[Thumbgps-devel] "Plain Jane" timing GPS is working!

[Thumbgps-devel] "Plain Jane" timing GPS is working!

[Eric S. Raymond]

A few minutes ago I received mail from the test engineers at NaviSys.
They have tested the "Plain Jane" concept (take a stock SiRf-III GPS
chip plus PL2303 USB adapter, connect the 1PPS pin on the former to
DCD input on the latter) with GPSD, and are actually seeing PPS pulse
messages in the logs.

The filter in GPSD doesn't like the pulse shape and is rejecting it, but
that can be fixed.  The essential point is that the hardware mod works
exactly as expected and is producing events visible on USB. 

I'm expecting my samples from UniTraq today or tomorrow and may soon have
more good news about our second Plain Jane variant.
-- 
		<a href="http://www.catb.org/~esr/">Eric S. Raymond</a>

Re: [Thumbgps-devel] "Plain Jane" timing GPS is working!

[Ron Frazier (NTP)]

Hi Eric,

I had my Sure board working a couple of weeks ago connected SERIAL -> 
Prolific adapter -> USB, with PPS patched to DCD, running NTPD on 
Windows using Dave Hart's new binaries, without running GPSD.  It was 
working fine within + / - 1 ms of GPS time.  It was definitely reacting 
to the pulses.  Maybe you could try with NTPD only.  I was using the 
"20" NMEA driver with the PPS flag 1 active.  I don't know for sure if 
the Linux version of NTPD has all the features necessary to do this.

PS  I haven't seen much traffic on the thumbgps-devel list lately.  Is 
this still "the place to be" for info on this project?

Sincerely,

Ron

On 4/3/2012 10:34 AM, Eric S. Raymond wrote:
> A few minutes ago I received mail from the test engineers at NaviSys.
> They have tested the "Plain Jane" concept (take a stock SiRf-III GPS
> chip plus PL2303 USB adapter, connect the 1PPS pin on the former to
> DCD input on the latter) with GPSD, and are actually seeing PPS pulse
> messages in the logs.
>
> The filter in GPSD doesn't like the pulse shape and is rejecting it, but
> that can be fixed.  The essential point is that the hardware mod works
> exactly as expected and is producing events visible on USB.
>
> I'm expecting my samples from UniTraq today or tomorrow and may soon have
> more good news about our second Plain Jane variant.
>    

-- 

(PS - If you email me and don't get a quick response, don't be concerned.
I get about 300 emails per day from alternate energy mailing lists and
such.  I don't always see new messages very quickly.  If you need a
reply and have not heard from me in 1 - 2 weeks, send your message again.)

Ron Frazier
timekeepingdude AT techstarship.com

Re: [Thumbgps-devel] "Plain Jane" timing GPS is working!

[Eric S. Raymond]

Ron Frazier (NTP) <timekeepingntplist@techstarship.com>:
> PS  I haven't seen much traffic on the thumbgps-devel list lately.
> Is this still "the place to be" for info on this project?

I don't know of any other place. :-)  You're right that traffic has been
light - I think people have been waiting on some of the tests that were
queued up.
-- 
		<a href="http://www.catb.org/~esr/">Eric S. Raymond</a>

Re: [Thumbgps-devel] "Plain Jane" timing GPS is working!

[Dave Taht]

[-- Attachment #1: Type: text/plain, Size: 1801 bytes --]

On Tue, Apr 3, 2012 at 4:35 PM, Eric S. Raymond <esr@thyrsus.com> wrote:

> Ron Frazier (NTP) <timekeepingntplist@techstarship.com>:
> > PS  I haven't seen much traffic on the thumbgps-devel list lately.
> > Is this still "the place to be" for info on this project?
>
> I don't know of any other place. :-)  You're right that traffic has been
> light - I think people have been waiting on some of the tests that were
> queued up.
>

I was waiting on the tests  queued up, (nice going everybody!) and also
thinking a bit about the network tomography problem, and ways to attack
measuring p2p connectivity across deployed boxes. For example I started
fiddling with openhip as a possible way to enable snmp again, even behind
natted boxes. openhip is pretty cool.

But primarily I've been focused on getting cerowrt out the door in time for
the world ipv6 launch day, and have run into problems with dnssec among
others. A dnssec enabled nameserver needs to *know* it's getting valid time
before enabling dnssec, somehow.

In other side news on that I've been getting spectacular results out of the
new AQM technologies we've been fiddling with on that project, but this
isn't the list to discuss that.

Expect light participation from me here through may 5. I would certainly be
open here to a discussion of 'now that we have good time, and geolocation,
what can we effectively measure with it, and with what tools'...


--
>                <a href="http://www.catb.org/~esr/">Eric S. Raymond</a>
> _______________________________________________
> Thumbgps-devel mailing list
> Thumbgps-devel@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/thumbgps-devel
>



-- 
Dave Täht
SKYPE: davetaht
US Tel: 1-239-829-5608
http://www.bufferbloat.net

[-- Attachment #2: Type: text/html, Size: 2734 bytes --]

Re: [Thumbgps-devel] "Plain Jane" timing GPS is working!

