On 08 Jul 2004 18:05:51 GMT, (Tdonaly) wrote:

|Wes wrote,
|Message-id:
|
|On Wed, 07 Jul 2004 18:34:15 -0400, wrote:
|
||Well here's what I did: I treated it like a peice of coax. I made a
||fixture to hold the "feedthru" with semi-rigid coax on one side and a
||50 ohm chip resistor on the other. I dropped in a short peice of coax
||and cal'ed to that. Then checked VSWR, was about 1.00something. Then I
||took out the coax, dropped in a few feedthrus and got VSWRs about 1.04
||- 1.06, which works out to about 53 - 55 ohms or so.
|| How's that sound?
|
|Like you don't know what you're doing.
|
|Your DUT is a two-port device. If you really have a network analyzer,
|why aren't you using it to measure the device as a two port?
|
|
|
|
|Is two port theory even being taught in engineering schools these days?

Beats me Tom. I'm retired, I know nuttin' 'bout engineering school.

I do know that if someone had brought me these devices to test, the
first thing I would have done would be to try and talk them out of
doing it.

I did that often. I used to see things like big LC power filters with
a solder lug on one side and two feet of shield wire coming out of the
other side with a pigtail on the far end and an attenuation spec that
went from 10 Hz to 10 GHz.

Damn fool engineers would think, well I've got an AMRAAM missile with
a 10 GHz transmitter, so I better have a filter that rejects 10 GHz.
Trying to explain that the device can't be measured to the design
engineer and the spec writer was almost a pointless exercise.

I can say that after my first retirement and subsequent rehire to
manage a components engineering group that was doing the spec writing,
we didn't do that crap anymore.

Specing and measuring feedthru filters was one of the most pointless
things we ever did. The specs all measure attenuation in a 50 ohm
system and the filters are never used in a 50 ohm system. In fact,
nobody ever knows what the source and load Z are at the test
frequencies so they don't know what they need or what they get.

Oh well, it was a pretty good living. Thank you taxpayers.

Wes wrote:

Specing and measuring feedthru filters was one of the most pointless
things we ever did. The specs all measure attenuation in a 50 ohm
system and the filters are never used in a 50 ohm system. In fact,
nobody ever knows what the source and load Z are at the test
frequencies so they don't know what they need or what they get.

Wes makes a valid point about the source and load impedances being
unknown in practical applications. The filter performance is going to be
different from the performance measured in a 50-ohm system.

But that doesn't make a 50-ohm measurement completely "pointless". In
order to select a filter, you have to make *some* attempt to
characterize the performance of the available options, under some kind
of standard conditions.

If the source and load impedances are totally unknown, a 50-ohm
environment is actually not a bad choice for a standard test, since it's
neither extremely high nor extremely low. If you have a better handle on
the practical source and load Z, then it makes sense to define a
different standard test environment - for example, the standard Line
Impedance Simulating Network (LISN) used for AC line/mains filters has a
much lower source Z.

The common-sense solution is to have standard tests, but understand
their limitations. Unfortunately - as Wes knows, all too well - military
spec writers aren't noted for either common sense or understanding.


--
73 from Ian G3SEK 'In Practice' columnist for RadCom (RSGB)
http://www.ifwtech.co.uk/g3sek