[time-nuts] Do reflections up/down the antenna cable cause a problem with GPS?

Attila Kinali attila at kinali.ch
Mon Nov 21 18:21:28 EST 2016


Sorry... I pressed the wrong button while editing the Mail and cut short...

Continuing where I left off

> In the case of a mismatched cable, there is no “single satellite” issue. 
> Everything is impacted by the mismatch. 
> Even if the mismatch is pretty bad, the “primary” wave is the one that will 
> dominate at the receiver end. The 
> reflections will always be lower in amplitude. That effectively guarantees 
> that you don’t have a multipath 
> issue from the coax. 

As above, weaker signals can still cause quite a bit of change in the
correlation peak, but in this case it will not matter because the
reflection acts the same on all signals. I.e. the net result is a small
time offset (but no position offset). Unfortunately, there is one big
assumption in here that does not hold true: for all signals to be affected
the same way by the reflection, the receiver must be exactly linear.
But we know that many of the components in the signal path of the receiver
are distinctly non-linear functions. So there is a slight change of the
position (and thus time) due to reflections in the cable. But, as we are
usually dealing with minute differences in impedance, the reflected signals
are heavily attenuated. Assuming we have a nominally 50Ω system and have
a variation of +/-2Ω, then the worst case 48Ω vs 52Ω would give a reflection
loss of 27dB. Even assuming the back reflection has no loss (which isn't
true) would mean that the cable reflection "multipath" is almost 30dB dampend.
More likely to be something in the order of 54dB (2*27dB). Even with a 50Ω
to 75Ω impedance missmatch the reflection loss is 14dB and the multipath
should be in the order of 28dB. So waaaay below the signal and thus distortion
should be very much limited and receiver non-linearities should be negligible.

			Attila Kinali


-- 
Malek's Law:
        Any simple idea will be worded in the most complicated way.



More information about the time-nuts mailing list