Konstant trouble?
In the early 70s we had a problem with
a multiplexer.
Nowadays such a piece of equipment would
fit in a shoebox.
Then, it stood six feet high in a substantial
19" metal rack.
There were shelves carrying beefy power
supplies, shelves carrying rows of plug-in printed circuit boards
and a large shelf carrying special interface cards supplied by
Telefunken who made the main-frame computers to which we were
to be connected.
At the top there was a panel containing
a row of switches and dials and at the bottom a row of powerful
fans for cooling it all.
We had had about a year to design it
all, get it made in the factory and to get it working. As I recall
there were ten small processors, each built onto a printed circuit
card, and these had to intercommunicate with each other and six
main frame computers connected to their interface cards, each
of which ran at the breathtaking speed of 20MHz.
Within each processor was a program
written by members of the engineering team and set into read
only memory.
There were two versions of the program;
one handling input and one handling output.
Connected to the multiplexer were 512
ports, each designed to carry asynchronous data, at 2400 bits
per second, originating at Air Traffic Control consoles.
During testing, data was supplied by
a special test set, constructed by more engineers, capable of
generating 512 outputs running at 2400 bits per second. The test
set was plugged into the multiplexer which was also connected,
via the six main-frame computer interface cards, to another special
test set supplied by Telefunken.
All went well except every so often
an alarm lamp came on and the system stopped working.
It transpired that it usually, but not
always went wrong about 8:30am. That is to say, it always went
wrong at about 8:30am but it also went wrong at other times as
well.
After much scratching of heads and puzzling
it turned out it was the goods lift just behind the wall at the
back of the laboratory.
Some engineers turning up for work in
the morning used the goods lift rather than the stairs, and clearly
electrical noise in the mains supply affected the multiplexer.
A large isolation transformer cured
the problem: however we were still plagued by intermittent faults…
not as often and seemingly completely random.
An exasperated engineer, with good hearing,
sitting watching the flashing lamps in the small hours one night
noticed that just as the alarm lamp came on he had heard a slight
click.
For the next hour or so, in the silence
of the lab, he walked round listening intently.
Sure enough, every so often, he could
hear a click.
It didn't always result in an alarm
but maybe once in ten times it did.
By the time the rest of us turned up
(via the stairs) he had it cracked.
"It's been running since 3am he
declared".
We were amazed, as the problem had seemed
insoluble and we had looked certain to miss our Factory Acceptance
date which was now only days away.
"How did you fix it" someone
asked.
"Easy", he said "I turned
off the soldering iron".
We used Weller irons, which used an
electro-mechanical form of thermostatic control, and the click
he'd heard had been the temperature sensor operating.
For some reason, despite our isolation
transformer, the iron had generated enough interference to get
into the multiplexer and cause a program to go wrong.
A few days later, the customer's Project
Manager arrived, was given a demonstration, and expressed his
satisfaction by signing the Acceptance Certificate.
Two production multiplexers duly arrived
in the lab and the old prototype was shipped off to somewhere
in France, complete with a couple of engineers, to help with
test-bed integration.
Both new machines worked well and were
in turn sent off, this time to Southern Germany.
The two engineers came back from France,
picked up drawings etc and a clean pair of socks and soon after
departed for Konstantz.
After they'd got settled in off they
went to the Prime Contractor's factory and helped unpack and
connect up the two new multiplexers.
Now, a whole system was being run; main-frame
computers; our two multiplexers and lots of radar consoles.
"How are things going", I
asked a week later.
"Terrible" was the response,
"The multiplexers keep falling over".
"YES we're using isolation transformers,
and before you ask, NO there aren't any Weller irons around".
I went out to see them.
Sure enough, after about five minutes
the alarm lamp came on.
"Check everything" I said.
All the boards came out and spares tried.
All the wiring was checked.
All was perfect.
The main problem now was the next Acceptance
Test.
This was supposed to be a formality
but a large stage payment depended on it.
"What are we going to do?"
I said.
"Leave it to us" Jim said,
winking at me.
Next day was the BIG Acceptance Test.
Problem number one.
Where was Jim and his colleague, co-incidentally,
also a "Jim"?
The Customer's representatives were
everywhere but, fortunately, we were scheduled last in the proceedings.
Soon it was our turn.
"After a coffee", was the
consensus.
Sigh of relief; the two Jims appeared
looking rather sheepish.
It turned out they'd been celebrating
the night before and tagged onto a German couple who'd offered
them a midnight trip on their motor boat on the nearby lake.
Jim said he remembers hanging spread-eagled
over the front bit of the boat and whizzing up and down Lake
Konstantz in a semi-paralytic state.
When he woke up that morning he was
in a tent on the wrong side of the Lake with the other Jim snoring
beside him.
They had only just managed to get back.
Coffee break was now over and a small
group gathered round the multiplexers.
Somewhere consoles were fired up and
data began to flow.
About fifty feet away a large line printer
hummed into life and started clattering.
Jim leaned nonchalantly on the Multiplexer
under test.
People went hither and thither poking
at keyboards, peering at printouts and generally looking busy.
A few times the steady clattering from
the printer seemed to hesitate as if the machine was catching
its breath.
Finally, after about half an hour, silence
reigned and a heap of printouts was gathered up.
"How did it go", I asked the
young Belgian chap in charge.
He smiled, "Fine he said; maybe
there's a few queries but all went OK".
I looked at Jim.
He was still standing in the same position
he'd adopted half an hour earlier.
I looked more closely.
His hand was over the alarm lamp and
his thumb was next to the RESET button.
"You didn't", I said.
"You don't want to know",
he said.
The next morning, in the Hotel lobby,
after a memorable evening wining and dining "the Customer",
I asked about the young Belgian.
"He's not very well" was the
response.
"We're due to fly back later this
morning", I said, "and I need the certificate so I
can claim a stage payment".
"Don't worry he's sharing your
minibus to the airport you can get him to sign it on the way".
He did, and that was the first time
I'd ever seen anyone with a green face before, as the bus wended
its way up and down the mountain roads between Konstantz and
Zurich.
Later, Jim was talking to an engineer
from a French Company who were also involved in the project.
They compared notes.
THEY had just received a communication
from Signetics, a big player in integrated circuits.
"Throw away your 82S06 chips and
fit 82S106 chips", I think was the gist.
"The old ROMs are fusible link
types and we've found that some of the fused links grow back
again causing programs to be intermittent".
That was the answer!
We threw away all the rogue chips (dozens
and dozens of them!) and programmed the newer devices.
We never had any more trouble after
that.
To this day the two Multiplexers are
probably still there, not in Konstantz but at the Air Traffic
Control Centre in ……..I even got £30 for patenting
them! (see Story 32)
