I obviously can't speak for the Italian or Norwegian companies you
mention, but I do have the direct experience of 17 years of circuit
design and project engineering management at Tektronix. It's hard to
imagine experiences like yours happening with Tek equipment.

During the time I worked there, and presumably up to the present, Tek
had what they called the "phase" system. The engineers would do their
very best to design the product to meet all the advertised
specifications, plus additional non-advertised in-house specs. These
included temperature, vibration, shock, humidity and other environmental
specs; certification by various safety agencies; and EMC requirements,
in addition to detailed electrical performance specs.

When this design was complete, a meeting was held. Attendees included
representatives from design engineering, marketing, product safety,
component and evaluation engineering, and others. Only by consensus of
this group was the milestone declared to have been completed.

This was just the beginning, though, of the first phase, called "A
Phase". A number of instruments were built, typically around 25 to 50.
Some were sent to the environmental lab to test performance over the
range of specified environmental conditions. Others were shaken and
shocked. Others were studded with temperature probes and tested for
excessive temperature at many internal points. A few were put on
accelerated long-term reliability testing at a greatly elevated
temperature. Some were cycled on and off at high temperature. The design
was carefully analyzed by the evaluation engineering group, looking for
overstressed components. And many of the units were tested against the
full specification list, to insure that they fully met every spec.

During this phase, many problems were of course found and fixed. The
engineers would generate change orders describing the fixes, and the
test units were modified accordingly.

When it was believed that the units all met the many requirements,
another milestone meeting was held. Again if the attendees agreed, the
milestone was declared met, and "B Phase" began.

B Phase was largely a re-run of A Phase. Again, a sizeable number of
instruments were built and fully tested. Problems which were found were
corrected. Only at the end of this phase was production started.

Production often started with a pilot build. The first hundred or so
instruments were given extra scrutiny, temperature cycled, and otherwise
tested in a way to overstress them. These instruments normally became
demo units for the sales force, and some were retained by engineering
for internal use.

After pilot production, volume shipment finally commenced. Some large
companies required an incoming inspection test where every one of the
electrical performance specifications was checked, and the instrument
rejected if any failed. I once had the job of collecting test equipment
and writing a procedure for customers to use for testing our 50 GHz
bandwidth sampling head to specification, and it was very difficult to
find equipment capable of verifying the performance. We weren't able to
claim 60 GHz bandwidth even though we were pretty sure our units would
do it, because we couldn't find a way for us or the customer to verify
it at the time.

After the units were in production, each field service center kept
records of repairs, and which components failed. They were sorted by
circuit number (e.g. Q123) and part number, as well as by instrument and
board. If any part showed a high failure rate, the design was modified
and future instruments were built using the new design.

I know that other quality manufacturers have similar development systems.

That's why a Tektronix instrument costs a lot more than some others. The
existence of companies putting out the shoddy sort of stuff you mention
shows that some people are willing to trade quality for price. That's
their choice.

But the environment I described is the one I, and Wes, are accustomed
to, and it's what our designs had to get through.

Roy Lewallen, W7EL
(formerly Principal Engineer, Tektronix)

J M Noeding wrote:
On Sat, 16 Oct 2004 15:17:47 -0700, Roy Lewallen
wrote:

When amateur constructors are mentioned, it is not only those who do
strange things. While many large telecommunication and instrument
factories like HP, Tektronic, Siemens, Wandel&Goltermann,
Rohde&Schwartz, LME, Philips, Telettra seem to have certain rules to
follow and you may even see certain ways the different factory solves
the problems, it is some very large companies in Norway, Great
Brittain and elsewhere who make rather strange solutions.

One Italian company forgot to put transient protection over a relay,
and the driver transistor was damaged ever so often. I've maintained
many different transmitters which were almost impossible to tune up
after replacing parts because the impedances changed a lot, adding a
resistor in the base circuit improved on this. A wellknown Norwegian
radiolink manufacturer designed local oscillators in 6-8GHz using
2N3866 with over 1.5W power consumption, a buffer with the same and
operated in class C, the next doubler to 200MHz in class C and a
2N3866 as well, and a 2N3375 in class C. The first and third
transistors were critical and had to be replaced every two years, and
the signal on 6cm was so noisy that SM6ESG couldn't find any beat
note. He modified the stages to class A, reduced the drive level on
all stages and the heat was considerable lower, and at least the
oscillator noise very much improved

So, one shouldn't only blame the amateurs for bad constructors, but
sometime the manufacturers may even be worse

73,
Jan-Martin
---
J. M. Noeding, LA8AK, N-4623 Kristiansand
http://home.online.no/~la8ak/c.htm