Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!cmcl2!beta!hc!ames!ucbcad!eros!max
From: max@eros.uucp (Max Hauser)
Newsgroups: rec.audio,sci.physics,sci.electronics
Subject: Specs
Message-ID: <1914@ucbcad.berkeley.edu>
Date: Sun, 4-Oct-87 09:16:49 EDT
Article-I.D.: ucbcad.1914
Posted: Sun Oct  4 09:16:49 1987
Date-Received: Wed, 7-Oct-87 05:38:15 EDT
References: <3816@watdcsu.waterloo.edu> <578@uthub.toronto.edu>
Sender: news@ucbcad.berkeley.edu
Reply-To: max@eros.UUCP (Max Hauser)
Distribution: rec
Organization: University of California, Berkeley
Lines: 70
Summary: Jeff Segawa on specs vs. quality
Xref: mnetor rec.audio:3554 sci.physics:2380 sci.electronics:1487

In article <899@uhccux.UUCP> cm450s02@uhccux.UUCP (jeff t. segawa) writes:

> Bill Blue: I couldn't agree more with this business about tecnical
> specifications and subjective sound quality.  I bought my first 
> stereo based on those distortion specs. ...
> ...    Maybe the source of a lot of these
> "looks good on paper, sounds awful" sort of problems have a lot to
> do with the WAY these measurements are taken. For example...
> [typical flaky-turntable example] ... Naturally, the spec
> sheet told me that the turntable had nearly perfect speed accuracy,
> and it did, so long as I played 1Khz test tones. 


No kidding. Manufacturers who want your money will actually 
measure specs that are easy, and don't tell you the whole story.
Welcome to the real world.

It's worse, by the way, in industrial technology, where you don't
usually even have your ears to fall back on. That is why I assume
that any well-trained engineer knows all about specsmanship
(indeed, the clever people who documented that turntable certainly
knew about it!).

When I was a graduate student at MIT, I took an "advanced 
circuit-hacking" course from an experienced engineer who 
understood things like this. One of our first assignments was to
take a commercial analog IC (the Intersil 8038 voltage-controlled 
oscillator), go into the lab, and as he put it verbally, "find 
five lies in the spec sheet."

What I object to in postings like the foregoing is the suggestion
-- and I hope it's only in my imagination -- that because a few 
simple specs per se, the kind usually published by manufacturers,
fail to accurately describe an audio component, then ALL specs are
useless and we must throw our hands up in the air, abandon
specifications entirely, use our ears exclusively, eschew 
electricity, eat only beets, what have you. Of COURSE little
numbers like distortion won't tell you how the component sounds.
That manufacturers can oversimplify and lie, and that some
engineers (just like physicians) are not familiar with this
particular problem, does not mean those specs are useless as
a tool, properly applied; or that you could not make further 
measurements that more completely characterize the component.
I think that Jeff also alluded to this in his posting.

What we REALLY need, of course, is consumers who are 
informed-and-critical rather than astonished-and-cynical.
Such consumers would know about specsmanship and about the limits
of simplistic manufacturers' numbers, with their tacit dependence
on assumed linearity, test conditions, and so forth. Then people 
would not be so shocked at the revelation that it is actually 
complicated to accurately specify the performance of a
sophisticated audio component, and that a few simplistic curves
don't tell the whole story indeed. With such enlightened consumers,
not only would it be unnecessary for the superstitious to throw
out the technical baby with the specsmanship bath (if you will),
but also, manufacturers would not get away with the shit they now
do because more people would call them on it.

I promise you that given any audio component, I can produce nice-
looking, and even sexily-named, performance curves that will not
reveal the full sound or the weaknesses of the component.

By the way, if you need to remove epoxy from a "proprietary" 
potted module inside of an audio product, as Dick Pierce
mentioned, use Dynasolve (tm) 160. It's much cleaner and gentler
to the circuitry than "chiseling." I have a whole file of
schematics that I got that way in 1971-72.

Max Hauser / max@eros.berkeley.edu / ...{!decvax}!ucbvax!eros!max