[Eric S. Raymond]

Dave Taht <dave.taht@gmail.com>:
> Expect light participation from me here through may 5. I would certainly be
> open here to a discussion of 'now that we have good time, and geolocation,
> what can we effectively measure with it, and with what tools'...

Good time isn't quite a done deal yet.  There are some relatively minor 
changes to gpsd and a bit more hardware testing between here and there.
I'd say two weeks, assuming we don't run into any boojums.  And I won't
sign off that the job is *really* finished until we can document hard bounds
on the jitter using something like the test rig Patrick Maupin is trying
to build.

By the way, I've had a third GPS vendor - a U.S. outfit called Parallax
Systems - get back to me with interest in the Plain Jane timing-GPS mod
(connect 1PPS to DCD in a stock design using SiRF+PL2303 or functional
equivalents). Had a long, friendly phone conversation with one of
their design guys.

For those of you not paying full attention recently, I have
engineering samples of Plain Jane from UniTraq - not yet verified that
these work because I have to mod gpsd so it won't time out the PPS
thread before it gets sat lock.

Meanwhile, Unitraq's test engineers in Shenzhen have shown me a gpsd log 
including PPS events seen over USB.  But we need to tell gpsd's noise
filter to be a bit less picky; for some reason it doesn't like the
pulse width and is rejecting those.

So we have two vendors actively cooperating on Plain Jane product
development and may have a third shortly.  This is all good; my
ideal outcome would be multiple sources, with other vendors starting 
to pick up that they need to supply 1PPS to stay competitive and it 
becoming a stock feature on all new low-end GPS mice within a year.

Actually, it still kind of blows my mind that I've been able to
social-engineer all this with a handful of emails.  Woulda been the
stuff of cyberpunk novels before the century turned...
-- 
		<a href="http://www.catb.org/~esr/">Eric S. Raymond</a>

Re: [Thumbgps-devel] "Plain Jane" timing GPS is working!

[Dave Taht]

[-- Attachment #1: Type: text/plain, Size: 2110 bytes --]

On Wed, Apr 4, 2012 at 6:36 AM, Eric S. Raymond <esr@thyrsus.com> wrote:

> Dave Taht <dave.taht@gmail.com>:
> > Expect light participation from me here through may 5. I would certainly
> be
> > open here to a discussion of 'now that we have good time, and
> geolocation,
> > what can we effectively measure with it, and with what tools'...
>
> Good time isn't quite a done deal yet.


Good p2p network tomography software is barely a half baked, half formed
idea as yet.

Fact checking ntp's raw stats filters is another thing worth thinking about.

Pretend we have good time already... then what?

There are some relatively minor
> changes to gpsd and a bit more hardware testing between here and there.
> I'd say two weeks, assuming we don't run into any boojums.  And I won't
> sign off that the job is *really* finished until we can document hard
> bounds
> on the jitter using something like the test rig Patrick Maupin is trying
> to build.
>

Tell me when a good time to pull is. I would certainly like to try current
against
the wndr, as what andrew had tested was gpsd 2.X.


>
> By the way, I've had a third GPS vendor - a U.S. outfit called Parallax
> Systems - get back to me with interest in the Plain Jane timing-GPS mod
> (connect 1PPS to DCD in a stock design using SiRF+PL2303 or functional
> equivalents). Had a long, friendly phone conversation with one of
> their design guys.
>
>
Did ublox ever get back to you? Sexy sadie is calling...


>
> Actually, it still kind of blows my mind that I've been able to
> social-engineer all this with a handful of emails.


and a couple blog posts!


>  Woulda been the
> stuff of cyberpunk novels before the century turned...
>

The whole maker movement is a barrel of fun! While I didn't
care for much of Stevenson's "Diamond age", the core concept
of a reprap and related is becoming more real by the day.

http://reprap.org/wiki/RepRap

Hmm... self replicating gps and networking gear... hmmm...

-- 
Dave Täht
SKYPE: davetaht
US Tel: 1-239-829-5608
http://www.bufferbloat.net

[-- Attachment #2: Type: text/html, Size: 3268 bytes --]

Re: [Thumbgps-devel] "Plain Jane" timing GPS is working!

[Eric S. Raymond]

Dave Taht <dave.taht@gmail.com>:
> Fact checking ntp's raw stats filters is another thing worth thinking about.

Yes.
 
> Pretend we have good time already... then what?

Then we start thuinking about data reduction for the rawstats, and what things 
are interesting to visualize.

> Tell me when a good time to pull is. I would certainly like to try current
> against the wndr, as what andrew had tested was gpsd 2.X.

I'll keep you posted via this list and personally.

> Did ublox ever get back to you? Sexy sadie is calling...

Alas, no.
-- 
		<a href="http://www.catb.org/~esr/">Eric S. Raymond</a>

Re: [Thumbgps-devel] "Plain Jane" timing GPS is working!

