Re: How inaccurate is a 555 or 7555 REALLY?


John Fields wrote:
On 10 Dec 2006 15:52:54 -0800, bill.sloman@xxxxxxxx wrote:


---
Still, you could have mounted it Jap style.

Not according to the QA department. This stuff went into power stations
and oil refineries and everybody involved was rather picky.

---
And mounting the resistor vertically would have made it more likely
to fail? What a bunch of wankers you were.
---

The local theory was that the resistors could vibrate on the mounting
wire and eventually the wire could fatigue through.

Life was a little too short to try and persuade QA any different - we
had better things to do.

---
So you're a proponent of recurrent production costs in lieu of
proper training? How much money do you think you've "saved" over
the years by doing it that way?
---

Quite a lot. In particular, service engineers who are prepared to
travel all over the globe are a rare breed, and both George Kent and
Cambridge Instruments hired as many as they could find. They could not
afford to be too picky about the their trainability. The design
engineers understood that keeping the service engineers happy was part
of their job - if the service engineer got fractious, we were next in
line to spend six weeks in Seoul.

---
Yeah, right. A service engineer who couldn't understand a 3-input
RDL AND? Get off it Sloman, you're as full of *** as ever.
---

It takes time to make sense of a complex circuit diagram, and realising
a digital function with analog components doesn't speed up the process.
You are welcome to *** on your service engineers as much as you like.
We treated ours as colleagues. They weren't servicing just one machine,
but a variety of different machines, so they did have to consult the
circuit diagrams pretty frequently.

And - granting our QA department - using that particular trick wouldn't
have saved us anything anyway.

The worst I copped was a week in Nice in the middle of winter, and that
was sorting ot a mess that the marketing department had got us into,
but we were all encouraged to be careful.

Final test was a different kind of problem. They worked down the
corridor, and if they got stuck the good ones had a tendency to stick
their head around the door and ask or advice, which used up design
time. The bad ones invented their own solutions to what they perceived
as the problem, which you'd find out about when it appeared in the
"proposed modifications" file - usually some two hundred items long,
which appeared on the desk of any design engineer when he (or she) was
between projects.

If the design wasn't transparent to the final test technicians, you
always ran the risk of the machine going out the door distinctly
maladjusted,

---
That's ridiculous. There always needs to be someone responsible to
test the equipment before it goes out the door, not some bunch of
half-assed techs who are twiddling pots.
---

Grow up. That sort of work is pretty boring, and if you try and get
smart people to do it they get bored and find themselves another job.

One time I designed a circuit that used an un-trimmed - thus cheap -
Analog Devices multiplier to do a job, and wrote a procedure that let
me set up the two trimpots required in about ten minutes. I got called
into final test when the first machine went through to find that the
boss of final test and his star technician had been fiddling all day
trying to do the adjustments that I then did in the usual ten minutes.

---
Then, I suspect, your "procedure" was badly written.

It was iterative, and the final test guys didn't follow it. I could
have rewritten it, but one go around had already cost more than using
the untrimmed multiplier was ever going to save.

Plus, you
should never use pots. When I worked for Racal-Milgo we weren't
allowed to design in pots for _any_ reason. Seems you learned your
lesson the hard way. Well, at least it had a good ending since your
customers didn't wind up inheriting a problem which should never
have been there in the first place.

I've designed out a lot of pots in my time, but we knew how much they
cost, and how long final test took to set them (and what that cost). In
the application the untrimmed multiplier, two pots and a ten minute
set-up procedure should have saved money - around $4 per machine.

So you think "transparent" design involves ten minutes of fiddling
with a pair of trimpots? What a load of crap.
---

It looked perfectly transparent to me and to my boss and to his
procedure-reading side-kick(who was a very intelligent guy who was very
good at "misunderstanding" ambiguous procedure descriptions). I still
don't know what threw the technicians - it certainly wasn't worth
spending the time to find out.

Total cost (in techniican time) was about $300. I promptly modified the
circuit to use the more expensive laser trimmed multiplier, and did the
offsets on the rest of that batch (about 5 electron microscopes)
myself. We weren't expecting to sell more than about fifty of that
model before we obsoleted it with the first fully computer controlled
electron microscope, so my blunder with the cheap part cost us more
than I could have saved over the whole production run by using the
cheaper part.

---
Don't you mean the more expensive part?

No, I don't - the cheaper part was some $10 cheaper than the trmmed
multiplier. Maximum total savings was thus $500 over 50 machines, less
$100 worth of pots and $200 worth of setting-up time, if the
technicians could have set it up as fast as I did.

--
Bill Sloman, Nijmegen (but in Sydney at the moment)

.