[[email protected]]

On Thursday, March 16, 2017 at 6:21:52 PM UTC-4, Peter Wieck wrote:
On Thursday, March 16, 2017 at 12:48:11 PM UTC-4, ScottW wrote:
ould I dispute that if it were the same thing to all people.
=20
People are different...that's a given. Not sure why that refutes the =
position=20
that any difference a person (any person) can repeatably detect is easi=
ly=20
measurable.
The trick is determining what parameters or combinations thereof is tri=
ggering=20
your detection.
=20
I would like to be able to predict that. Here is one example, however:
=20
I happen to have a serious weakness for Dynaco products, tube and solid-s=
tate. I have rebuilt enough ST-120s in my time such that I do not even need=
a schematic and I can pretty much go directly to the problem based on the =
symptoms. Early versions using 2N3055s for drivers sound much like glass-in=
-a-blender despite "good specs."=20

What are "good specs"? Well, they are most assuredly NOT good
measurements. "Specs", as published by manufacturers, are more
marketing and sales tools, somewhat less are they legally binding
agreed minimum requirements to meet the definition of "fitness=20
for purpose" and almost NEVER are they reliable indicators of=20
actual performance.

Specifications are not measurements.

If this is not clear, let me say it differently:

SPECIFICATIONS ARE NOT MEASUREMENTS.

And the 2N3055 example you give is a classic example of this. I,=20
too, have come across 2N3055s that "spec'ed" the same but worked=20
VERY differently in situ. And taking them back out of the circuit
and actually measuring them, they measure VERY differently. Not an
insignificant number of them didn't even met "spec." Those that
did fell within the manufacturer's stated statistical variance of
their target specs, that that variance is wide for many of the
device's many parameters.

Another example: analog high-order filters. Try and find capacitors
with the kinds of manufacturing tolerances needed to implement a high-
order filter: you need tolerances of perhaps 1% or better. Now, go=20
shopping for them. You're not going to find them, not for any price
that would fit in the budget of a piece of electronics intended to=20
be sold to all but the most anally retentive consumer. The capacitors
that were sued, most assuredly "met specs", but the variance in
actual performance among, say, a dozen filters implemented with 5%=20
and 10% capacitors could be enormous.

If one equates "specs" with "measurements", that's simply a piece=20
of folly.

If someone comes to me and shows me two different pieces of equipment
that sound different but with the same specs, I will, politely as
I am able, send that person back telling them they ain't even started=20
their homework yet.

And the "specs as measurements" folly is one I have pointed out more
times than I can count.

Dick Pierce