At 10:45 8/27/99 -0700, you wrote:
>I don't dispute that there must of necessity be delay between arrival time
>of the input signal and the display update, which must always be playing
>catch up.

That wasn't my point; I was addressing your next question just below.  But
keep on reading. :)

>  What I'm interested in is whether the input signal is constantly
>monitored, or read, in real time, so that a sudden glitch like a 0 V. signal
>lasting, say 300 MS, will be captured and held for later retrieval with the
>unit in Min-Max recording mode.

Just for context, 300 ms is a *long* time in the terms I'm used to thinking
of.  It may or may not be a long time in terms of a flaky connection.
NOMINAL RESPONSE         ACCURACY

>100ms to 80%

Ok, this says that if a glitch lasts one tenth of a second, your reading
will be within twenty per cent of what it would be if the glitch lasted at
least two tenths of a second, and it will be of lower magnitude than the
true reading.  For context, the spec for a VU meter (like the recording
meters on older tape decks, but very tightly specified and expensive) is
90% of reading in 100 ms, and no overshoot regardless.  That takes a
powerful movement to fling the needle around, which is why they're expensive.

>                     Specified accuracy +/- 12 digits for
>changes > 200ms in duration

If I'm reading this right it says that if it lasts more than two tenths but
less than one second, the accuracy will be within twelve digits plus or
minus of the actual value, in addition to the basic accuracy specified for
a steady reading.  So if your basic spec is +/- 0.5% +/- one digit, the
min/max spec would be +/- 0.5% +/- thirteen digits.  Note that on a 3 1/2
digit meter you have either 300 or 400 digits to play with, so if the
observed reading is 275 then the actual value could be btw 260 and 290.  On
the other hand if the reading is 5 is could fall btw minus 7 and plus
18.  Obviously that's a lousy percentage, but in practice it ought to be
better than that.  In any case if you're looking for dropouts you don't
*care* what the value is, only that it happened.

>1 Sec                                  Same as specified accuracy for
>changes > 200ms in duration

Well, here I'm confused.  I'd expect the one-second reading to be better
than the 200 ms one, but maybe not.  [Paragraph of electronic yak deleted
out of pity, I'll reinsert it if you care.]

>It is apparent you have a deeper working knowledge of these things than I
>do

If you can't dazzle them with science, baffle them with...uh...uh...I
forget. :)

>  - do these specs support what I'm thinking?

Yes.

david

David Beierl - Providence, RI
'84 Westy "Dutiful Passage"
'85 GL "Poor Relation"