On Sat, 7 Feb 2015 10:07:44 -0800
Tom McDermott tom.n5eg@gmail.com wrote:

The usual way is to calibrate the whole setup, including antenna, LNA,
cable and receiver. Ie. you drive to the national lab, set up your whole
system, then measure the timing difference of your GPS receiver to the
one of the lab, drive back home, and apply the correction.

Not surprising. Group delay is not considered of any importance in most
RF designs.

I would be very much interested in those papers. Could you list their titles
and authors at least?

Unlikely. LC filters are not sharp enough and difficult to build reliably
at those frequencies. I would rather assume that there are no filters
at all (beside the antenna characteristics).

		Attila Kinali

--
It is upon moral qualities that a society is ultimately founded. All
the prosperity and technological sophistication in the world is of no
use without that foundation.
-- Miss Matheson, The Diamond Age, Neil Stephenson

Attila,

On 02/08/2015 11:11 AM, Attila Kinali wrote:

I've seen a few different approaches.

Indeed.

LC filters should avoid too high Q in pass-band, but should have a bunch
of zeros a bit further out to punch out the stop-bands properly.
SAW-filters should similarly avoid high Q notches, but there it is easy
to achieve higher degree systems such that you achieve the filtering
without going to high-Q systems.

Works great unless you have a radio amateur doing L-band (23 cm)
transmissions. Here in Sweden you can transmit 1 kW in that band, just a
handfull of MHz from L2.

Cheers,
Magnus

I think that the surplus HP/Agilent GPS splitters may have an SAW
filter.  If so, measuring the delay of one of those could yield at least
an approximation.

I may have that data laying around; I'll do some digging.

John

On 02/08/2015 05:11 AM, Attila Kinali wrote:

On 2/8/15 2:11 AM, Attila Kinali wrote:

One has to be "very" careful about reading group delay specs on wideband
devices. Sometimes, the group delay (or its flatness/deviations from a
straight line) is measured ONLY over the frequency band of interest,
which might not be the filter passband.

You could have wild fluctuations of phase vs frequency somewhere, but as
long as dphase/dfreq is constant in the desired area, the
filter/amplifier meets spec.

There might be a wideband (500MHz) filter in front of the LNA, and then
separate narrow band filters for each of the three frequencies.  The
wideband filter could be LC or coupled microstripline equivalents.

Hi all.

The papers are:

1. "SAW Filter Modeling in MATLAB for GNSS Receivers", S.H. Abbas, et al.,
   IJECE Oct 2013, ISSN: 2088-2078
   The authors de-embed the group delay using FFT and MATLAB. Eyeball about
   15-20 nsec. for a pretty wide filter.

2. "The Effects of SAW Group Delay Ripple on GPS and Glonass Signals",
   Simon Adams, Novatel, Inc., Calgary AB
   The author computes a group delay ripple of 38 nsec for a specific SAW
   filter due to triple-transit reflections.

3. "GPS + Modernized GPS + Gallileo Signal Timing Biases", Chris Hegerty,
   Ed Powers, Blair Fonville, all of USNO, GPS World, March 2006 pp 49-54.
   Figure 2 shows group delay minima of about 65 nsec. for the RF/IF filtering.

4. Arbiter Systems Datasheet for AS0087800 active timing antenna, the
   datasheet is numbered PD0050600A. Arbiter systems, Paso Robles, CA. They
   specify (or measured?) the antenna delay exclusive of any cable as 43 nsec.

In reading through the various SAW filter specifications on the web, I've
found a few that specify the group delay ripple, but none that specify the
group delay itself.

-- Tom, N5EG

On Sun, Feb 8, 2015 at 2:11 AM, Attila Kinali attila@kinali.ch wrote:

OK, so I had an HP 58535A two-port GPS splitter handy and put it on the
VNA.  It clearly has a filter of some sort, as shown by the S21
frequency response.  The delay at the center of the passband is about
21ns, and it increases to about 26ns at the edges.

That delay consists of the physical length of the signal path in the
splitter, plus the effects of a 6dB amplifier, SAW filter, and a hybrid
2-way splitter.

The noise in the delay plot is because I had to avoid overdriving the
splitter amp, so the input signal was 40dB lower than normal for the
VNA.  Thus the analyzer receivers had pretty weak signals to work with
(and I didn't do anything heroic to compensate, other than use a large
averaging factor plus smoothing).

I don't know how applicable this would be to the circuit in an antenna.
I suspect the biggest difference might be due to the higher gain amp
in the antenna vs. 6dB in the splitter.

On 02/08/2015 08:29 AM, John Ackermann N8UR wrote:

On 7 Feb 2015 19:18, "Tom McDermott" tom.n5eg@gmail.com wrote:

I have never done this, but suspect that using a VNA is the best way to
go.  With a simple passive dipole on one port, and the active antenna and
cable on the second port, a transmission measurement (S21) is probably the
way to go. But I was unsure of exactly how to do it.

Your question prompted me to ask on the Keysight VNA forum. There's a reply
by Dr. Joel Dunsmore - one of the worlds leading authorities on VNAs.

http://www.keysight.com/owc_discussions/thread.jspa?threadID=39151&tstart=0

He asks what accuracy is needed. Being time-nuts, I think the answer is "as
high as possible". He writes

"With this method, 1 nsec is reasonable, but if you need 100 psec or 10
psec, then we will have to be much more careful."

You might want to contribute to that forum post.

Dave.

Hi Dave - agree that VNA is one good way to measure the delay.  If required
accuracy is less than about
0.5 nsec, then Tx antenna to Rx antenna mutual impedance starts to become
an issue. Above about
1 nsec error probably most of these can be ignored.  No access to a vector
VNA that works at 1.5 GHz.
unfortunately.

Normally I would use a difference measurement (substitute known reference
Rx antenna for Rx
under test, and difference the two), but am afraid that the (reference
Rx-to-Tx) and the (Rx under test-to-Tx)
might have different mutual Z.

Thanks for the pointer to the Keysight VNA discussion list.

-- Tom, N5EG

On Mon, Feb 16, 2015 at 6:12 PM, Dr. David Kirkby (Kirkby Microwave Ltd) <
drkirkby@kirkbymicrowave.co.uk> wrote:

On Tue, 17 Feb 2015 18:33:24 -0800
Tom McDermott tom.n5eg@gmail.com wrote:

You could try tinyVNA[1]. I have used it once, it has some quirks
(it's half hobby, half commercial project and that shows) but works
otherwise. I have no idea how accurate it is.

		Attila Kinali

[1] http://miniradiosolutions.com/minivna-tiny

--
It is upon moral qualities that a society is ultimately founded. All
the prosperity and technological sophistication in the world is of no
use without that foundation.
-- Miss Matheson, The Diamond Age, Neil Stephenson