Empathy List Archives

AW

Anders Wallin

Sat, Sep 21, 2019 8:08 AM

Hi all, we are testing an uBlox F9T (RCB-F9T timing-board) for the past few
days. Small uBlox puck antenna in a not-so-great location.
I get an ADEV of 3e-8/tau long-term (see pictures or blog)

Through the UART there are lots and lots of settings to tweak with the
uBlox software - I am wondering if anyone figured out optimal settings for
stationary operation and best stability when the pulse1/pulse2 outputs are
configured for 1PPS and 10 MHz. Is it using just GPS (or multi-GNSS?), dual
or single-frequency, as the default setting?

Also, it seems that on 10MHz there is dithering (8ns granularity?) - is
there any simple way to generate a round-number out of the uBlox, and use a
(simple) multiply/divide to get dithering-free 10MHz?

we made a carrier-board for the RCB-F9T with an USB/UART conversion and
buffered dual BNC outputs for pulse1/pulse2 - this might be published
later...

Further down the road, if the F9T really can do dual-frequency
observations, and either the receiver clock 1PPS measured (TICC?) against
an external 1PPS, or the entire receiver clocked from external 10M/1PPS -
and dual-frequency RINEX generated from this - then there seems to be an
obvious opportunity for making a low-cost dual-frequency time-transfer
setup? Or am I missing something?
PPP with F9T seems to be possible:
https://gpsd.gitlab.io/gpsd/ppp-howto.html

thanks,
Anders
board https://www.u-blox.com/en/product/rcb-f9t-timing-board
antenna https://www.u-blox.com/en/product/ann-mb-series
blog http://www.anderswallin.net/2019/09/ublox-f9t-test/

[image: uBlox_oadev_tdev.png]
[image: uBlox_F9T_timeseries.png]

Hi all, we are testing an uBlox F9T (RCB-F9T timing-board) for the past few days. Small uBlox puck antenna in a not-so-great location. I get an ADEV of 3e-8/tau long-term (see pictures or blog) Through the UART there are lots and lots of settings to tweak with the uBlox software - I am wondering if anyone figured out optimal settings for stationary operation and best stability when the pulse1/pulse2 outputs are configured for 1PPS and 10 MHz. Is it using just GPS (or multi-GNSS?), dual or single-frequency, as the default setting? Also, it seems that on 10MHz there is dithering (8ns granularity?) - is there any simple way to generate a round-number out of the uBlox, and use a (simple) multiply/divide to get dithering-free 10MHz? we made a carrier-board for the RCB-F9T with an USB/UART conversion and buffered dual BNC outputs for pulse1/pulse2 - this might be published later... Further down the road, if the F9T really can do dual-frequency observations, and either the receiver clock 1PPS measured (TICC?) against an external 1PPS, or the entire receiver clocked from external 10M/1PPS - and dual-frequency RINEX generated from this - then there seems to be an obvious opportunity for making a low-cost dual-frequency time-transfer setup? Or am I missing something? PPP with F9T seems to be possible: https://gpsd.gitlab.io/gpsd/ppp-howto.html thanks, Anders board https://www.u-blox.com/en/product/rcb-f9t-timing-board antenna https://www.u-blox.com/en/product/ann-mb-series blog http://www.anderswallin.net/2019/09/ublox-f9t-test/ [image: uBlox_oadev_tdev.png] [image: uBlox_F9T_timeseries.png]

BK

Bob kb8tq

Sat, Sep 21, 2019 3:15 PM

Hi

All of the current crop of uBlox modules do the “dither stuff” when you ask them
to generate an output frequency. Indeed pretty much every device on the market
(that is not a GPSDO) does the same thing ( = free running oscillator and hop edges
to get the output).

Bob

On Sep 21, 2019, at 2:08 AM, Anders Wallin anders.e.e.wallin@gmail.com wrote:

Hi all, we are testing an uBlox F9T (RCB-F9T timing-board) for the past few
days. Small uBlox puck antenna in a not-so-great location.
I get an ADEV of 3e-8/tau long-term (see pictures or blog)

Through the UART there are lots and lots of settings to tweak with the
uBlox software - I am wondering if anyone figured out optimal settings for
stationary operation and best stability when the pulse1/pulse2 outputs are
configured for 1PPS and 10 MHz. Is it using just GPS (or multi-GNSS?), dual
or single-frequency, as the default setting?

Also, it seems that on 10MHz there is dithering (8ns granularity?) - is
there any simple way to generate a round-number out of the uBlox, and use a
(simple) multiply/divide to get dithering-free 10MHz?

we made a carrier-board for the RCB-F9T with an USB/UART conversion and
buffered dual BNC outputs for pulse1/pulse2 - this might be published
later...

