TenTec
[Top] [All Lists]

[TenTec] Are Specs Guaranteed?

To: <tentec@contesting.com>
Subject: [TenTec] Are Specs Guaranteed?
From: N4NT@chartertn.net (Mike Hyder -N4NT-)
Date: Sun, 7 Apr 2002 23:32:17 -0400
You have 30 days to return a Ten-Tec.  Test away.

Mike

----- Original Message -----
From: "LORONA,AL (A-USA,ex3)" <al_lorona@agilent.com>
To: "'Duane Grotophorst'" <n9dg@yahoo.com>
Cc: <tentec@contesting.com>
Sent: Sunday, April 07, 2002 11:09 PM
Subject: [TenTec] Are Specs Guaranteed?


|
| You know, Duane brings up something I have always wondered about.
|
| An example: When a manufacturer says that their radio has 100 dB of
dynamic
| range, is that a guaranteed figure?
|
| Or is it an average number? Maybe a typical number?
|
| If it's guaranteed, I could conceivably measure the dynamic range
myself,
| and if it doesn't meet the spec, I could return the radio for my money
back.
| But I doubt if you could. I have not heard of a policy where you could
| return a radio simply because it didn't meet some spec. That leads me
to
| believe that the numbers you see on a spec sheet or color brochure are
| either typical or average values. If they are typical or average
values,
| there is a Gaussian distribution of that spec in a group of many
radios of
| that model. So the next logical question is, what is the sigma or
width of
| the Gaussian curve? If sigma is large, meaning the variation around
the mean
| is large, then you could get lucky and get a radio with much better
than 100
| dB of dynamic range, or get stuck with one with significantly less
than 100
| dB. So the one unit that ARRL tested... was it an average unit, a
golden
| unit, or a dog?
|
| Once the product review is published, it becomes ham gospel. For ever
and
| ever, hams will point to the review and say, "See, that radio is
junk," or,
| "Hey, pretty darn good radio!", and the differences could largely be
due to
| nothing more than the statistical variation between units.
|
|
| >
| >
| > One of my other major reservations of the ARRL tests
| > is that they typically use a single test specimen,
| > running tests on 5 or so units will provide a better
| > overall picture of a particular models traits.
| >
| > Duane
| > N9DG>
|


<Prev in Thread] Current Thread [Next in Thread>