[time-nuts] Next Generation Time/Frequency Standards May Require Provisions Preventing Vertical Displacement

jimlux jimlux at earthlink.net
Thu Sep 30 04:14:25 UTC 2010

Magnus Danielson wrote:
> For the case of telecommunication networks, the receiver will recover 
> the symbol rate of the signal in order to sample the symbols and later 
> those symbols is converted into bits. That the transmitter and receiver 
> has different gravitational potential causes a small offset in 
> frequency, but since the receiver PLL tracks the frequency then 
> frequency errors due to oscillator offsets, temperature changes, doppler 
> effects as one moves around etc.

Unless you're also using the telecommunications signal as a measurement 
tool (e.g. radio science).. To be honest, this is a curse for modern 
deep space telecom designers.  Back in the day ( a couple decades ago or 
so).. the data rates were very low (e.g. 7-10 bps) so at very low SNR, 
one needs a pretty stable oscillator to be able to effectively use a 
narrow filter.  Assuming you have that oscillator, it's not so hard to 
also use it to do precision Doppler measurements (e.g. Transit and Argos 
are examples of moderate precision nav using Doppler), so the 
"requirement" to have an oscillator with good short term stability so 
you can do nav also didn't drive cost.
Now, jump forward and we want to send and receive Megabits/second from 
deep space.  With a multiMHz wide signal, it's not like we need to know 
the carrier center frequency to a gnat's eyelash, so, in theory, as long 
as the overall phase noise is OK, we don't need good Allan deviation 
performance or even particularly good frequency accuracy.. 1 ppm out of 
32 GHz is 32 kHz, which is pretty small compared to a 1 MHz wide data 
signal.  So you'd think we could cheap out on the oscillator...  But, 
no... the navigators and radio scientists are used to getting that 
really rock solid signal for free...

But I digress...

> Doppler effects is much more important, and it's effects is being 
> treated regularly, such as when talking in the GSM phone while driving 
> the car...
Hmm.. I think crystal oscillator frequency variation in the phone is a 
bigger factor.  Let's say you're zipping down the road at 200 km/hr 
(55m/s), texting your friends.  That's about 0.2ppm Doppler or around 
400 Hz (for a 2 GHz carrier).. as noted.. the XO probably has 1ppm (at 
best.. more like 10ppm)

An even bigger problem in a mobile environment is multipath, which is 
far worse than the Doppler.. The effective propagation path distance 
could easily change 1000 meters in a fraction of a second.

More information about the time-nuts mailing list