Further down the road, if the F9T really can do dual-frequency
observations, and either the receiver clock 1PPS measured (TICC?) against
an external 1PPS, or the entire receiver clocked from external 10M/1PPS -
and dual-frequency RINEX generated from this - then there seems to be an
obvious opportunity for making a low-cost dual-frequency time-transfer
setup? Or am I missing something?
PPP with F9T seems to be possible:
https://gpsd.gitlab.io/gpsd/ppp-howto.html

thanks,
Anders
board https://www.u-blox.com/en/product/rcb-f9t-timing-board
antenna https://www.u-blox.com/en/product/ann-mb-series
blog http://www.anderswallin.net/2019/09/ublox-f9t-test/

[image: uBlox_oadev_tdev.png]
[image: uBlox_F9T_timeseries.png]
<uBlox_F9T_timeseries.png><uBlox_oadev_tdev.png>_______________________________________________
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Hi All of the current crop of uBlox modules do the “dither stuff” when you ask them to generate an output frequency. Indeed pretty much every device on the market (that is not a GPSDO) does the same thing ( = free running oscillator and hop edges to get the output). Bob > On Sep 21, 2019, at 2:08 AM, Anders Wallin <anders.e.e.wallin@gmail.com> wrote: > > Hi all, we are testing an uBlox F9T (RCB-F9T timing-board) for the past few > days. Small uBlox puck antenna in a not-so-great location. > I get an ADEV of 3e-8/tau long-term (see pictures or blog) > > Through the UART there are lots and lots of settings to tweak with the > uBlox software - I am wondering if anyone figured out optimal settings for > stationary operation and best stability when the pulse1/pulse2 outputs are > configured for 1PPS and 10 MHz. Is it using just GPS (or multi-GNSS?), dual > or single-frequency, as the default setting? > > Also, it seems that on 10MHz there is dithering (8ns granularity?) - is > there any simple way to generate a round-number out of the uBlox, and use a > (simple) multiply/divide to get dithering-free 10MHz? > > we made a carrier-board for the RCB-F9T with an USB/UART conversion and > buffered dual BNC outputs for pulse1/pulse2 - this might be published > later... > > Further down the road, if the F9T really can do dual-frequency > observations, and either the receiver clock 1PPS measured (TICC?) against > an external 1PPS, or the entire receiver clocked from external 10M/1PPS - > and dual-frequency RINEX generated from this - then there seems to be an > obvious opportunity for making a low-cost dual-frequency time-transfer > setup? Or am I missing something? > PPP with F9T seems to be possible: > https://gpsd.gitlab.io/gpsd/ppp-howto.html > > thanks, > Anders > board https://www.u-blox.com/en/product/rcb-f9t-timing-board > antenna https://www.u-blox.com/en/product/ann-mb-series > blog http://www.anderswallin.net/2019/09/ublox-f9t-test/ > > [image: uBlox_oadev_tdev.png] > [image: uBlox_F9T_timeseries.png] > <uBlox_F9T_timeseries.png><uBlox_oadev_tdev.png>_______________________________________________ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there.

MW

Michael Wouters

Sat, Sep 21, 2019 11:49 PM

Hello Anders,

We did some work on single-frequency time-transfer with the F9P earlier
this year which we presented at IFCS-EFTF in April.
http://www.openttp.org/downloads/Multi-GNSS_IFCS-EFTF_2019.pdf
There's a paper too, which I should upload.
In short, the F9P is very suitable for code-based time-transfer, and we
will be using it , or the F9T, in the next iteration of our time-transfer
system.

But, I don't see how you can feed an external 10 MHz and 1 pps to the F9T.
There don't appear to be any inputs for this on the chip. The TIMEMARK
inputs only seem to be useable for measurements. Have I missed something?

Regards
Michael

On Sun, 22 Sep 2019 at 12:30 am, Anders Wallin anders.e.e.wallin@gmail.com
wrote:

Hi all, we are testing an uBlox F9T (RCB-F9T timing-board) for the past few
days. Small uBlox puck antenna in a not-so-great location.
I get an ADEV of 3e-8/tau long-term (see pictures or blog)

Through the UART there are lots and lots of settings to tweak with the
uBlox software - I am wondering if anyone figured out optimal settings for
stationary operation and best stability when the pulse1/pulse2 outputs are
configured for 1PPS and 10 MHz. Is it using just GPS (or multi-GNSS?), dual
or single-frequency, as the default setting?

Also, it seems that on 10MHz there is dithering (8ns granularity?) - is
there any simple way to generate a round-number out of the uBlox, and use a
(simple) multiply/divide to get dithering-free 10MHz?

we made a carrier-board for the RCB-F9T with an USB/UART conversion and
buffered dual BNC outputs for pulse1/pulse2 - this might be published
later...

