Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!mnetor!uunet!seismo!rutgers!ames!amdcad!rpw3 From: rpw3@amdcad.AMD.COM (Rob Warnock) Newsgroups: sci.electronics Subject: Re: TTL Questions Message-ID: <17725@amdcad.AMD.COM> Date: Tue, 28-Jul-87 04:02:54 EDT Article-I.D.: amdcad.17725 Posted: Tue Jul 28 04:02:54 1987 Date-Received: Wed, 29-Jul-87 04:32:39 EDT References: <1395@crash.CTS.COM> <1008@speech1.cs.cmu.edu> <294@uvicctr.UUCP> Reply-To: rpw3@amdcad.UUCP (Rob Warnock) Distribution: na Organization: [Consultant] San Mateo, CA Lines: 54 In article <294@uvicctr.UUCP> collinge@uvicctr.UUCP (Doug Collinge) writes: +--------------- | I read a neato book once by a guy who actually went out and tested things | like this,... He said that open TTL inputs WILL NOT | GLITCH even under the most extreme conditions. Remember, before you flame, | he actually made circuits and tried it out... +--------------- Then he (whoever this anonymous person was) was very, very lucky, and was (in my opinion) grossly negligent in recommending such behavior to others! Yes, on a *small* breadboard on your bench in a quiet (electrically) lab, you can get away with it. I have. I do, still. But I (see signature below for a real name you can quote) have also personally been burned very badly (some 25 years ago) by that sort of thing, in production products. I had a manager giving me the same sort of stuff, "You don't need to pull those guys up! Pullups are expensive! The design time to wire them is expensive!", and I let him intimidate me into saving a few bucks. Of course, when the product failed in the field, it was my fault! (And those things are *hard* to diagnose!) In a high-speed, high-power system (and if you think TTL is not high power, calculate the total instantaneous peak power when you pull all 32 lines of a data bus from TTL high to low within a few nanoseconds, when the bus has several hundred pF of load per pin -- we're talking *amps*!), unterminated inputs *do* pick up trash and cause glitches, *especially* inputs such as preset/clear on a flip-flop. (Yep! They fail only on "some" data patterns...) *DO NOT* leave inputs floating on commercial products. (What you do in your home lab is another thing.) You may find yourself with a product liability suit otherwise (or a criminal charge, if the circuit fails in such a way as to harm humans, as in an elevator or an airplane). For the last 25 years, I have preferred to be safe & sure. As a quick & dirty rule of thumb, put a 1000 ohm to +5 and tie no more than 10-15 inputs to it. Use "SIP" packages of several 1k pullups to cheaply & safely terminate all sucj inputs. Don't use long wires on your pullups (they can pick up noise), use more pullups instead (closer to the packages being pulled up). Incidentally, if you are careful to never tie the set and preset inputs of a flip-flop to the same pullup, you can also win some side-effects in testability (when using "bed-of-nails" testers), since you can set/clear groups of flops by pulling down the pullup. (In fact, a testability engineer I once worked with wouldn't even let us tie constant-zero inputs to ground, but made us use a low-ohm resistor, so that the testers could override them for fault isolation. But that's another story...) Rob Warnock Systems Architecture Consultant UUCP: {amdcad,fortune,sun,attmail}!redwood!rpw3 ATTmail: !rpw3 DDD: (415)572-2607 USPS: 627 26th Ave, San Mateo, CA 94403