Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!rutgers!ucla-cs!zen!ucbvax!zion.berkeley.edu!max
From: max@zion.berkeley.edu (Max Hauser)
Newsgroups: sci.electronics
Subject: Re: TTL Questions
Message-ID: <20036@ucbvax.BERKELEY.EDU>
Date: Sun, 9-Aug-87 09:33:47 EDT
Article-I.D.: ucbvax.20036
Posted: Sun Aug  9 09:33:47 1987
Date-Received: Sun, 9-Aug-87 21:52:39 EDT
References: <7105@alice.UUCP> <764@sol.ARPA> <128@umich.UUCP>
Sender: usenet@ucbvax.BERKELEY.EDU
Reply-To: max@eros.berkeley.edu (Max Hauser)
Organization: U.C. Berkeley
Lines: 52
Keywords: TTL, pullups
Summary: Watch out!

In article <128@umich.UUCP> don@umich.UUCP (Don Winsor) writes:
>In the recent discussion on TTL there seem to be numerous
>recommendations to use a series resistor when connecting an unused
>input to +5 volts.  This is an unnecessary waste of a resistor, the PC
>board space to mount it, and the foil trace to connect it.  Can't we
>finally give pullup resistors a decent burial?  Unused inputs should
>be connected DIRECTLY to the appropriate (+5 Volt or Gnd) supply rail!


Speaking from a historical point of view, I humbly remind the net of
the excellent reason for including the resistor with pre-Schottky
TTL families. TI went so far as to publish an application note for
the 54/74 series, explicitly admonishing against connecting inputs
to the positive power rail under any circumstances.

The reason is because standard TTL inputs, consisting of emitter
diffusions in an NPN transistor, exhibit typical Zener breakdown
voltages of 5.5 volts. Since standard TTL power supplies can be
5.25 volts within spec, a noise spike of 250 mV on the positive
supply would be enough to break down the input and possibly damage
the chip in the absence of a series resistor. If the Schottky
families (where the inputs are isolated Aluminum-Silicon Schottky
junctions) do indeed have a higher breakdown voltage, information
I don't have at hand, then the resistor is unnecessary (although
still harmless).

Perhaps I am old-fashioned but I prefer stating the justification
both for the original requirement of resistors and the contemporary
waiver of that genuine need, so that people will know the reason why,
not just follow commands.

I should add that one never, of course, uses a resistor in series
with inputs tied to ground, for current-sinking logic such as TTL
families (of all kinds), since the major input current flows in the
"low" state, causing a voltage drop that can cloud the logic "0";
and reverse junction breakdown is not an issue in that case.

Also, one can indeed tie unused logic-gate inputs to used inputs;
the penalty is an increased capacitive load to the preceding stage
(not an increase in "AC noise", since (1) the dominant capacitance
at each gate input by far is from package parasitics, not internal 
to the die and (2) two inputs being driven together causes essentially
the same parasitic capacitive currents to flow in the driven chip
as one input driven, for little real change in the AC noise picture).

Respectfully,

Max Hauser, UC Berkeley EECS Department, IC Design Group

   UUCP: ...{!decvax}!ucbvax!eros!max
   Internet (old style): max%eros@berkeley
   Internet (domain style): max@eros.berkeley.edu