Further down the road, if the F9T really can do dual-frequency
observations, and either the receiver clock 1PPS measured (TICC?) against
an external 1PPS, or the entire receiver clocked from external 10M/1PPS -
and dual-frequency RINEX generated from this - then there seems to be an
obvious opportunity for making a low-cost dual-frequency time-transfer
setup? Or am I missing something?
PPP with F9T seems to be possible:
https://gpsd.gitlab.io/gpsd/ppp-howto.html

thanks,
Anders
board https://www.u-blox.com/en/product/rcb-f9t-timing-board
antenna https://www.u-blox.com/en/product/ann-mb-series
blog http://www.anderswallin.net/2019/09/ublox-f9t-test/

[image: uBlox_oadev_tdev.png]
[image: uBlox_F9T_timeseries.png]

time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Hello Anders, We did some work on single-frequency time-transfer with the F9P earlier this year which we presented at IFCS-EFTF in April. http://www.openttp.org/downloads/Multi-GNSS_IFCS-EFTF_2019.pdf There's a paper too, which I should upload. In short, the F9P is very suitable for code-based time-transfer, and we will be using it , or the F9T, in the next iteration of our time-transfer system. But, I don't see how you can feed an external 10 MHz and 1 pps to the F9T. There don't appear to be any inputs for this on the chip. The TIMEMARK inputs only seem to be useable for measurements. Have I missed something? Regards Michael On Sun, 22 Sep 2019 at 12:30 am, Anders Wallin <anders.e.e.wallin@gmail.com> wrote: > Hi all, we are testing an uBlox F9T (RCB-F9T timing-board) for the past few > days. Small uBlox puck antenna in a not-so-great location. > I get an ADEV of 3e-8/tau long-term (see pictures or blog) > > Through the UART there are lots and lots of settings to tweak with the > uBlox software - I am wondering if anyone figured out optimal settings for > stationary operation and best stability when the pulse1/pulse2 outputs are > configured for 1PPS and 10 MHz. Is it using just GPS (or multi-GNSS?), dual > or single-frequency, as the default setting? > > Also, it seems that on 10MHz there is dithering (8ns granularity?) - is > there any simple way to generate a round-number out of the uBlox, and use a > (simple) multiply/divide to get dithering-free 10MHz? > > we made a carrier-board for the RCB-F9T with an USB/UART conversion and > buffered dual BNC outputs for pulse1/pulse2 - this might be published > later... > > Further down the road, if the F9T really can do dual-frequency > observations, and either the receiver clock 1PPS measured (TICC?) against > an external 1PPS, or the entire receiver clocked from external 10M/1PPS - > and dual-frequency RINEX generated from this - then there seems to be an > obvious opportunity for making a low-cost dual-frequency time-transfer > setup? Or am I missing something? > PPP with F9T seems to be possible: > https://gpsd.gitlab.io/gpsd/ppp-howto.html > > thanks, > Anders > board https://www.u-blox.com/en/product/rcb-f9t-timing-board > antenna https://www.u-blox.com/en/product/ann-mb-series > blog http://www.anderswallin.net/2019/09/ublox-f9t-test/ > > [image: uBlox_oadev_tdev.png] > [image: uBlox_F9T_timeseries.png] > _______________________________________________ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe, go to > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there. >

AW

Anders Wallin

Sun, Sep 22, 2019 7:22 AM

Hi Michael, thanks for the reminder about your slides!
Your page 4 has the two setups. I think you are right in that the F9T
doesn't seem to take an external receiver clock (even if working with the
bare part, not the RCB-board).
(FWIW, AFAIK the not-so-low-cost septentrios use the left external receiver
clock scheme, while Dicom GTRs use the right - both in the 15k cost range)

Did you use an external TIC (like the TICC?) - to get lower than the 8ns
granularity of the TIMEMARK?

If I understand CGGTTS the TIC and sawtooth-correction are used to provide
one single time-offset value into the CGGTTS file, for each 780s=13 min
satellite track observed?
https://tycho.usno.navy.mil/cggtts.html
is that correct?
Can the same TIC-based sawtooth correction work for 30s RINEX data instead?
If gpsd can produce dual-frequency RINEX from the F9T, then the new
(software)part needed is the sawtooth-correction at 30s intervals applied
to the raw RINEX from the F9T.
https://gpsd.gitlab.io/gpsd/ppp-howto.html

Anders

On Sun, Sep 22, 2019 at 3:00 AM Michael Wouters michaeljwouters@gmail.com
wrote:

Hello Anders,

We did some work on single-frequency time-transfer with the F9P earlier
this year which we presented at IFCS-EFTF in April.
http://www.openttp.org/downloads/Multi-GNSS_IFCS-EFTF_2019.pdf
There's a paper too, which I should upload.
In short, the F9P is very suitable for code-based time-transfer, and we
will be using it , or the F9T, in the next iteration of our time-transfer
system.

