Re: Step-down switching regulator advice ?

Joerg wrote:
Winfield Hill wrote:
Eeyore (Graham) wrote:
mpm wrote:
Eeyore wrote
I need a popular (easily available), low cost step-down switching
regulator suitable for an input voltage of up to ~ 18V that will
deliver ~ 1A @11V with Vin ideally as low as 11.25V.

LM2575's switchers are pretty popular. Second sourced by
ON Semi, and possibly others. Not sure about the 0.25V
drop though. And you'd need the adjustable one - requires
2 external programming resistors.

The low voltage drop is important to me. I suppose an alternative
might be a buck-boost circuit. This would actually be even
better suited to my needs. In which case my spec would be
1A @ 12.6V output with 11-18V in.

Let's try electronics design with Google.

At the right of your post in Google Groups there's
a "related pages" panel, where I see National Semi's
LM2587 chip is recommended. Two clicks gets us to
http://www.national.com/pf/LM/LM2587.html where we
see that a simple flyback-transformer converter might
work. In a panel to the right we see NSC's WEBENCH®
tool to test the thought. OK, we enter 10 to 20V in,
12.6V out, 1.5A max, and press START YOUR DESIGN.
Oops, they want me to log on. Sigh. OK, press START
again, and whammo, there it is -- our design, pretty
routine, except a custom transformer is called out.
But it's an easy one, with Ns/Np = 0.78, Lp = 23.4uH,
and Lell no more than 0.47uH, or less than 2% of Lp,
which is pretty easy to wind. They specify Rp under
35 milliohms, which should help pin down the bobbin's
winding area, and a ferrite-core size. They say the
LM2587 will dissipate 3.5 watts, so we'll need a heat
sink clip on the TO-220 tab.

3.5W at 11W out? Ouch.

Hmm, DigiKey wants $8.32 just for an LM2587 TO-220-5
IC. Maybe the $10 TI module isn't so bad after all.

Double-Ouch. I usually do SEPICs with the LM3478. About
a buck in quantities and that external FET ain't expensive
either. If it has to be dirt cheap I'll use a Schmitt
inverter. And it usually has to be dirt cheap.

I think a $1 LM3478 controller is better than a Schmitt
logic IC, but you're right, a small-die IC and a MOSFET
makes more sense than an expensive high-current part.
Especially when said part's switch dissipates so much
power. A decent MOSFET (looking for a cheap part) like
a 100V 0.1-ohm IRLR3410 (21.6 cents at DigiKey, 170,775
in stock, wow!) would only dissipate about 0.4 watts,
plus a little bit more for switching.

WRT using Google for engineering, presumably NSC would
be more interested in paying to promote their $4.95 part
(qty 500) on Google than a $1.10 part, but since Google's
"related pages" are not paid links, SFAIK, they used some
other rubric to pick the LM2587 IC instead of the superior
LM3478 + MOSFET, etc. Well, computers aren't too smart.

It's like TV dinners. Either they are bland or expensive.
When you whip up your own from scratch it costs less and
tastes better.

.