[time-nuts] Alternative to GPS?

jimlux jimlux at earthlink.net
Wed Oct 6 10:19:38 EDT 2010

Neville Michie wrote:
> Hi,
> How far could you get passing time around from amateur station to station
> with a two way handshake system that establishes the instantaneous delay
> on the two way path and assumes a delay of half that value? A time relay.
> The stations would need their own short term clocks so they could keep
> their own time between contacts, and somewhere you would need heros
> with primary standards to synch the whole system.
> You would not be able to find a good position, because you would not 
> know the
> propagation mode.
> cheers, Neville Michie
> ___

You could look at the data from things like OTH-B radar (Over the 
Horizon-Backscatter) to get a feel for this.

You can also estimate position by looking at direction of arrival from 
multiple stations (HF-DF).

My gut feel is that you'd be doing real well to estimate your position 
within a few km.

OTOH, if you already know your position (e.g. you want some sort of 
time/frequency distribution that persists after the cataclysm that wiped 
out WWV and GPS), and you don't move, then position accuracy isn't a big 

Over a sufficiently long period of time (years), one could probably 
calibrate out a lot of the variability in propagation(but what would you 
calibrate against?).

I still think that if you want "infrastructure-light" calibration of a 
local standard, astronomical means are going to be easiest.  Harrison's 
H4 did about 10 ppm over a month and a half, just to put things in 
perspective. Astronomical methods were the competitor, and presumably 
are in the same accuracy range.

Then, there's the meridian/zenith crossing telescope thing used for 
astronomical time hacks.  (Can we presume that one has an almanac, even 
if WWV or GPS  are unusable?)  A standard surveying theodolite (like a 
Wild T2) can be read directly to 1 second of arc, and one could easily 
see a star cross the hairline in the 28x scope as several distinct 
events (touch hairline, centered, not touching). By hand, you could 
probably time that to less than a second, but probably not to 100 ms. 
However, I can easily conceive of a telescope and sensor that could do 
it (especially because you could track the centroid of the star image 
down to 0.1 pixel, and do some form of interpolation to get the 
"meridian crossing" to, perhaps, 10 milliseconds?

Once a day gives you 0.1 ppm, roughly.

Atmospheric seeing problems might be the limiting factor here, but they 
are random bubbles, and so you could probably do some sort of image 
processing (like astronomical blind deconvolution) to get a "reduced 
uncertainty" estimate of star position vs time.

More information about the time-nuts mailing list