But, I don't see how you can feed an external 10 MHz and 1 pps to the F9T.
There don't appear to be any inputs for this on the chip. The TIMEMARK
inputs only seem to be useable for measurements. Have I missed something?

Regards
Michael

Further down the road, if the F9T really can do dual-frequency
observations, and either the receiver clock 1PPS measured (TICC?) against
an external 1PPS, or the entire receiver clocked from external 10M/1PPS -
and dual-frequency RINEX generated from this - then there seems to be an
obvious opportunity for making a low-cost dual-frequency time-transfer
setup? Or am I missing something?
PPP with F9T seems to be possible:
https://gpsd.gitlab.io/gpsd/ppp-howto.html

Hi Michael, thanks for the reminder about your slides! Your page 4 has the two setups. I think you are right in that the F9T doesn't seem to take an external receiver clock (even if working with the bare part, not the RCB-board). (FWIW, AFAIK the not-so-low-cost septentrios use the left external receiver clock scheme, while Dicom GTRs use the right - both in the 15k cost range) Did you use an external TIC (like the TICC?) - to get lower than the 8ns granularity of the TIMEMARK? If I understand CGGTTS the TIC and sawtooth-correction are used to provide one single time-offset value into the CGGTTS file, for each 780s=13 min satellite track observed? https://tycho.usno.navy.mil/cggtts.html is that correct? Can the same TIC-based sawtooth correction work for 30s RINEX data instead? If gpsd can produce dual-frequency RINEX from the F9T, then the new (software)part needed is the sawtooth-correction at 30s intervals applied to the raw RINEX from the F9T. https://gpsd.gitlab.io/gpsd/ppp-howto.html Anders On Sun, Sep 22, 2019 at 3:00 AM Michael Wouters <michaeljwouters@gmail.com> wrote: > Hello Anders, > > We did some work on single-frequency time-transfer with the F9P earlier > this year which we presented at IFCS-EFTF in April. > http://www.openttp.org/downloads/Multi-GNSS_IFCS-EFTF_2019.pdf > There's a paper too, which I should upload. > In short, the F9P is very suitable for code-based time-transfer, and we > will be using it , or the F9T, in the next iteration of our time-transfer > system. > > But, I don't see how you can feed an external 10 MHz and 1 pps to the F9T. > There don't appear to be any inputs for this on the chip. The TIMEMARK > inputs only seem to be useable for measurements. Have I missed something? > > Regards > Michael > > > Further down the road, if the F9T really can do dual-frequency > > observations, and either the receiver clock 1PPS measured (TICC?) against > > an external 1PPS, or the entire receiver clocked from external 10M/1PPS - > > and dual-frequency RINEX generated from this - then there seems to be an > > obvious opportunity for making a low-cost dual-frequency time-transfer > > setup? Or am I missing something? > > PPP with F9T seems to be possible: > > https://gpsd.gitlab.io/gpsd/ppp-howto.html > >

PV

Peter Vince

Sun, Sep 22, 2019 8:54 PM

Hello Michael,

 Thank you for your Powerpoint presentation slides - if the paper you

mentioned added some descriptive words, I would love to see that.

Might I ask why you didn't use the F9P in dual-frequency mode, as I
understood that gives a much better result, being able to (almost?)
eliminate ionospheric effects?

 Regards,

      Peter

On Sun, 22 Sep 2019 at 01:00, Michael Wouters michaeljwouters@gmail.com
wrote:

We did some work on single-frequency time-transfer with the F9P earlier
this year which we presented at IFCS-EFTF in April.
http://www.openttp.org/downloads/Multi-GNSS_IFCS-EFTF_2019.pdf
There's a paper too, which I should upload.
In short, the F9P is very suitable for code-based time-transfer, and we
will be using it , or the F9T, in the next iteration of our time-transfer
system.
...
Regards
Michael

Hello Michael, Thank you for your Powerpoint presentation slides - if the paper you mentioned added some descriptive words, I would love to see that. Might I ask why you didn't use the F9P in dual-frequency mode, as I understood that gives a much better result, being able to (almost?) eliminate ionospheric effects? Regards, Peter On Sun, 22 Sep 2019 at 01:00, Michael Wouters <michaeljwouters@gmail.com> wrote: > > We did some work on single-frequency time-transfer with the F9P earlier > this year which we presented at IFCS-EFTF in April. > http://www.openttp.org/downloads/Multi-GNSS_IFCS-EFTF_2019.pdf > There's a paper too, which I should upload. > In short, the F9P is very suitable for code-based time-transfer, and we > will be using it , or the F9T, in the next iteration of our time-transfer > system. > ... > Regards > Michael

MW

Michael Wouters

Sun, Sep 22, 2019 9:54 PM

Hello Anders,
Yes, we used TAPR TICCs do the testing. It’s a very economical solution
for testing multiple receivers in parallel.

