplies can excite parasitic resonances in other components and the layout of the power circuit and cause ringing on the voltages and current waveforms. The effect of this ringing is three fold. First, ringing can drive voltage and current waveforms outside the forward biased (turn-on) and reverse biased (turn-off) safe operating areas of the power switches causing device reliability degradation (long term failure) or near instant failure of the power semiconductors and other components. Second, the ringing of high voltage and current waveforms can create local EMI that enters and upsets the low power control circuitry, in the worse case, commanding the circuit to destroy the power supply. Third, as the load increases, the increased switching current can cause more EMI and power dissipated in the power semiconductors can drive the junction temperatures beyond their ratings.

Snubber circuits are the usually solutions to these problems. They damp the ringing and shape the waveforms to stay within safe operation areas. They greatly attenuate the EMI generated by the power switches, although careful design and layout practices are also needed around the control circuits to desensitize them from EMI. Finally snubbers can be used to reduce the switching power losses of the total design or divert the losses from overheating semiconductors to snubber resistors.

But snubber circuits are not as simple as they seem. A February 16, 2008 search of the IEEE Xplore data base yielded 1,075 IEEE papers on the subject and through the linked GlobalSpec data base, 3,464 application notes. Virtually every text book on power supply design discusses some aspects of snubber design, but not all aspects. This should give a clue that they are not simple circuits in their application. Yet surprisingly, there is no book devoted to the subject. But that has changed with the release of Rudy Severns ebook on the subject, Snubber Circuits for Power Electronics, the first book every devoted to this complex subject.

Original: January 28, 2008, revised September 27, 2008


Webmaster and editor: Jerrold Foutz