FE-.5680A trimming resolution
On 2/2/12 1:28 AM, Attila Kinali wrote:
On Wed, 01 Feb 2012 21:53:16 -0800
Jim Luxjimlux@earthlink.net wrote:
You don't need the ADC: you just need a limiter/comparator.
Yes, but this degrades sensitivity quite a lot.
You don't need insane sampling rates. Think in terms of subharmonic
sampling.
This requires that you have an ADC that has the anlog bandwidth of
the signal. And ADCs with a analog BW in the GHz range are damn expensive
and hard to get.
Also a problem is to get the sampling frequency right if you want to
sample more than one band. Downmixing solves both of these problems
at the cost of higher complexity and a bit more noise.
Yes, if you need lots o'bits, but a single bit sampler with wide
bandwidth is easy (which is why they do it). It's basically a D-latch
at the end of the amplifier/limiter chain.
There is a sampling rate around 38-39 MHz that works out nicely for all
three bands (actually, any rate in that range probably works..I haven't
looked).. It helps that the 3 GPS frequencies are related to a common
base. A few minutes work with an Excel spreadsheet trying frequencies
will probably find something that works: You want the carrier to alias
to about a quarter of the sample rate (so the entire signal is in the
sample bandwidth without aliasing), but not exactly in the center
(because having some known frequency offset means your Doppler tracking
doesn't have to go through zero)
40MHz gives you a sample bandwidth of 20 MHz, so you could probably
sample slower, but I think having more samples/chip makes the tracking
easier (if nothing else, oversampling is like having more bits in your ADC)
Is there a publically-available antenna design that's easy to
fabricate, has a stable phase center, covers 1100--1600 MHz, and has a
good pattern over this band with low cross-polarization? Even a large
choke-ring design would be okay if it's fully specified.
I think there are some crossed dipole designs around. What about quad
helix?
Crossed dipole are narrow band and not easy to build as dual band designs
at least at those frequencies. Quad helix needs quite a precision to get
the right frequency and dual band designs (stacked helixes) get even more
difficult.
I suspect that you're right.. the actual antenna may be simple, the
design is hard. The antennas we use for multiband look like a crossed
dipole on the surface of a hemisphere, but the actual elements are a
very odd shape: generally a wide strip, but there are some lumps and
bumps in the outline.
I'm going to guess that they were designed with some FEM code, and then
iterated by hand. If you knew the shapes, it would be pretty easy to
build, though: copper foil tape on an appropriate substrate. As you
note, precision is important.
I'd go hunting through patents assigned to Dorne & Margolin. (part of
EDO, these days, I think). Or even maybe looking at their datasheets.
There's also what they call the "helibowl" antenna which is some form of
helix in a bowl shaped reflector/ground plane. googling that might turn
up something.
On Thu, 02 Feb 2012 07:49:53 -0800
Jim Lux jimlux@earthlink.net wrote:
[Limiting / Downmixing converter]
Yes, if you need lots o'bits, but a single bit sampler with wide
bandwidth is easy (which is why they do it). It's basically a D-latch
at the end of the amplifier/limiter chain.
Yes, but you lose IIRC about 3dB of performance compared to a 2bit ADC.
There is a sampling rate around 38-39 MHz that works out nicely for all
three bands (actually, any rate in that range probably works..I haven't
looked).. It helps that the 3 GPS frequencies are related to a common
base.
Only if you sample them seperately. Which requires seperate, sharp
filters for all of them. Also something that isn't that easy to do.
Also do not forget that Galileo E1 signals have about a 20MHz Bandwidth.
The combined E5 frequencies have about 50MHz. I think i've read somewhere
that you can get away with 8MHz for the E1 signal. Don't know how
the E5 behaves if you limit its bandwith.
40MHz gives you a sample bandwidth of 20 MHz, so you could probably
sample slower, but I think having more samples/chip makes the tracking
easier (if nothing else, oversampling is like having more bits in your ADC)
Yes.
[Antennas]
Crossed dipole are narrow band and not easy to build as dual band designs
at least at those frequencies. Quad helix needs quite a precision to get
the right frequency and dual band designs (stacked helixes) get even more
difficult.