The OpenTTP software mktimetx applies sawtooth corrections to the TIC
measurements, as you guessed. The corrected TIC measurements are combined
with the raw GNSS pseudo ranges, and then this is what goes into the output
RINEX and what is used for CGGTTS generation.

One complication is that the raw pseudo ranges have ms ambiguities in them.
But mktimetx fixes that.

I’ll put the paper on the openttp website.
Unfortunately, it appears that it will otherwise be behind a paywall.

Cheers
Michael

On Mon, 23 Sep 2019 at 12:00 am, Anders Wallin anders.e.e.wallin@gmail.com
wrote:

Hi Michael, thanks for the reminder about your slides!
Your page 4 has the two setups. I think you are right in that the F9T
doesn't seem to take an external receiver clock (even if working with the
bare part, not the RCB-board).
(FWIW, AFAIK the not-so-low-cost septentrios use the left external receiver
clock scheme, while Dicom GTRs use the right - both in the 15k cost range)

Did you use an external TIC (like the TICC?) - to get lower than the 8ns
granularity of the TIMEMARK?

If I understand CGGTTS the TIC and sawtooth-correction are used to provide
one single time-offset value into the CGGTTS file, for each 780s=13 min
satellite track observed?
https://tycho.usno.navy.mil/cggtts.html
is that correct?
Can the same TIC-based sawtooth correction work for 30s RINEX data instead?
If gpsd can produce dual-frequency RINEX from the F9T, then the new
(software)part needed is the sawtooth-correction at 30s intervals applied
to the raw RINEX from the F9T.
https://gpsd.gitlab.io/gpsd/ppp-howto.html

Anders

On Sun, Sep 22, 2019 at 3:00 AM Michael Wouters <michaeljwouters@gmail.com

wrote:

Hello Anders,

We did some work on single-frequency time-transfer with the F9P earlier
this year which we presented at IFCS-EFTF in April.
http://www.openttp.org/downloads/Multi-GNSS_IFCS-EFTF_2019.pdf
There's a paper too, which I should upload.
In short, the F9P is very suitable for code-based time-transfer, and we
will be using it , or the F9T, in the next iteration of our time-transfer
system.

But, I don't see how you can feed an external 10 MHz and 1 pps to the

F9T.

There don't appear to be any inputs for this on the chip. The TIMEMARK
inputs only seem to be useable for measurements. Have I missed something?

Regards
Michael

Further down the road, if the F9T really can do dual-frequency
observations, and either the receiver clock 1PPS measured (TICC?)

against

an external 1PPS, or the entire receiver clocked from external

10M/1PPS -

and dual-frequency RINEX generated from this - then there seems to be

an

obvious opportunity for making a low-cost dual-frequency time-transfer
setup? Or am I missing something?
PPP with F9T seems to be possible:
https://gpsd.gitlab.io/gpsd/ppp-howto.html

time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Hello Anders, Yes, we used TAPR TICCs do the testing. It’s a very economical solution for testing multiple receivers in parallel. The OpenTTP software mktimetx applies sawtooth corrections to the TIC measurements, as you guessed. The corrected TIC measurements are combined with the raw GNSS pseudo ranges, and then this is what goes into the output RINEX and what is used for CGGTTS generation. One complication is that the raw pseudo ranges have ms ambiguities in them. But mktimetx fixes that. I’ll put the paper on the openttp website. Unfortunately, it appears that it will otherwise be behind a paywall. Cheers Michael On Mon, 23 Sep 2019 at 12:00 am, Anders Wallin <anders.e.e.wallin@gmail.com> wrote: > Hi Michael, thanks for the reminder about your slides! > Your page 4 has the two setups. I think you are right in that the F9T > doesn't seem to take an external receiver clock (even if working with the > bare part, not the RCB-board). > (FWIW, AFAIK the not-so-low-cost septentrios use the left external receiver > clock scheme, while Dicom GTRs use the right - both in the 15k cost range) > > Did you use an external TIC (like the TICC?) - to get lower than the 8ns > granularity of the TIMEMARK? > > If I understand CGGTTS the TIC and sawtooth-correction are used to provide > one single time-offset value into the CGGTTS file, for each 780s=13 min > satellite track observed? > https://tycho.usno.navy.mil/cggtts.html > is that correct? > Can the same TIC-based sawtooth correction work for 30s RINEX data instead? > If gpsd can produce dual-frequency RINEX from the F9T, then the new > (software)part needed is the sawtooth-correction at 30s intervals applied > to the raw RINEX from the F9T. > https://gpsd.gitlab.io/gpsd/ppp-howto.html > > Anders > > > > On Sun, Sep 22, 2019 at 3:00 AM Michael Wouters <michaeljwouters@gmail.com > > > wrote: > > > Hello Anders, > > > > We did some work on single-frequency time-transfer with the F9P earlier > > this year which we presented at IFCS-EFTF in April. > > http://www.openttp.org/downloads/Multi-GNSS_IFCS-EFTF_2019.pdf > > There's a paper too, which I should upload. > > In short, the F9P is very suitable for code-based time-transfer, and we > > will be using it , or the F9T, in the next iteration of our time-transfer > > system. > > > > But, I don't see how you can feed an external 10 MHz and 1 pps to the > F9T. > > There don't appear to be any inputs for this on the chip. The TIMEMARK > > inputs only seem to be useable for measurements. Have I missed something? > > > > Regards > > Michael > > > > > Further down the road, if the F9T really can do dual-frequency > > > observations, and either the receiver clock 1PPS measured (TICC?) > against > > > an external 1PPS, or the entire receiver clocked from external > 10M/1PPS - > > > and dual-frequency RINEX generated from this - then there seems to be > an > > > obvious opportunity for making a low-cost dual-frequency time-transfer > > > setup? Or am I missing something? > > > PPP with F9T seems to be possible: > > > https://gpsd.gitlab.io/gpsd/ppp-howto.html > > > > > _______________________________________________ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe, go to > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there. >