[Dave Hart]

On Tue, Apr 3, 2012 at 14:59, Ron Frazier (NTP)
<timekeepingntplist@techstarship.com> wrote:
> I had my Sure board working a couple of weeks ago connected SERIAL ->
> Prolific adapter -> USB, with PPS patched to DCD, running NTPD on Windows
> using Dave Hart's new binaries, without running GPSD.  It was working fine
> within + / - 1 ms of GPS time.  It was definitely reacting to the pulses.
>  Maybe you could try with NTPD only.  I was using the "20" NMEA driver with
> the PPS flag 1 active.  I don't know for sure if the Linux version of NTPD
> has all the features necessary to do this.

The Windows port of ntpd is the only one which has the crude user-mode
PPS hack needed to use PPS without PPSAPI.  On Linux, gpsd is the way
to achieve the same.  Incidentally that user-mode PPS hack will be
moving from ntpd's Windows serial I/O layer to a separate PPSAPI
provider DLL before long, so that it will be disabled by default and
enabled only when desirable by administrator action.  It will also
behave better via PPSAPI, so that for example NMEA's fudge time1 is
used to offset PPS timestamps as designed, rather than fudge time2
(which is intended to control only NMEA timestamps, but the user-mode
PPS hack substitutes the first NMEA timestamp each cycle with the PPS
timestamp.  Also currently the fact that PPS is used is invisible, but
when it's done with a PPSAPI provider DLL, ntpd will show the 'o'
tally code for the PPS-enabled refclock rather than '*'.

Cheers,
Dave Hart

Re: [Thumbgps-devel] "Plain Jane" timing GPS is working!

[Patrick Maupin]

On Tue, Apr 3, 2012 at 9:34 AM, Eric S. Raymond <esr@thyrsus.com> wrote:
> A few minutes ago I received mail from the test engineers at NaviSys.
> They have tested the "Plain Jane" concept (take a stock SiRf-III GPS
> chip plus PL2303 USB adapter, connect the 1PPS pin on the former to
> DCD input on the latter) with GPSD, and are actually seeing PPS pulse
> messages in the logs.

That's great news.  My OCXO order seems to have stalled; I need to
ping the seller about those.  It sounds like people are fairly
confident that the PPS signals are working correctly (not wobbling)
anyway, but I have some other reasons I want to get a precise local
timebase going.

In other news, TI apparently has some new GPS chips.  Haven't
investigated, but would assume the cost of GPS is coming down even
farther soon:

http://newscenter.ti.com/index.php?s=32851&item=127833

Re: [Thumbgps-devel] [gpsd-dev] "Plain Jane" timing GPS is working!

[Håkan Johansson]

[-- Attachment #1: Type: text/plain, Size: 1589 bytes --]


Is it the pulse shape (i.e. length) or the pulse repetition that gpsd 
rejects?  As I recall, a SiRf-III GPS delivers a PPS pulse that is only 1 
us long, so does not record a full state change, but this is handled.

Timing over USB is distored by the inherent USB polling interval of 1 ms, 
so could make the inter-pulse interval fail those tests.  (They actually 
allow +/-1 ms, so should make it?).  Logfile?

Due to the USB effectively downsampling the measurement to the USB 
'polling clock', such a setup might be prone to seeing jumps.  I.e. if the 
USB polling frequency is close to a multiple of the PPS signal, then it 
for long times drifts slowly (which will not be detected), and then jumps 
one polling cycle.  (Possibly even usable.  Unhandled it could make life 
miserable for chrony/ntp.)

Cheers,
Håkan


On Tue, 3 Apr 2012, Eric S. Raymond wrote:

> A few minutes ago I received mail from the test engineers at NaviSys.
> They have tested the "Plain Jane" concept (take a stock SiRf-III GPS
> chip plus PL2303 USB adapter, connect the 1PPS pin on the former to
> DCD input on the latter) with GPSD, and are actually seeing PPS pulse
> messages in the logs.
>
> The filter in GPSD doesn't like the pulse shape and is rejecting it, but
> that can be fixed.  The essential point is that the hardware mod works
> exactly as expected and is producing events visible on USB.
>
> I'm expecting my samples from UniTraq today or tomorrow and may soon have
> more good news about our second Plain Jane variant.
> --
> 		<a href="http://www.catb.org/~esr/">Eric S. Raymond</a>
>
>

Re: [Thumbgps-devel] [gpsd-dev] "Plain Jane" timing GPS is working!

[Håkan Johansson]

[-- Attachment #1: Type: text/plain, Size: 263 bytes --]


> I should change the log message because people freak when they see all
> the 'reject' messages.  Any suggestions?

How about 'ignored'?  Should reduce the fear factor by at least 50 %.
Or even in case the reason is know: 'ignored trailing edge'.

Cheers,
Håkan

Re: [Thumbgps-devel] [gpsd-dev] "Plain Jane" timing GPS is working!

[Eric S. Raymond]

Håkan Johansson <f96hajo@chalmers.se>:
> Is it the pulse shape (i.e. length) or the pulse repetition that
> gpsd rejects?

I don't know.  The pulse filter code is Gary Miller's and has a lot
of black art in it; I don't understand it at all.
-- 
		<a href="http://www.catb.org/~esr/">Eric S. Raymond</a>