From: "Mike Engelhardt"
Subject: Re: power supply pwm design help
Date: 29 Sep 2002 20:01:38 GMT
Organization: Concentric Internet Services
X-Newsreader: Microsoft Outlook Express 5.50.4807.1700
> First, the topology I was referring to is *NOT* a SEIPC. What
> *I* was referring to is indeed called a Buck-Boost. This
> topology has 2 transistors and two diodes..
Oh, I've always heard that called a 4-switch. A SEPIC is
a better choice here. It also has the advantage that there
only one switch drop.
> My comment was not directed at you personally, but rather the apparent
> suggestion one can substitute actual undersanding of the the underlying
> principles of SMPS design, by having the software do the work.
But the software does give an excellent solution for the original
poster's problem. My impression was that you only have a
problem what you assume the problem is and what you assume
people are saying.
> When the design doesn't work for some reason, the newbie
> (to this field of design) is left without a clue how to
> make it work because he has no understanding of
> the basics.
In general, the synthesizer just gives an initial design. Then the
user is free to modify it an re-simulate it before bence
verification. In the case of the original poster's problem, the
synthesizer's solution is a good one.
> My opinion of Switchercad is that it is a nice piece of
> sales material intended to lock you into using Linear
> Tech's devices. Sure, P-Spice (upon which it is based)
SwitcherCAD III is not based in P-SPICE but LTspice. There
shouldn't be a need here to comment on things one doesn't
understand. It is not intended to lock you into using
Linear parts, as the license allows the user to perform
general circuit simulation simulation and allows the
importation of 3rd party models. Of course, it does help
people to use LT products, as it includes about 650 models
of LT products.
> is a good tool, but not for an amateur who doesn't
> understand the principles. I am reminded of new engineers
> at a former job who spent weeks (even months) "designing"
> RF devices with their nose glued to the computer monitor.
> Once the design was built, it generally did not work as
> expected (or often, at all). The comments generally ran
> along the line of "Well, there's NOTHING wrong with the
> design, because it worked in SPICE!". They were generally
> told something like "Welcome to the REAL world".
But the same thing could happen to someone who spent months
doing analytical analysis of the problem. So what? I've
always thought that the best understanding of electronics
comes from a combination of bench work, analytical
analysis, and simulation.
The original posters would still be well advised to punch
his problem into SwCADIII and get a practical solution.
LT does sell a one chip four-switch solution that I
think works here, but the SEPIC is, in my opinion, a