I read a compendium on frequency measurement, probably put out by the
IRE or IEEE, likely 25 years ago. I think its still in a certain pile
but I don't want to disturb that pile at the moment. Its not propagation
delay alone, its the variation in propagation delays. Think of it this
way, if you were 1000 miles from WWV, a path length change of 1 part per
million is 0.001 mile, or 52.8 feet. To hold frequency via the
ionosphere to even 1 part per million the reflection layer AND
reflection point (which can be quite diffuse) has to hold within 52.8
feet of the initial point. As I recall the article comparing HF to VLF
propagation of standard frequencies, one would be required (having a
local standard with stability say a part in 1000 million) to measure the
receive frequency every day about the same time (ignoring the days with
truly upset propagation the result of solar flares or sporadic E events)
for a month averaging the results. Then the frequency transferred would
only be precise to about 1 part per million. It is not possible to
predict the error, nor to estimate it.
N4PY claims it should average in a few minutes. Not so, for half of each
day the ionosphere tends to rise and for half a day it tends to fall,
but has a lot of variations including multiple paths that are only
observable, they are not predictable or removable. While N4PY and WWV
are not moving relative to one another, the RF propagation paths between
them are constantly in flux. Sometimes one path, sometimes multiple
paths, sometimes a single path with a bounce near one end, sometimes
multiple hops, sometimes the "optical" path bounce point has moved from
one end to the other end as the ionospheric layer responsible for
reflection (really a volumetric refraction instead of a mirror like
reflection and that refraction adds to the uncertainty) undulates under
the influence of the solar wind and solar flux.
The ionosphere refraction layer has NOTHING to hold it to a specific
position. It can vary from a few tens of miles to a couple hundred miles
of elevation and does that daily. It only takes 52.8 feet of path change
to make a part per million change.
According to one ARRL Frequency Measuring test, decades ago, I achieved
better than 1 ppm precision using the 1 KHz divisions of the dial of my
75S3B. I did do a careful calibration of the receiver using uncommon
calibration points, but I also must have made a VERY good guess getting
within 7 Hz (1/140th of a dial division)...
If one adjusts the TT TCXO to match the frequency readout of radios
without TCXO, one is probably using the wrong reference to set frequency
and those rice boxes setting the net frequency were probably never
exactly on frequency (unless they drifted past one day) and are still
drifting in the very fine precision.
Crystals do drift with age, tending to drift upward as damaged molecules
from grinding rattle loose reducing the mass of the crystal slab. The
best crystals are cut for a few MHz, 3rd or 5th overtone operation and
are operated in a carefully temperature controlled environment. My two
standards use a simple outside oven over a thermos type container with a
proportionally controlled oven for the inside oven. And they hold 1 to 5
ppb (parts per billion) per month compared to WWVB. Terrible compared to
an atomic standard, not great compared to a really good crystal
standard, but fine enough for my needs.
When doing short term frequency checks against WWVB I see random motions
of time of arrival of 10 to 50 nanoseconds. Some of that is randomness
in my local standard, some is randomness at VLF of the propagation path,
and I suspect some is wind induced motion of the suspended vertical
antenna wire at Ft. Collins. Averaged over hours or days, which is the
standard technique, traveling clocks have proved that 20 and 60 KHz
propagation is pretty good. Better than my local standards, for sure.
I find it is important to be sure to know that there are no flare
products approaching the earth, that there are no disturbances to
propagation, and that the weather is consistent between here and Ft.
Collins before I try to do a short term VLF frequency calibration. Then
I still have to ignore times when the receive signal is spread over 100
ns or more due to noise and undocumented short term path variations.
73, Jerry, K0CQ
--
Entire content copyright Dr. Gerald N. Johnson, electrical engineer.
Reproduction by permission only.
|