OT: Alternative clock display

[Jim Thompson]

On Wed, 08 Jun 2005 16:42:28 GMT, Tim Hubberstey <bogus@bogusname.com>
wrote:

Jim Thompson wrote:
On Wed, 08 Jun 2005 04:10:44 GMT, Tim Hubberstey <bogus@bogusname.com
wrote:

If you're running Windows 2000 (and probably XP), there is an NTP client
built into the OS (called W32time, IIRC). All you need to do is enable
it and point it at a server.


All I can find on my Win2K system is w32time.dll

How do you use it?

Socketwatch queries multiple servers. It's presently showing 11
active servers.

Sorry, I don't remember the exact process for setting it up. I'll tell
you what I remember.

1. w32time.dll runs as a service named "Windows Time". In addition to
syncing with a NTP server, it also figures out a fudge factor for your
PC's clock and applies it to maintain accuracy between server queries.
Find it in the Admin Tools -> Services tool and change the startup type
to "Automatic".

2. Open a DOS box and type:
NET TIME /SETSNTP[:ntp server list]

You may need to reboot to have this take effect. If things are working,
you should have registry entries something like this:

[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\W32Time\Parameters]
"LocalNTP"=dword:00000000
"Period"="SpecialSkew"
"type"="NTP"
"ntpserver"="timelord.uregina.ca"
"Adj"=dword:0001873c
"msSkewPerDay"="603.0000"

There is a white paper on setting up the service on the Micro$oft
website. I think I found it by searching for NTP or SNTP. There is also
documentation in the Windows help system.

I was using the US Navy NTP server (TICK/TOCK...), but I found it to be
very unreliable. It is possible (and not unreasonable) that it was
filtering out-of-country requests so YMMV.

It was a bit of a pain to setup initially, but it has been chunking
along without intervention for quite a while now.

I think I'll just stick with SocketWatch, it's working just fine, cost
$10, and requires no mental capacity on my part. (Plus it has a neat
"countdown" feature for setting your watch.)

Presently reporting....

cuckoo.nevada.edu - ***ACTIVE SERVER***
ntp.ucsd.edu - OK
rolex.peachnet.edu - OK
ntp3.cs.wisc.edu - OK
ntp1.cs.wisc.edu - OK

(I have it set to maintain 5 servers minimum.)

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice480)460-2350 | |
| E-mail Address at Website Fax480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.

 

[Tim Hubberstey]

Jim Thompson wrote:

I think I'll just stick with SocketWatch, it's working just fine, cost
$10, and requires no mental capacity on my part. (Plus it has a neat
"countdown" feature for setting your watch.)

Yep. If it ain't broke, don't fix it.

However, you might want to verify that it keeps the PC clock adjusted
between server queries. Sudden time changes can really screw up some
build processes, with "interesting" results. If your PC clock runs slow,
it's not usually an issue, but if it runs fast and gets set back, watch
out! This is why Sun boxes allow you to speed up or slow down the clock
to change the time.
--
Tim Hubberstey, P.Eng. . . . . . Hardware/Software Consulting Engineer
Marmot Engineering . . . . . . . VHDL, ASICs, FPGAs, embedded systems
Vancouver, BC, Canada . . . . . . . . . . . http://www.marmot-eng.com

 

[Jim Thompson]

On Wed, 08 Jun 2005 17:21:31 GMT, Tim Hubberstey <bogus@bogusname.com>
wrote:

Jim Thompson wrote:

I think I'll just stick with SocketWatch, it's working just fine, cost
$10, and requires no mental capacity on my part. (Plus it has a neat
"countdown" feature for setting your watch.)

Yep. If it ain't broke, don't fix it.

However, you might want to verify that it keeps the PC clock adjusted
between server queries. Sudden time changes can really screw up some
build processes, with "interesting" results. If your PC clock runs slow,
it's not usually an issue, but if it runs fast and gets set back, watch
out! This is why Sun boxes allow you to speed up or slow down the clock
to change the time.

