[time-nuts] To use or not to use transmission line splitters for GPS receivers

Magnus Danielson magnus at rubidium.dyndns.org
Tue Oct 9 20:55:58 UTC 2012


On 10/09/2012 09:27 PM, John Ackermann N8UR wrote:
> Here's a link to a USNO paper that measured the tempco of three GPS
> amplifiers: http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA490830
>
> They found that amplifier filtering was the prime cause of tempco, and
> the narrowest bandpass amplifier they looked at had a group delay range
> of 4 nanoseconds over the range of -15 to +45 degrees C.

This is a good paper. I've read it before. It presents three strategies 
for GPS amplifiers:

1) Wide-band amplifier, represented by the AOA Wideband amplifier
2) Narrow-band amplifier with peaks, represented by the AOA narrow band 
amplifier
3) Narrow-band amplifier with no peaks, represented by the KW microwave 
phase-stable narrow band amplifier.

The wide-band amplifier has around 4 ns group delay, and it is fairly 
flat and stable. Since there isn't much delay to start with, it doesn't 
change a whole lot either. Since the amplifier isn't very flat, it also 
has some variations in group delay. It's fairly natural. The downside is 
that it has no suppression of interference, so we should do some damping.

The second case tries to achieve just that, but in order to create steep 
slopes around the pass-band, they have used two resonances, one on each 
side of the pass-band. You see the peaking effect on the gain curve of 
figure 1, but oh... they show up clearly in the group delay measurement 
of figure 2 too. This is expected from the theory, as these two 
pole-pairs has fairly high Q, their group delay will show this property 
in the direct vicinity of their respective resonances, just as their 
contribution to gain will do. So, nice steep slopes and good 
suppression, but lots of group delay, and by that higher sensitivity to 
environmental effects, i.e. temperature.

The third example shows wider but much flatter amplitude response, and 
essentially flat group delay. This is what you expect from maximum flat 
group delay filters such as Bessel/Thompson. No wonders those are 
specified as measuring filters for digital transmission. Lesser delay, 
and lesser sensitivity. The downside is that the cost of steep slopes 
comes from a higher number of needed poles/zeros.

Just as I expect from traditional signal theory.
Again, you get what you pay for.

Now you know why I want a network analyzer reaching this area at home.

Cheers,
Magnus



More information about the time-nuts mailing list