MW

Michael Wouters

Sun, Sep 22, 2019 10:01 PM

Hello Peter

The aim of our study was to characterise the performance of single
frequency receivers for time-transfer. The F9P was a late addition to our
study and extending our software to dual frequency was too much work at the
time. Having a measured ionosphere was not such a huge improvement anyway,
because there are other ways to get an improved ionosphere correction when
post-processing.

Regards
Michael

On Mon, 23 Sep 2019 at 7:03 am, Peter Vince petervince1952@gmail.com
wrote:

Hello Michael,

  Thank you for your Powerpoint presentation slides - if the paper you

mentioned added some descriptive words, I would love to see that.

Might I ask why you didn't use the F9P in dual-frequency mode, as I
understood that gives a much better result, being able to (almost?)
eliminate ionospheric effects?

  Regards,

       Peter

On Sun, 22 Sep 2019 at 01:00, Michael Wouters michaeljwouters@gmail.com
wrote:

We did some work on single-frequency time-transfer with the F9P earlier
this year which we presented at IFCS-EFTF in April.
http://www.openttp.org/downloads/Multi-GNSS_IFCS-EFTF_2019.pdf
There's a paper too, which I should upload.
In short, the F9P is very suitable for code-based time-transfer, and we
will be using it , or the F9T, in the next iteration of our time-transfer
system.
...
Regards
Michael

time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Hello Peter The aim of our study was to characterise the performance of single frequency receivers for time-transfer. The F9P was a late addition to our study and extending our software to dual frequency was too much work at the time. Having a measured ionosphere was not such a huge improvement anyway, because there are other ways to get an improved ionosphere correction when post-processing. Regards Michael On Mon, 23 Sep 2019 at 7:03 am, Peter Vince <petervince1952@gmail.com> wrote: > Hello Michael, > > Thank you for your Powerpoint presentation slides - if the paper you > mentioned added some descriptive words, I would love to see that. > > Might I ask why you didn't use the F9P in dual-frequency mode, as I > understood that gives a much better result, being able to (almost?) > eliminate ionospheric effects? > > Regards, > > Peter > > > On Sun, 22 Sep 2019 at 01:00, Michael Wouters <michaeljwouters@gmail.com> > wrote: > > > > We did some work on single-frequency time-transfer with the F9P earlier > > this year which we presented at IFCS-EFTF in April. > > http://www.openttp.org/downloads/Multi-GNSS_IFCS-EFTF_2019.pdf > > There's a paper too, which I should upload. > > In short, the F9P is very suitable for code-based time-transfer, and we > > will be using it , or the F9T, in the next iteration of our time-transfer > > system. > > ... > > Regards > > Michael > _______________________________________________ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe, go to > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there. >

RS

Ragnar Sundblad

Mon, Sep 23, 2019 6:11 AM

Hello Michael,

On 22 Sep 2019, at 23:54, Michael Wouters michaeljwouters@gmail.com wrote:

The OpenTTP software mktimetx applies sawtooth corrections to the TIC
measurements, as you guessed. The corrected TIC measurements are combined
with the raw GNSS pseudo ranges, and then this is what goes into the output
RINEX and what is used for CGGTTS generation.

Have you considered using the UBX-TIM-TP qErr correction (the receivers own
estimate of the quantisation error for the TIMEPULSE output) instead of
sawtooth filtering?