I suspect that you're right.. the actual antenna may be simple, the
design is hard. The antennas we use for multiband look like a crossed
dipole on the surface of a hemisphere, but the actual elements are a
very odd shape: generally a wide strip, but there are some lumps and
bumps in the outline.
I thought about combining an antenna simulator with a genetic algo
to see whether it produces any usable shapes. But i havent had time
for this yet (and it's actually way down in my priority list).
I'm going to guess that they were designed with some FEM code, and then
iterated by hand. If you knew the shapes, it would be pretty easy to
build, though: copper foil tape on an appropriate substrate. As you
note, precision is important.
That's why i said that probably a patch antenna build out of PCBs
is the best solution. You can get the copper sheet at 0.1mm precision
which would define frequency and polarity properties quite well.
The only thing that would have to be done by hand would be the distance
from the ground plate. I guestimate that this value is not as critical
and that 0.5mm variation should be ok.
I'd go hunting through patents assigned to Dorne & Margolin. (part of
EDO, these days, I think). Or even maybe looking at their datasheets.
There's also what they call the "helibowl" antenna which is some form of
helix in a bowl shaped reflector/ground plane. googling that might turn
up something.
From my understanding of antenna theory (which is very little),
these are mostly variations on the directivity characteristic
(ie to get a more favorable distribution), but do not change
much the frequency characteristics. Ie if you don't have the
frequency characteristics right with a straight design, there
wont be much chance to get them right with a "shaped" design.
Attila Kinali
--
Why does it take years to find the answers to
the questions one should have asked long ago?
On 2/2/12 9:39 AM, Attila Kinali wrote:
On Thu, 02 Feb 2012 07:49:53 -0800
Jim Luxjimlux@earthlink.net wrote:
[
There is a sampling rate around 38-39 MHz that works out nicely for all
three bands (actually, any rate in that range probably works..I haven't
looked).. It helps that the 3 GPS frequencies are related to a common
base.
Only if you sample them seperately. Which requires seperate, sharp
filters for all of them. Also something that isn't that easy to do.
The filters don't have to be all that sharp. What you typically do is a
chain of amp/filter/amp/filter/amp/filter, etc, for about 6 stages.
I'll ask around about the filters, but I suspect they're a pretty
standard ceramic thing (it's a bit high frequency to be a SAW), and
since GPS frequencies are "standard" it's likely to be a "catalog part".
Also do not forget that Galileo E1 signals have about a 20MHz Bandwidth.
The combined E5 frequencies have about 50MHz. I think i've read somewhere
that you can get away with 8MHz for the E1 signal. Don't know how
the E5 behaves if you limit its bandwith.
yes, that might be tricky
[Antennas]
That's why i said that probably a patch antenna build out of PCBs
is the best solution. You can get the copper sheet at 0.1mm precision
which would define frequency and polarity properties quite well.
The only thing that would have to be done by hand would be the distance
from the ground plate. I guestimate that this value is not as critical
and that 0.5mm variation should be ok.
I've seen dual band patches that were pretty simple. One was air
dielectric, so the interplate spacing was set mostly by the spacers.
I'd go hunting through patents assigned to Dorne& Margolin. (part of
EDO, these days, I think). Or even maybe looking at their datasheets.
There's also what they call the "helibowl" antenna which is some form of
helix in a bowl shaped reflector/ground plane. googling that might turn
up something.
From my understanding of antenna theory (which is very little),
these are mostly variations on the directivity characteristic
(ie to get a more favorable distribution), but do not change
much the frequency characteristics. Ie if you don't have the
frequency characteristics right with a straight design, there
wont be much chance to get them right with a "shaped" design.
True in some designs.. however, in general "fat" elements have wider
bandwidth. Adding oddball protrusions and notches can flatten out a
response quite nicely.
Rob,
Can you upload it to my site?
THANKS IN ADVANCE,
Didier KO4BB
Sent from my BlackBerry Wireless thingy while I do other things...
