On Sat, 3 Aug 2002 10:30:35 +1200, Russell McMahon wrote:
>>That's not the one off chip price -- that's the price for Qty 1=
 of
>>their PS demo board for the chip.
>
>True - my mistake (reading too fast and/or uptoo late :-) ).

I did a double-take also. Then I reread it, I couldn't believe=
 National
would have a $22/each switcher chip of ANY kind! ;-)

>A better choice MAY be the LTC1624 controller
>
>http://www.linear.com/prod/datasheet.html?datasheet=3D395
>
>Similar functionality but
>
>- Drives an N Channel FET (implements a high side pump for gate=
 drive
>supply)
>
>- =A0 =A0Wider input and output voltage ranges
>
>- Able to implement other topologies.

OUCH. Without even looking I knew... $4.30 (Qty. 1-500) for that=
 part.
Linear Tech makes great parts, but they are DAMNED proud of them.=


>They use an external current sense resistor but it is in the=
 high side
>=A0and drops only about 160 mV (AFAIR) so this is fairly=
 bearable.

Have you priced low ohm, precision current sense resistors=
 (especially
SMD ones)? Another OUCH. In a high side application, taking=
 advantage
of a FET's disadvantage - Rds(on) - is a good thing.

>N Channel FETs are generally cheaper, more available, have=
 better
>Rdson and are available in higher voltage ratings (not an issue=
 here).
>
>All these small pinout modern chips tend to have lower=
 functionality
>compared with many existing "full house" switching regulator ICs=
 with
>more pins and therefore access to more functionality. Despite=
 this
>they look very useful for simple dedicated applications. Even=
 some
>VERY old SMPS ICs (20 + years)(eg xx384x family and many more)=
 will
>implement many of these functions with a little design work -=
 external
>=A0component count may be higher. If you want absolute minimum=
 parts
>count then finding an IC with internal pass FET and flyback=
 diode
>adequately rated for your application is the "way to go". Cost=
 MAY be
>higher.

I'm a big fan of TCA (total cost of assembly) since most all my=
 designs
are destined to be built and tested in some quantity for some=
 customer.
PC board space, component cost, assembly labor and test labor are=
 all
factors in this. Generally, a bullet-proof switcher design that=
 has
fewer parts but slightly higher parts cost will be the best=
 solution
for all but the highest volume, cost sensitive applications.

For my personal hobby use, I could care less about parts cost=
 most of
the time since I building one-offs and generally using free=
 parts.

That's the great thing about engineering -- there are a large=
 numbers
of solutions to a given problem based on your requirements and
perspective. None of them may be absolutely right or wrong, but=
 some
will certainly be better than others... ;-)

Matt Pobursky
Maximum Performance Systems

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.