On 8/14/13 8:49 PM, Peter Gutmann wrote:

Another thing about switchmode design is that unless it's pretty simple, you're better off outsourcing it (meaning, get a pre-built one).

WhatHeSaid. Switchers are notorious for being persnickety about components tolerances, and are sheer hell to troubleshoot as well. And the bigger you get, the harder they fail... I love the smell of burning bakelite^H^H glass-fiber^H^H electrolyte in the morning...

Thanks for the links & design pointers. I'll willing to admit that I'm not an expert, but always eager and willing to learn something new! [yes I know the lurker finally speaks...lol] I'll try yo keep this brief. A little history about my background. I was born in the chicago area a dual citizen (American/German[Transylvanian]), from good stock. Ingenuity, design, art, music, and sciences, are some of the traits I share with my family & relatives. I studied electrical engineering for 3 years in high school (graduated early), then 2 more in college before switching into the computer network field. While in college I worked at the sba patent office, reviewing patents and researching designs. Going back a little further, when I was a kid I ran a small repair shop, fixing tv's stereos, etc (anything with cord or ran on batteries). I had a real knack for fixing broken electronics so they wouldn't break again. In the early 80's I got interested in computers and have been online since 1985. I used to be an avid reader of 2600, and frequented the meeting in chicago during the 90's. [I admit I still have my first redbox]. It was also during this time frame I became heavily involved in amateur and cb radio. The radio experience led me down the path of becoming an On-Air personality, where my first radio program attracted 3 million listeners. :) I'm not afraid to learn theory and good design before embarking on a new electronics project or any project for that matter. I do admit to suffering from hobby overload from time to time, especially since I'm interested in almost everything. I have learned almost everything I know by reverse engineering. At my present job I am well known for having the ability to reverse engineer a new technical process, document it and train others. Lately I've found myself burning out of the network engineering field, so starting projects such as these help me keep things in perspective [house, car, food, wife, etc] -Christopher Olesch *"Affordable IT Services for Non-Profit & Small Business"* || http://www.ngotechnology.org/ || http://www.linkedin.com/in/chrisoleschjr *Masonic Affiliations:* || http://www.scottishritechicago.org || http://www.supremecouncil.org/ || http://www.ilmason.org/ *Online Artistic Portfolio* || http://cjolesch.deviantart.com/ On Wed, Aug 14, 2013 at 12:44 PM, Riad S. Wahby wrote:

Chris Olesch wrote:

...

reading thru rsw's thesis as well.

I doubt you want to use anything so zany :)

What you're talking about is a relatively high power design, which means that in addition to worrying about designing the switching supply, you're going to have to sweat the practical details regarding PCB layout, et cetera.

I'd have a look at some of the reference designs and application notes from the usual suspects (onsemi.com, ti.com, maxim-ic.com, linear.com), since they will have a lot of good practical information on getting the supply built. For example, http://www.onsemi.com/PowerSolutions/taxonomy.do?id=448 seems to have a substantial amount of reference material available.

For more general information on theory and practice of dc/dc converter design, "Fundamentals of Power Electronics 2e" by Erickson and Maksimovic is my go-to reference. It more or less assumes you're already an electrical engineer, which may or may not be useful to you. Along these lines there's also "Principles of Power Electronics" by Kassakian et al, a book dear to me but certainly not to all.

If you're looking for a gentler introduction, Google should be able to point you to a few tutorials, e.g., http://www.maximintegrated.com/app-notes/index.mvp/id/2031

I've been kicking around an idea for a weird little buck converter controller that I haven't actually built yet. It's not particularly well- suited for kW converters, though, since the most basic form of the idea involves using a P-channel device for the high side switch. For any reasonable amount of power you end up wanting to use an N-ch, as they're generally higher performance devices than their P-ch counterparts (in silicon, hole mobility is about 1/3 of electron mobility, so devices that control electrons are generally higher performance than devices that control holes).

Cypherpunks build circuits,

-=rsw

-- -- -- http://cjolesch.deviantart.com/