-----Original Message-----
From: "Rob Kimberley" robkimberley@btinternet.com
Sender: time-nuts-bounces@febo.com
Date: Tue, 31 Jan 2012 21:14:59
To: 'Discussion of precise time and frequency measurement'time-nuts@febo.com
Reply-To: Discussion of precise time and frequency measurement
time-nuts@febo.com
Subject: Re: [time-nuts] FE-.5680A trimming resolution
If anyone is interested, I have just got hold of a PDF of the Technical Manual TM 5680-0211 for 5680A series Rubidiums.
Please contact me off list for a copy (1M, so too large to post on time-nuts@febo.com)
Rob Kimberley
-----Original Message-----
From: time-nuts-bounces@febo.com [mailto:time-nuts-bounces@febo.com] On Behalf Of Attila Kinali
Sent: 31 January 2012 20:47
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] FE-.5680A trimming resolution
On Tue, 31 Jan 2012 21:06:04 +0100
Javier Herrero jherrero@hvsistemas.es wrote:
El 31/01/2012 20:43, Attila Kinali escribi :
My current progress is that the uC i wanted to use does not do what
i want. Can anyone recommend a uC with 32bit timers and IEEE 1588 support?
You can have a look on these
http://www.ti.com/mcu/docs/mculuminaryfamilynode.tsp?sectionId=95&tabI
d=2597&familyId=1756 Some of them have IEEE-1588, like the LM3S9B96
This was exactly the device i intended to use.
But it doesnt really have 32bit timers. They cascade two 16bit timers to get 32bit, but then all kind of restrictions apply which make the timers unusable. And when using 16bit timers, i'll get an overflow every 800us (at 80MHz clock).
I browsed trough the internets and could find the following devices:
LPC18xx (NXP):
Look nice, but still in development or early production.
I especially like that they (will be) are available in a 200 pin QFP.
This would enable to use all the functions of the chip while still being able to solder one by hand.
K60 (Freescale): Hell of a confusing documentation. Also quite new.
Have only 16bit timers. A big issue is that they have a crypto unit on chip, which makes them export restricted. Ie the only way to buy them is from a local distributor which makes them expensive.
STM32-F2/F4 (ST): ST doesn't want to give me the documentation to those.
(website fails w/o error message)
I havent found any other chips yet. Sofar the options i see are:
- wait for the LPC18xx become available in quanities (can take a year)
- Use a LM3S9B96 together with a small FPGA to implement the counter functions.
Any hints appreciated
Attila Kinali
--
Why does it take years to find the answers to the questions one should have asked long ago?
time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
On 02/02/2012 10:13 AM, Azelio Boriani wrote:
OK, got it: no need to lock the receiver clock to birds to get stable data
(e.g. Oncore+Cs) but the clock can be locked to birds to get even better
data and obtain "for free" a reference clock (TBolt). The use of a stable
clock (not locked to birds) feeding the receiver can show the various
errors that affect the downlink having removed the clock instabilities.
Not quite right. If you lock up the clock, you do not lock to the birds,
but to GPS time or UTC as received over GPS. The observed time of the
birds would be a bad solution since you can't see a particular bird
continously unless you is in geosync orbit.
Cheers,
Magnus
And then all that can be done is lock the oscillator to a solution derived
from the birds... this can explain why a receiver can fail to correctly
lock an oscillator or give a strange PPS. That is, it is possible, in case
of errors, to have, for example, a PPS displaced by 1mS but state that the
solution is good.
On Sun, Feb 5, 2012 at 1:31 PM, Magnus Danielson <magnus@rubidium.dyndns.org
wrote:
On 02/02/2012 10:13 AM, Azelio Boriani wrote:
OK, got it: no need to lock the receiver clock to birds to get stable data
(e.g. Oncore+Cs) but the clock can be locked to birds to get even better
data and obtain "for free" a reference clock (TBolt). The use of a stable
clock (not locked to birds) feeding the receiver can show the various
errors that affect the downlink having removed the clock instabilities.
Not quite right. If you lock up the clock, you do not lock to the birds,
but to GPS time or UTC as received over GPS. The observed time of the birds
would be a bad solution since you can't see a particular bird continously
unless you is in geosync orbit.
Cheers,
Magnus
_____________**
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/**
mailman/listinfo/time-nutshttps://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.