I have played with it a little on a u-blox M8, and it sure seemed to smooth
things out, but I haven’t made any real measurements on the result.

Anders,
Thanks for sharing your results! I am a but surprised that you have so large
variations, I wonder how much the antenna placement affects the results.
Just for fun - can you tell us more about how your antenna is place now?

Best regards,

Ragnar

Hello Michael, > On 22 Sep 2019, at 23:54, Michael Wouters <michaeljwouters@gmail.com> wrote: > > The OpenTTP software mktimetx applies sawtooth corrections to the TIC > measurements, as you guessed. The corrected TIC measurements are combined > with the raw GNSS pseudo ranges, and then this is what goes into the output > RINEX and what is used for CGGTTS generation. Have you considered using the UBX-TIM-TP qErr correction (the receivers own estimate of the quantisation error for the TIMEPULSE output) instead of sawtooth filtering? I have played with it a little on a u-blox M8, and it sure seemed to smooth things out, but I haven’t made any real measurements on the result. Anders, Thanks for sharing your results! I am a but surprised that you have so large variations, I wonder how much the antenna placement affects the results. Just for fun - can you tell us more about how your antenna is place now? Best regards, Ragnar

MW

Michael Wouters

Mon, Sep 23, 2019 7:36 AM

Hello Ragnar

Mktimetx uses the receiver-provided sawtooth correction in the TP message.
The TDEV plots in the presentation I linked show its effectiveness.

Cheers
Michael

On Mon, 23 Sep 2019 at 4:21 pm, Ragnar Sundblad ragge@kth.se wrote:

Hello Michael,

On 22 Sep 2019, at 23:54, Michael Wouters michaeljwouters@gmail.com

wrote:

The OpenTTP software mktimetx applies sawtooth corrections to the TIC
measurements, as you guessed. The corrected TIC measurements are combined
with the raw GNSS pseudo ranges, and then this is what goes into the

output

RINEX and what is used for CGGTTS generation.

Have you considered using the UBX-TIM-TP qErr correction (the receivers own
estimate of the quantisation error for the TIMEPULSE output) instead of
sawtooth filtering?

I have played with it a little on a u-blox M8, and it sure seemed to smooth
things out, but I haven’t made any real measurements on the result.

Anders,
Thanks for sharing your results! I am a but surprised that you have so
large
variations, I wonder how much the antenna placement affects the results.
Just for fun - can you tell us more about how your antenna is place now?

Best regards,

Ragnar

time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Hello Ragnar Mktimetx uses the receiver-provided sawtooth correction in the TP message. The TDEV plots in the presentation I linked show its effectiveness. Cheers Michael On Mon, 23 Sep 2019 at 4:21 pm, Ragnar Sundblad <ragge@kth.se> wrote: > > Hello Michael, > > > On 22 Sep 2019, at 23:54, Michael Wouters <michaeljwouters@gmail.com> > wrote: > > > > The OpenTTP software mktimetx applies sawtooth corrections to the TIC > > measurements, as you guessed. The corrected TIC measurements are combined > > with the raw GNSS pseudo ranges, and then this is what goes into the > output > > RINEX and what is used for CGGTTS generation. > > Have you considered using the UBX-TIM-TP qErr correction (the receivers own > estimate of the quantisation error for the TIMEPULSE output) instead of > sawtooth filtering? > > I have played with it a little on a u-blox M8, and it sure seemed to smooth > things out, but I haven’t made any real measurements on the result. > > > Anders, > Thanks for sharing your results! I am a but surprised that you have so > large > variations, I wonder how much the antenna placement affects the results. > Just for fun - can you tell us more about how your antenna is place now? > > > Best regards, > > Ragnar > > > _______________________________________________ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe, go to > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there. >

MW

Michael Wouters

Mon, Sep 23, 2019 8:41 AM

I have uploaded the paper that appears in the conference proceedings:
http://www.openttp.org/downloads/Multi-GNSS_IFCS-EFTF_2019_paper.pdf

Cheers
Michael

On Sun, Sep 22, 2019 at 9:49 AM Michael Wouters
michaeljwouters@gmail.com wrote:

Hello Anders,

We did some work on single-frequency time-transfer with the F9P earlier this year which we presented at IFCS-EFTF in April.
http://www.openttp.org/downloads/Multi-GNSS_IFCS-EFTF_2019.pdf
There's a paper too, which I should upload.
In short, the F9P is very suitable for code-based time-transfer, and we will be using it , or the F9T, in the next iteration of our time-transfer system.

But, I don't see how you can feed an external 10 MHz and 1 pps to the F9T. There don't appear to be any inputs for this on the chip. The TIMEMARK inputs only seem to be useable for measurements. Have I missed something?

