[Elecraft] UHF connectors [was: Array Solutions Lightning Arrestor]

[Alan]

On 01/23/2018 11:05 AM, Jim Brown wrote:
> On 1/23/2018 10:29 AM, Walter Underwood wrote:
...
>> But I mostly like them better because they are engineered instead of a 
>> historical accident.:-)
> 
> The technical superiority of N-connectors for use at HF is a wild 
> exaggeration, to the extent of being an urban myth. Yes, there is a 
> SMALL impedance difference at a junction, but it simply doesn't matter 
> at 6M and below, both because the difference in Zo is relatively small, 
> because the length is small as a fraction of a wavelength, and because 
> as frequency increases, small mismatches are reduced by the loss in the 
> feedline (and NOT loss due to mismatch).
> 
> There is, of course, a FICTIONAL loss called "mismatch loss," which 
> shows up in the lab with test equipment that is carefully engineered to 
> have 50 ohm output Z.  ...
Yup.  Here is a posting I made 25 years ago that has actual data:


From: ... (Alan Bloom)

Date: Fri, 18 Sep 1992 23:03:13 GMT

Subject: The Truth about UHF Connectors

Organization: Hewlett-Packard, Santa Rosa, CA

Newsgroups: rec.radio.amateur.misc



Ya gotta feel sorry for UHF connectors. Recent strings on this notes

group lambasted them as worthless at VHF and above, and barely tolerable

at HF. One poster called them "5 dB attenuators", and many agreed that

there must be some sort of conspiracy among ham equipment manufacturers

to inflict such garbage connectors on the amateur community.



Today I finally remembered to bring some UHF adapters from home so I

could do some relative measurements of UHF versus type-N.  As expected,

the type-N showed lower insertion loss at high frequencies, but the UHF

connectors were hardly "5 dB attenuators."



For the test I connected an HP8753 RF network analyzer through two short

BNC cables into the following arrangement:

    _______    ____________    ___________    ____________    _______

   |       |  | BNC female |  | N female- |  | N male to  |  |       |

__| 10 dB |__| to N male  |__| N female  |__| BNC female |__| 10 dB |__

   | Atten.|  | adapter    |  | adapter   |  | adapter    |  | Atten.|

   |_______|  |____________|  |___________|  |____________|  |_______|



Then I repeated the measurement with the N adapters replaced with UHF.

I normalized the measurements by replacing the 3 adapters with a BNC

double-female. (That is, this was assumed to have 0 dB loss.)



Since two N or UHF adapters were used, I assume the loss per connector

is half the total. The vertical scale was .1 dB/division, so I estimated

the insertion loss to the nearest .01 dB or so:



             --------- Type N --------   ---------- UHF ----------

FREQ (MHz)  TOTAL  LOSS PER CONNECTOR   TOTAL  LOSS PER CONNECTOR

1.8         0 dB   0 dB                 0 dB   0 dB

30          0      0                    0      0

100         0      0                    0      0

150         0      0                    0.02   0.01

200         0      0                    0.03   0.015

450         0      0                    0.18   0.09

600         0      0                    0.26   0.13

900         0      0                    0.66   0.33

1000        0.05   0.025                0.8    0.4

1300        0.1    0.05                 0.86   0.43

1600        0.05   0.025                0.5    0.25

2000        0.05   0.025                0.02   0.01



Insertion loss increases until about 1300 MHz, and then starts to

decrease until it is almost zero for the UHF connector at 2 GHz!  At

that frequency, the connectors are about 1/4 wave long (1 inch,

assuming .66 velocity factor), so I assume that the two adapters are

providing a conjugate match to each other. This confirms my assumption

that the insertion loss is due to reflections (impedance mismatch), not

absorption (true power loss).



Bottom line: UHF connectors work fine through the VHF range, and are not

too bad even on the 420 MHz band if you can stand about .1 dB mismatch

loss per connector.



By the way, I did not do the full 2-port calibration on the HP8753, so

there was a couple hundredth's dB ripple in the plots. I averaged this

out by eye to come up with the numbers in the above chart.



AL N1AL



P.S. Sorry, I guess I violated the Usenet rule against posting objective

data... :=)