Path: utzoo!news-server.csri.toronto.edu!cs.utexas.edu!usc!randvax!edhall
From: edhall@rand.org (Ed Hall)
Newsgroups: sci.electronics
Subject: Re: 12 to 9 volts DC
Message-ID: <1991Mar3.221651.14040@rand.org>
Date: 3 Mar 91 22:16:51 GMT
References: <5170111@hplsla.HP.COM> <4596@ash17.UUCP>
Sender: news@rand.org
Reply-To: edhall@rand.org
Organization: The RAND Corporation, Santa Monica, CA, USA
Lines: 35
Originator: edhall@ives

In article <4596@ash17.UUCP> dunne@motcid.UUCP (Jim E. Dunne) writes:
>From article <5170111@hplsla.HP.COM>, by tomb@hplsla.HP.COM (Tom Bruhns):
> 
>> (hee-hee-hee --  you're with Motorola and can't find semiconductors?? ;-)
>
>   OK, I'm a software hack in Cellular.  Hardware devices crash when
>   they see me comin'!  Actually, right upstairs from me, where I knew
>   they had scads of data books, I found out they had a cache of sample
>   parts too.  I raided the cabinet, only to find 5, 8, and 12 V
>   regulators!  Oh well, I grabbed an LM317 adjustable.  To the mailers 
>   who said the adjustable needed only two resistors, well the data book
>   shows otherwise.

You are correct--the databook shows two bypass capacitors, both optional.
One is at the regulator input, and is needed when the power supply filter
capacitor is an appreciable distance from the regulator (e.g. in a wall
wart, as is quite common these days).  The other is at the regulator
output, and improves transient response (i.e. better regulation when
the load varies at a high frequency, such as with digital logic).

If you look at the databook for the 78XX, you'll see the same optional
capacitors.  All that the LM317 adds is two resistors, and a great
deal of convenience.  I find that using a single part for a variety
of voltages (including odd values) is a LOT simpler than having to use
a different part for a different voltage.

>   Long live the KISS principle...
>    -- Keep It Simple Stupid (Softwarejock)!

As a software jock myself, I agree.  That's why I prefer to have a
single procedure that I can call with a variety of parameters rather
than a different procedure for each parameter.

		-Ed Hall
		edhall@rand.org