"Richard Clark" wrote in message
...
On Mon, 4 Oct 2004 15:19:41 -0500, "John Smith"
wrote:

Here are the results computed from the data:

Freq (MHz) Impedance (Ohms)

410 46.4 + 6.0i (50 Ohm terminator)

380 9.7 - 12.5
390 3.5 - 5.7
400 5.1 + 1.3i
410 5.1 + 6.5i
415 4.0 + 10.0i
425 2.5 + 15.7i

Hi John,

How about the raw data? How about a detailed description of the NARDA
coupler? It should have a calibration plate affixed to the side of it
with freq vs. coupling marks (generally pretty close).

The Narda coupler has no calibration plate. It says model 30611, serial no.
4235. It also appears to have been made for Motorola, as their part number
appears on it. One end says "BTS" and the opposite end says "ANT". At the
BTS end, on the side, there is a connector which says -30 dB. Similarly on
the antenna end.


Raw data:

Reference Measurement
Freq A1 B1 ?1 A2 B2 ?2
415 1 0.76 180 0.82 0.535 158
410 1 0.77 180 0.905 0.57 165
400 1 0.79 180 1.08 0.695 177
390 1 0.81 180 1.02 0.72 -167
425 1 0.743 180 1.06 0.72 145
380 1 0.815 180 0.86 0.485 -151
410 1 0.749 180 0.535 0.029 118
425 1 0.695 180 1.04 0.695 143


I would note you have a considerable SWR, but this was expected going
into the test (however, in an inverse proportion, which may be
meaningful here).

73's
Richard Clark, KB7QHC

On Mon, 4 Oct 2004 19:30:19 -0500, "John Smith"
wrote:

The Narda coupler has no calibration plate. It says model 30611, serial no.
4235. It also appears to have been made for Motorola, as their part number
appears on it.

Hi John,

This is not a NARDA model number according to their own catalogue,
although I see it described as NARDA in more than one ebay auction. I
would say this is a special run for Motorola (which is probably their
contract number with NARDA).

However, all ebay auctions list this as an 960 MHz device.

Typically, directional couplers are within their nominal ratings only
over a octave range and some of those octaves from within their
catalogue are 500 MHz to 1 GHz. Others are 450 MHz to 900 MHz. Some
are listed as 0.05 GHz to 1 GHz, but the coupling is VERY MUCH
different than nominal outside the octave range (by as much as 10 - 15
dB). Generally, you don't suffer this much variation near the octave
band edges, but they do track off from their otherwise flat response.

It looks like you need to measure the coupling directly (at both
ports) across your frequencies of interest. The coupling factor is
not so important as is the ports tracking. I will discuss the raw and
finished data separately.

73's
Richard Clark, KB7QHC

On Mon, 4 Oct 2004 19:30:19 -0500, "John Smith"
wrote:
400 5.1 + 1.3i
Reference Measurement
Freq A1 B1 ?1 A2 B2 ?2
400 1 0.79 180 1.08 0.695 177

Hi John,

Well, from the two results above, and referencing my copy of Appl.
Note 77-3, Page 7, section "Measuring Rho 100 to 1000 MHz," there are
a number of issues here.

Your B1/A1 is quite off the mark (but certainly correctable, afterall,
that is the purpose of its measurement).

|Rho| = B2·A1/A2·B1 = 0.695·1/1.08·0.79 = 0.815

Casting the magnitude and angle onto a Smith Chart would suggest your
5.1 + 1.3i Ohms is close enough given your data.

The only kicker is port tracking, but I have a hunch that probably is
not an issue.

This bears further consideration. Too bad Kraus did not choose to
elaborate.

73's
Richard Clark, KB7QHC

"Richard Clark" wrote in message
...
On Mon, 4 Oct 2004 19:30:19 -0500, "John Smith"
wrote:
400 5.1 + 1.3i
Reference Measurement
Freq A1 B1 ?1 A2 B2 ?2
400 1 0.79 180 1.08 0.695 177

Hi John,

Well, from the two results above, and referencing my copy of Appl.
Note 77-3, Page 7, section "Measuring Rho 100 to 1000 MHz," there are
a number of issues here.

Your B1/A1 is quite off the mark (but certainly correctable, afterall,
that is the purpose of its measurement).

|Rho| = B2·A1/A2·B1 = 0.695·1/1.08·0.79 = 0.815

Casting the magnitude and angle onto a Smith Chart would suggest your
5.1 + 1.3i Ohms is close enough given your data.

The only kicker is port tracking, but I have a hunch that probably is
not an issue.