Regards
Michael

On Sun, 22 Sep 2019 at 12:30 am, Anders Wallin anders.e.e.wallin@gmail.com wrote:

Hi all, we are testing an uBlox F9T (RCB-F9T timing-board) for the past few
days. Small uBlox puck antenna in a not-so-great location.
I get an ADEV of 3e-8/tau long-term (see pictures or blog)

Through the UART there are lots and lots of settings to tweak with the
uBlox software - I am wondering if anyone figured out optimal settings for
stationary operation and best stability when the pulse1/pulse2 outputs are
configured for 1PPS and 10 MHz. Is it using just GPS (or multi-GNSS?), dual
or single-frequency, as the default setting?

Also, it seems that on 10MHz there is dithering (8ns granularity?) - is
there any simple way to generate a round-number out of the uBlox, and use a
(simple) multiply/divide to get dithering-free 10MHz?

we made a carrier-board for the RCB-F9T with an USB/UART conversion and
buffered dual BNC outputs for pulse1/pulse2 - this might be published
later...

Further down the road, if the F9T really can do dual-frequency
observations, and either the receiver clock 1PPS measured (TICC?) against
an external 1PPS, or the entire receiver clocked from external 10M/1PPS -
and dual-frequency RINEX generated from this - then there seems to be an
obvious opportunity for making a low-cost dual-frequency time-transfer
setup? Or am I missing something?
PPP with F9T seems to be possible:
https://gpsd.gitlab.io/gpsd/ppp-howto.html

thanks,
Anders
board https://www.u-blox.com/en/product/rcb-f9t-timing-board
antenna https://www.u-blox.com/en/product/ann-mb-series
blog http://www.anderswallin.net/2019/09/ublox-f9t-test/

[image: uBlox_oadev_tdev.png]
[image: uBlox_F9T_timeseries.png]

time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

I have uploaded the paper that appears in the conference proceedings: http://www.openttp.org/downloads/Multi-GNSS_IFCS-EFTF_2019_paper.pdf Cheers Michael On Sun, Sep 22, 2019 at 9:49 AM Michael Wouters <michaeljwouters@gmail.com> wrote: > > Hello Anders, > > We did some work on single-frequency time-transfer with the F9P earlier this year which we presented at IFCS-EFTF in April. > http://www.openttp.org/downloads/Multi-GNSS_IFCS-EFTF_2019.pdf > There's a paper too, which I should upload. > In short, the F9P is very suitable for code-based time-transfer, and we will be using it , or the F9T, in the next iteration of our time-transfer system. > > But, I don't see how you can feed an external 10 MHz and 1 pps to the F9T. There don't appear to be any inputs for this on the chip. The TIMEMARK inputs only seem to be useable for measurements. Have I missed something? > > Regards > Michael > > On Sun, 22 Sep 2019 at 12:30 am, Anders Wallin <anders.e.e.wallin@gmail.com> wrote: >> >> Hi all, we are testing an uBlox F9T (RCB-F9T timing-board) for the past few >> days. Small uBlox puck antenna in a not-so-great location. >> I get an ADEV of 3e-8/tau long-term (see pictures or blog) >> >> Through the UART there are lots and lots of settings to tweak with the >> uBlox software - I am wondering if anyone figured out optimal settings for >> stationary operation and best stability when the pulse1/pulse2 outputs are >> configured for 1PPS and 10 MHz. Is it using just GPS (or multi-GNSS?), dual >> or single-frequency, as the default setting? >> >> Also, it seems that on 10MHz there is dithering (8ns granularity?) - is >> there any simple way to generate a round-number out of the uBlox, and use a >> (simple) multiply/divide to get dithering-free 10MHz? >> >> we made a carrier-board for the RCB-F9T with an USB/UART conversion and >> buffered dual BNC outputs for pulse1/pulse2 - this might be published >> later... >> >> Further down the road, if the F9T really can do dual-frequency >> observations, and either the receiver clock 1PPS measured (TICC?) against >> an external 1PPS, or the entire receiver clocked from external 10M/1PPS - >> and dual-frequency RINEX generated from this - then there seems to be an >> obvious opportunity for making a low-cost dual-frequency time-transfer >> setup? Or am I missing something? >> PPP with F9T seems to be possible: >> https://gpsd.gitlab.io/gpsd/ppp-howto.html >> >> thanks, >> Anders >> board https://www.u-blox.com/en/product/rcb-f9t-timing-board >> antenna https://www.u-blox.com/en/product/ann-mb-series >> blog http://www.anderswallin.net/2019/09/ublox-f9t-test/ >> >> [image: uBlox_oadev_tdev.png] >> [image: uBlox_F9T_timeseries.png] >> _______________________________________________ >> time-nuts mailing list -- time-nuts@lists.febo.com >> To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com >> and follow the instructions there.

time-nuts@lists.febo.com

uBlox F9T testing - best settings?