How are drivers QC'd?
Chu Gai
ought to be QC'd for with incoming drivers and approximately how long
might it take? Are 100% of the drivers tested and if not, how does one
arrive at a statistical number and sampling technique for batches that
come in? Just curious.
Eeyore
Chu Gai wrote:
> For a speaker manufacturer,
You mean a speaker cabinet assembler ?
In normal commercial situations I'd expect minimal if no testing at all.
Graham
Ethan Winer
> Are 100% of the drivers tested
I have no idea what's done today, but years ago I worked for a summer at
Bozak, and they tested every single driver by running a full range sweep
tone at a fairly high volume. As I recall it took about 5 seconds for the
tone to sweep from low to high. The tester simply listened manually for
buzzes and rattles etc. That was done in the room adjacent to where I worked
in building and testing power amps. Which is why I stayed for only one
summer. :->)
--Ethan
dpi...@cartchunk.org
Having both consulted to quite a few speaker system
manufacturers as well as ran the engineering department
for a driver manufacturer, I can say with some reasonable
authority that the range of testing is all over the map.
Some manufacturers test every driver for what they
consider relevant parameters. Other have a contractual
agreement with their suppliers to provide this service
for them. Others assume that if the drivers made it to
them, they must work.
What's required vs what's done is a different question,
and does not have a single, simple answer either.
A manufacturer making a sealed box system with a
high tuning ratio can tolerate a far wider variance in
the woofer suspension compliance than can a
manufacturer of a high-order vented/equalized system,
for example.
A lot of manufacturers require custom drivers that are
not available off-the-shelf, often the case for woofers.
If they don't make the drivers themselves, it behooves
them to have an arrangement with their supplier to assure
that the proper components are being used, that the
drivers are being properly assembled and tested. It's
to no one's economic advantage to skimp on any of
these stages.
One of my clients was making a system for which I had
designed a 6 1/2 mid-woofer. I had spelled out certain
crucial parameter carefully and unambiguously, even to
the point of specifying means, variances and standard
deviations acceptable. Unfortunately, when the first 2,000
drivers delivered, and when I did a spot check. I found
quickly (within the first 20 drivers I had randomly selected)
that the factory had failed to follow the terms of the
agreement. The measurement clearly showed that they
had used to very different suspensions that even though
they were visually identical, the result was 60% of the
woofers had a compliance that was too low, 40% were
too high.
Now, the simple mean was what I wanted, but the
variance and deviations were far out of what I had
specified. I basically rejected the whole lot. The
factory had to swallow them and start all over. You
bet they got it right the next time!
Another instance illustrates a case of lack of attention
to detail. This was for an 8" woofer. The polybutadene-
styrene surround had a fairly large gluing "flat" where it
was attached to the frame, and there were six rectangular
cutouts to provide clearance for the mounting bolts. The
factory never bothered to include a procedure for
aligning the cutouts with the bolt holes, with the result
that about 90% of the drivers were unsuitable: if you
attempted to mount the driver, the bolt head tore the
suspension from the frame. All 1500 of them were rejected
as well (different factory).
The short answer is every manufacturer is different and
has different procedures and requirements. Some of them
are excessive, some are sufficient for the task, and some
fall woefully short.
rustybx
October 1987. My how time flies.
For the newbies do a google groups search on "Dick Pierce anecdotes".
Many, many fun and interesting stories.
Date: 02 Oct 87 19:02:31 GMT
From: r...@teddy.UUCP (Richard D. Pierce)
Subject: The Best of Audio Anecdotes of the <interval>
Organization: GenRad, Inc., Concord, Mass.
Audio Anecdote of the <interval>
Today's episode: How to hit a moving target by shooting yourself
in the foot.
A decade past, I worked for a brief period for the great
consortium of companies known as Harmon International. Included
in this collection of companies was, of course, Harmon-Kardon, as
well as JBL, Ortofon, and Tannoy (Remember- Tannoy without the
"T" spells "annoy"!)
I was fortunate (?) to visit the JBL facilities in and around Los
Angeles, and there gave witness to some of the most incredible
things I have ever seen.
First, who is the largest producer of electric fog-horns for use
in marine navigation, such as lighthouses? Why, JBL, of course.
They took over that title when the Wurlitzer Organ company gave
up the ghost.
I was given the 50 cent tour of the factory. First, to the
cabinetry department. Here is a model of reasonably efficient,
good-quality high volume cabinetmaking. Panels are cut to size
and routed using appropriate jigs. Glue is then spread on the
joints and the cabinet is assembled and held together using
gigantic rubber bands. The whole affair is popped into the
woodworking equivalent of a microwave oven (actually, an MHF RF
heater) and the glue is "cooked" for about 5 minutes, after
which, the cabinet is ready for finishing. So far, so good. The
cabinet are inspected for any damage, and bad units rejected. So
far, about $25 has been invested in a cabinet for something like
a JBL L-100. Now the fun begins. The rejected cabinets, rather
than being tossed away, are shuttled over to a room occupied by a
bevy of middle-European craftsmen, who take the cabinets, sand
off the veneer, and glue on a whole new veneer skin. This takes
several hours per cabinet, and, as I later calculated, costs
about $100 per repaired cabinet. Now, given a reject rate of
about 20%, this means that it costs nearly $40 per cabinet when
you spread the cost of repairs over the entire production run!
Now you know why JBL products can be so expensive!
By far the most shoking discovery was the "quality control"
department for drivers. At first glance, it seemed
straightforward enough. A Hewlett-Packard dual-channel audio
spectrum analyzer was used. The driver was placed in a test
chamber, and a sweep signal applied. This was read into one
channel of the analyzer, and compared to the response of a
"reference" driver stored in the other channel. Any response
falling outside certain pre-established limits causes the driver
to be rejected. Ok, so what's the beef here? Sure, you might
argue, such a sweep signal cannot possibly measure all
performance aspects of a driver. Well, you're probably right.
What about distortion, you might ask? Yup, ain't measured. But
the problem with JBL's technique is that the reference driver
used is the first driver off the production line that day.
Travel with me now, as we recall that fateful day, 10 years ago:
ME: So you use the first driver off the line as your
standard, eh?
THEM: Yep!
ME: Why?
THEM: That way, we don't have the problem of having a
reference that ages and changes with time!
ME: Aha. Do you measure that first driver and compare
IT to anything.
THEM: (very innocently), Why, no, of course.
ME: Of course (short pause) What, uh, happens if that
first driver is a piece of unmittigated s**t?
THEM: Why, of course, then that means that, uh,
(pause), well, ... uhm (pause), see, it's not what you
think because we, uh, (pause) well, (very, very long
pause), uh... (total silence for a long time).
ME: Thank you, gentlemen, you've answered my questions
to my satisfaction.
We then proceeded to get a half-dozen drivers out of stock, each
with a different date code, and measure them. Surprise, surprise,
no two were the same, in fact, they weren't even close. As this
had been going on for several YEARS, needless to say, there was
much weeping, wailing and gnashing of teeth.
One of the other things I witnessed was the unveiling of the
prototype JBL state-of-the-art monitor loudspeaker (something
like the L-600). It had and 18 inch sub-sub-something and a 15
inch sub-hoochy and a 10 inch mid-sub dingo and a 8 inch mid-
dingy and a horn-horn here and a horn-horn there, here a ring-
radiator, there a diffusion lens, everywhere a driver. It was, I
think I can state unequivocably, the very worst loudspeaker, bar
none, I have ever heard.
I resigned my position at Harmon International the next day.
Dick Pierce