This bears further consideration. Too bad Kraus did not choose to
elaborate.

73's
Richard Clark, KB7QHC


I don't understand. Please enumerate the issues. What do you mean B1/A1 is
off the mark? If they are off the mark, how can the Z be close enough?

Are you saying that I have calculated the Z correctly from the data and you
think port tracking is not at fault? What further consideration?

You're right. Too bad Kraus didn't tell the feedpoint impedance. I suspect I
would not have embarked on this folly if he had. The antenna may have its
uses elsewhere, but I don't need the headaches of matching 5 Ohms at 440
MHz.

John

On Tue, 5 Oct 2004 09:10:19 -0500, "John Smith"
wrote:

I don't understand. Please enumerate the issues. What do you mean B1/A1 is
off the mark?

Hi John,

You should get 1.0 @ 180° (the definition of a short).

If they are off the mark, how can the Z be close enough?

Because what you DID measure, was used as a correction factor per:
|Rho| = B2·A1/A2·B1 = 0.695·1/1.08·0.79 = 0.815
(or you skipped that step)

Are you saying that I have calculated the Z correctly from the data and you
think port tracking is not at fault?

Well, that is really your job to confirm or deny. There is very
little I can accomplish short of that.

What further consideration?

I suppose I could visit my Engineering Library. I will be on campus
for my Nanotechnology seminar today anyway.

73's
Richard Clark, KB7QHC

"Richard Clark" wrote in message
...
On Tue, 5 Oct 2004 09:10:19 -0500, "John Smith"
wrote:

I don't understand. Please enumerate the issues. What do you mean B1/A1 is
off the mark?

Hi John,

You should get 1.0 @ 180° (the definition of a short).

Well, that's not possible when feeding a length of RG58 at 400 MHz, is it?
Remember, I said that there was about a 5 foot length of RG58 between the
directional coupler and the load. How can one get 1.0 reflected to the
coupler when the load is a short? That requires zero loss coax.


If they are off the mark, how can the Z be close enough?

Because what you DID measure, was used as a correction factor per:
|Rho| = B2·A1/A2·B1 = 0.695·1/1.08·0.79 = 0.815
(or you skipped that step)

I did that. As far as I can determine, I did it like the HP application note
said to do it.

Thanks for your comments.

John

On Tue, 5 Oct 2004 12:26:06 -0500, "John Smith"
wrote:

You should get 1.0 @ 180° (the definition of a short).

Well, that's not possible when feeding a length of RG58 at 400 MHz, is it?
Remember, I said that there was about a 5 foot length of RG58 between the
directional coupler and the load. How can one get 1.0 reflected to the
coupler when the load is a short? That requires zero loss coax.

Hi John,

As Wes suggests, butt up the load against the directional coupler
output and eliminate this arbitrary loss of the 5 foot RG58. It
should also shift the readings too (you are simply walking around the
circle of constant SWR). One question that would be obviated in this
process (but I have to ask anyway) is WHERE was this short you
applied? At the output port of the coupler, or at the end of this 5
foot RG58? (Same question applies to the calibrated load).

The other measurements that you reported in response to Wes indicate
you have tracking ports (even if they are off by 4dB). As I said, it
seemed unlikely this would be a problem and it confirms the
out-of-octave specification.

73's
Richard Clark, KB7QHC

"Richard Clark" wrote in message
news
On Tue, 5 Oct 2004 12:26:06 -0500, "John Smith"
wrote:

You should get 1.0 @ 180° (the definition of a short).

Well, that's not possible when feeding a length of RG58 at 400 MHz, is it?
Remember, I said that there was about a 5 foot length of RG58 between the
directional coupler and the load. How can one get 1.0 reflected to the
coupler when the load is a short? That requires zero loss coax.

Hi John,

As Wes suggests, butt up the load against the directional coupler
output and eliminate this arbitrary loss of the 5 foot RG58. It
should also shift the readings too (you are simply walking around the
circle of constant SWR). One question that would be obviated in this
process (but I have to ask anyway) is WHERE was this short you
applied? At the output port of the coupler, or at the end of this 5
foot RG58? (Same question applies to the calibrated load).


I used a 66 inch piece of RG58 between the directional coupler and the load.
It was at the load end of this piece of coax that I calibrated with a short
and made the load measurements.


The other measurements that you reported in response to Wes indicate
you have tracking ports (even if they are off by 4dB). As I said, it
seemed unlikely this would be a problem and it confirms the
out-of-octave specification.

73's
Richard Clark, KB7QHC