[Elecraft] Installed K3EXREF

[Paul Christensen]

I installed it on Friday in about 30 minutes, but I'm comfortable with my 
way around the K3.  I would think that most users could install it in about 
an hour or less.  Installation is actually very simple with very few parts 
involved.  K3EXREF came with an SMA connector and SMA to BNC between-series 
adapter.

Performance is exactly as promised by Elecraft.  Although K3EXREF will not 
phase lock the K3 to the external frequency source, I'm finding that it 
stays within 1 Hz on 20m when locked to CHU at 14.670 MHz.   I'm also 
finding that testing for zero-beat to WWV at 10 MHz is not a good idea in my 
shack, and perhaps yours too.  There are many 10 MHz signals being emitted 
by nearby appliances, LAN routers, security system, Ethernet boards, etc. 
That common, even frequency is everywhere in a household these days.

In much of North America, CHU is an excellent source for testing since it 
keeps cesium accuracy like WWV but the odd frequency of CHU is a real 
benefit since the chance for beating against extraneous signals is much 
smaller than at 10 MHz.  Testing at a higher frequency rather than say the 
MW broadcast band also provides for a reasonably good account of how the 
K3EXREF performs.   After installation, I would also suggest not trying to 
test zero beat K3EXREF against commercial MW or SW BC stations since their 
carriers are allowed to deviate much more than the cesium-based time and 
frequency standards of CHU and WWV.  Commercial AM broadcast station 
frequencies are typically only accurate to about +/- 20 Hz although most are 
better than that.

The K3's 49 MHz oscillator is only adjusted at intervals between 4 and 8 
seconds by K3EXREF in order to minimize jitter.  The K3 retains its original 
phase noise performance since the external reference is not used internal to 
the K3's frequency generating scheme.  Think of it this way -- K3EXREF is 
like lightly touching the 49 MHz oscillator with a feather every few seconds 
if necessary to keep it within 1 Hz or so.  Also, the K3 frequency stays 
accurate even at the moment of powering up the K3.  Kinda' fun to watch the 
49 MHz oscillator display change during warm-up as it's being corrected and 
compared against the external reference.

External references:  There are three primary types, and all can be 
purchased for less than USD $150 on the used market.  Cesium standards (a 
fourth type) are also available but are generally much more expensive, 
complicated, and won't offer any benefit to K3 users since K3EXREF limits 
resolution and accuracy to about 1 Hz.

The first type is the GPS Disciplined Oscillator (GPS-DO).   This type of 
standard locks itself onto the visible GPS satellites orbiting the sky. 
Requires a GPS antenna and is self-calibrating to the GPS satellites.  The 
GPS-DO's time and frequency averages from the visible satellites, each 
satellite having it's own on-board cesium-based oscillator.  Long-term 
accuracy is usually good to at least 1 x 10^-11.  Extreme accuracy and only 
a magnitude or two behind cesium.  Look for Hewlett-Packard Z3801A, Z3816A, 
Trimble Thunderbolt, Brandywine, Symmetricon, and Datum models.  Short list, 
but many others available.

Rubidium is the next type.  No GPS tracking needed.  Accuracy commensurate 
with GPS-DO.  No antenna required.  Long-term performance is highly 
dependent on the quality of the optical beam from the rubidium pump lamp. 
Look for EF Efratom, Datum,

Third, is a high quality oven-controlled crystal oscillator (OCXO). 
Requires periodic calibration.  Accuracy good to at least +/- 0.05 ppm and 
offer the best phase noise performance among the three types of oscillators, 
but as noted earlier, the excellent phase noise attributes of the OCXO will 
not be carried over to the K3.

A shack reference oscillator is good for use with other equipment too (e.g., 
frequency counter).   Although a bit lossy, I currectly use a Mini-Circuits 
passive three-way splitter, but one can purchase a distribution amp to feed 
many pieces of equipment from a single oscillator.  Unlike the splitter, a 
DA offers unity gain and very high port isolation.

Finally, be mindful that most GPS-DO and rubidium units use internal 
switch-mode DC-DC converters.  Switching noise was bad enough on two of my 
units that I decided to gut the converters and used an outboard 
triple-output linear supply.  The Trimble Thunderbolts that have been 
retired from the cellular industry require an external triple-output supply. 
If choosing this model, get a linear type and avoid the headaches.

Paul, W9AC