[TowerTalk] Connector loss measurements

[Hsu, Aaron (NBC Universal)]

A brief search on Google found this on K1TTT's website.  It's an archived
thread posted many years ago on the CQ-Contest reflector.  I know there was
also something more recent, but I haven't found it yet.

  - Aaron Hsu, NN6O

-----------------------------------

To: W8JITom at aol.com
From: k7fr at ncw.net (Gary Nieborsky)
Reply-To: k7fr at ncw.net (Gary Nieborsky)
Subject: Re: Cable Attenuation Question
CC: cq-contest at tgv.com

Tom,

35 watts = pencil soldering iron.

Your comments made me go out and dig through the College Archives.

Back in senior year at  Washington State U (W7YH, Go Cougs!)  we had to do a
measurement project in Measurements Lab.  Since there were two hams in the
Lab we decided to measure losses in coax connectors (the Prof was a ham
too).  We set up a calorimeter and measured I**2R losses from DC to 2 GHz
for a PL259/SO239 combo (did it for BNC and N too...hey it was a senior
project).

Here are some of the results from my Lab Notes:

Input power = 1,000 watts  (100V, 10A @ DC, homebrew 4-1000 .1-30 MHz,
borrowed USAF signal source 30-2,000 MHz (black box from Fairchild AFB),
Bird dummy load)

(We used a kW because neither of us had ever run more than 100 watts...power
trip!)

f (MHz) Loss (W)        dB
0.1     1       -0.00435
1       1.2     -0.00521
10      1.3     -0.00565
20      1.5     -0.00652
30      1.8     -0.00782
50      2.2     -0.00957
100     2.6     -0.01131
200     3.5     -0.01523
300     5       -0.02177
400     7       -0.03051
500     10      -0.04365
1000    15      -0.06564
1250    18      -0.07889
1500    28      -0.12334
1750    39      -0.17277**
2000    100     -0.45757**

**  Connector failed before calorimeter stabilized.

We attributed the steep upswing after 100MHz to the finish on the connector,
not the connector design.  Nickel plating seems to exhibit non-linearity
above 100MHz.  The N and BNC runs were much better.  BNC went flakey above
600MHZ (RG-58 size, RG-8 BNC went to 1000 MHz).  We were able to isolate
cable loss from connector loss by building a teflon box around the connector
body and only "viewing"  the inside of the box with the sensor.  The
Department Chair was not at all happy that this teflon box cost $750 to
build (teflon was rare in 1977).

As you can see from the table we experienced two failures.  Both were due to
the solder melting in the probe part of the connector.  The 1250 and 1500
watt runs showed discoloration but no melting.  The values for 1750 and 2000
MHz were the calculated values at the time of failure.  Each run took 1
hour, these two failed 28 and 17 minutes into the test.

We experienced a failure of an N connector at 2000MHz.   We ran the output
up in 100 watt steps until we observed a sharp up turn in losses.  We were
able to boil the water in the calorimeter at 15,000 watts and at 17,100
watts the fingers inside the connector relaxed and started arcing.

Before this experiment I was paranoid about my connectors.  Since then I
have only been concerned with the quality of the assembly and water ingress.

My take on it.......

73 Gary K7FR


At 02:25 PM 9/15/96 -0400, you wrote:
>In a message dated 96-09-14 19:24:27 EDT, you write:
>
>>loss with the formula  dB = 10(log PO/PI).  Record this value as a
reference
>>so that some day when the cable is down and about to go up again you can
see
>>if the cable is still as good as when you bought it.  (allow .1 dB for the
>>connectors at each end or .2 dB total).
>
>Hi,
>
>I keep seeing this thing about connector loss everywhere I look. It's
become
>accepted as fact in our community, but it's folklore. Here's an example
>why....
>
>With .1 dB loss per connector, power loss in a PL-259 SO-239 combo would be
>about 35 watts at 1500 watts. Loss is concentrated in the center pin and
>dielectric of the SO-239 section, in an area less than one half inch long.
>The PL-259 section has almost no loss or impedance bump when properly
>installed. 
>
>35 watts of connector heat (with a 1500 watt transmitter), when
concentrated
>in the inside of the SO-239, would quickly make the connector so hot it
would
>be untouchable. In a few minutes melt the solder connection and dielectric.
>If you doubt this, turn on a 35 watt lamp for a few minutes and touch the
>glass. Now imagine how hot the galss woud be if it was all within a quarter
>inch of the filament.
>
>At 30 MHz, the loss of a 239-259 combo is totally unimportant. One foot of
>9913 cable has much more loss.
>
>73 Tom
>
>