I do once-per-hour updates. Last correction (typical) was -55msec.

(And I don't do "build processes"

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice480)460-2350 | |
| E-mail Address at Website Fax480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.

 

[Frank Bemelman]

"Joerg" <notthisjoergsch@removethispacbell.net> schreef in bericht
newsiIpe.1338$bv7.22@newssvr21.news.prodigy.com...

Hello Guy,

I thought that too, but such delays really can't explain why
[ http://tycho.usno.navy.mil/cgi-bin/anim ] would be two seconds
off. If you look at the traffic, once the page is loaded it
sends a handful of bytes once per second, regular as clockwork.
It's hard to come to any conclusion other than tycho.usno.navy.mil
sending late or whatever Joerg is using to detect WWVB being early.
(The latter is remotely plausable if he is using one of those low-
cost "atomic clocks", impossible if he is listening to it.)

Well, both. The atomic clock was low cost, about $20. But it perfectly
corroborates the WWV time signal on 15MHz. The signal via the web was
again about two seconds late this morning, right now it is a little over
one second late.

Maybe your WWV(B?) clock is early. These clocks only try to synchronize
once a day, typically around midnight. The rest of the day the clock
runs on a poorly calibrated crystal, easily loosing seconds per day.
You might want to check by removing the batteries, and reinserting
them and wait until it synchronizes again.

We do use a HW firewall, maybe that has indeed something to do with it.
Also, I did not use any SW that performs latency averaging.

Don't think that would make a difference.

--
Thanks, Frank.
(remove 'q' and 'invalid' when replying by email)

 

[Joerg]

Hello Frank,

Well, both. The atomic clock was low cost, about $20. But it perfectly
corroborates the WWV time signal on 15MHz. The signal via the web was
again about two seconds late this morning, right now it is a little over
one second late.

Maybe your WWV(B?) clock is early. These clocks only try to synchronize
once a day, typically around midnight. The rest of the day the clock
runs on a poorly calibrated crystal, easily loosing seconds per day.
You might want to check by removing the batteries, and reinserting
them and wait until it synchronizes again.

No, as I said I compared it with WWVH this morning and this afternoon.
Both times it was right on and WWVH is realtime, with voice announcing
and all. There is a delay because of the shortwave path but that is minimal.

Regards, Joerg

http://www.analogconsultants.com

 

[Guy Macon]

Frank Bemelman wrote:

"Joerg" schreef...

Hello Guy,

I thought that too, but such delays really can't explain why
[ http://tycho.usno.navy.mil/cgi-bin/anim ] would be two seconds
off. If you look at the traffic, once the page is loaded it
sends a handful of bytes once per second, regular as clockwork.
It's hard to come to any conclusion other than tycho.usno.navy.mil
sending late or whatever Joerg is using to detect WWVB being early.
(The latter is remotely plausable if he is using one of those low-
cost "atomic clocks", impossible if he is listening to it.)

Well, both. The atomic clock was low cost, about $20. But it perfectly
corroborates the WWV time signal on 15MHz. The signal via the web was
again about two seconds late this morning, right now it is a little over
one second late.

Maybe your WWV(B?) clock is early.

Impossible. He said he listened to the WWV time signal on 15MHz.
There is a slight delay because of the speed of light from WWV to
his antenna, a slight delay because of the speed of sound from his
speaker to his ear, then whatever error his nervous system has when
comparing audio and visual events. All far less than a second, and
not able to make the signal come early.

So we have narrowed it down to five possibilities:

[1] tycho.usno.navy.mil, home of the US Naval Observatory Master Clock
(based on dozens of independently operating cesium atomic clocks and a
dozen hydrogen maser clocks [ http://tycho.usno.navy.mil/mc_to.html ]
and the primary reference for the GPS system), is off by 2 seconds
one day and one second the next, and nobody has noticed that the GPS
system is giving answers that are off by many thousands of miles.

