Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!cs.utexas.edu!know!samsung!dali.cs.montana.edu!milton!whit
From: whit@milton.u.washington.edu (John Whitmore)
Newsgroups: sci.electronics
Subject: Re: 2 questions
Message-ID: <4935@milton.u.washington.edu>
Date: 11 Jul 90 06:03:42 GMT
References: <36807@shemp.CS.UCLA.EDU>
Distribution: usa
Organization: University of Washington, Seattle
Lines: 40

In article <36807@shemp.CS.UCLA.EDU> loving@lanai.cs.ucla.edu (Mike Loving) writes:
>1)  Can someone out there recommend one/some good suppliers of electronics
>parts.  I am already familiar with Digi-Key (expensive) and Jameco...

	Try Allied Electronics (1-800-433-5700 or the Tustin, CA branch
at (714) 669-4190), or Active Electronics (1-800-ACTIVE4).  Both have
a sizeable selection (down to discrete transistors and diodes), and
a small minimum order.  I trust they send catalogs on request...
>
>2)  I am working on an EPROM programmer.  One of the requirements is to be
>able to switch someof the lines to various (well, regulated) voltages and
>supply anywhere from 30 to 100 ma.
>... using an LM317 adjustable regulator
>...  the specs ... are so loose as to require a feedback control loop with an
>A/D converter to insure accurate voltages.

	It's not that grim; there's two resistors in the regulator
circuit, and they DON'T interact too badly.
	You can trim the resistor from output to sense so that the 
1.2v balance on the sense terminal corresponds to EXACTLY 5 mA of
current in the upper resistor; then the lower resistor will be chosen
so that 
	(5 mA) * R = Vpp-1.2V
Note that, having trimmed the sense-to-output resistor for a known
current at balance, the uncertainty in that '1.2V' value is ONLY ADDED
to the uncertainty in Vpp, NOT multiplied by a gain and added.
The achieved accuracy depends on the accuracy of R (presumably 1%
or better) and the (approximately 0.05V) unit-to-unit deviation in 
the sense voltage.  This amounts to about 0.25V uncertainty in
a 21V programming voltage, which is certainly acceptable.
You DO need to trim that one resistor; the only ICs that have
accuracies of 1% or better in voltage references are those
that are factory-trimmed (at great cost to you, the consumer).
	The Vpp voltage is then resistor-programmed by the lower
resistor selection.  Wire in all the resistors you could want,
and ground only one at a time (or put in R/2R/4R/8R/16R/32R and
ground some combination... make your own DAC!).

I am known for my brilliance,             John Whitmore
 by those who do not know me well.