[2] WWV, which is derived from the USNO master clock as well, is off
by 2 seconds one day and one second the next and only one of the many,
many people who listen to WWV has noticed this.

[3] Joerg is an alien invader and his interstellar fleet was two
light seconds away yesterday and one light-second away today, but
for some reason having to do with alien technology he can hear WWV
with no delay.

[4] Rich is pretending to be Joerg and is laughing at us as we
puzzle over this clever deception of his.

[5] There can be a 1-2 second delay between tycho.usno.navy.mil
updating it's webpage and Joerg's browser displaying the change,
and I and others were wrong about how big delays through the
Internet can get.

 

[John Perry]

Joerg wrote:

Hello Mike,

NIST has link to info on firewalls on their time server page:

http://www.boulder.nist.gov/timefreq/service/its.htm


Thanks. I found the firewall hints. The gist is that it must be opened
for outbound connection. Nah, I'd rather stay with WWVB and the clock is
just a few feet from my desk. Keeping things simple ;-)

Regards, Joerg

http://www.analogconsultants.com

Well, my home firewall (Netgear RP614v2) offers "port forwarding", which
sounds like what Microsoft is telling you, and "port triggering".

They describe forwarding as opening the port permanently for outside
access. I don't blame you, I wouldn't do this, either.

Port triggering, on the other hand, allows computers inside the firewall
to open a port for a limited time, so that outsiders (ntp servers) can
get through for the limited period.

Their minimum opening is 1 minute, which makes me a bit uneasy; on the
other hand, after the minute, it's closed tight again. I wish the
minimum limit were shorter, but I expect (hope?) it would take longer
than a minute for a hacker to get control of even my wife's XP machine.

Of course, we run other intrusion checks, too.

John Perry

 

[Richard H.]

Joerg wrote:

No, as I said I compared it with WWVH this morning and this afternoon.
Both times it was right on and WWVH is realtime, with voice announcing
and all. There is a delay because of the shortwave path but that is
minimal.

I expect if you ask NIST, that web page is intended as a concept /
gimmick / demo. It'd be highly susceptible to rendering delays in your
browser, which is what you're probably seeing.

The page is actually pretty clever under the hood. It's a multi-part
MIME document - each second, the server sends the next part, which is
configured to overlay the prior page in-place. It looks like a
reloading page, but it's really one page that takes 31 seconds to load
fully. Each block of info is small enough to fit a single packet, and
there wouldn't be a perceptable delay in the Internet transport.

For a simple real-time check, try this one:
telnet time-a.timefreq.bldrdoc.gov 13
Info at http://tf.nist.gov/service/its.htm

Cheers,
Richard

 

[Guy Macon]

Richard H. wrote:

It'd be highly susceptible to rendering delays in your
browser, which is what you're probably seeing.

Any idea why the rendering delay might be 2 seconds one day
and 1 second the next?

I just looked at it using a 56K modem on a 100Mhz 486 running
W2K, testing with IE, Firefox, and Opera, and the digits seemed
to be changing on my screen at the same time the data light on
the modem would blink.

 

[Steve Sousa]

"Guy Macon" <_see.web.page_@_www.guymacon.com_> wrote in message
news:11af4bbdr0u99b0@corp.supernews.com...

Impossible. He said he listened to the WWV time signal on 15MHz.
There is a slight delay because of the speed of light from WWV to
his antenna, a slight delay because of the speed of sound from his
speaker to his ear, then whatever error his nervous system has when
comparing audio and visual events. All far less than a second, and
not able to make the signal come early.

So we have narrowed it down to five possibilities:

[1] tycho.usno.navy.mil, home of the US Naval Observatory Master Clock
(based on dozens of independently operating cesium atomic clocks and a
dozen hydrogen maser clocks [ http://tycho.usno.navy.mil/mc_to.html ]
and the primary reference for the GPS system), is off by 2 seconds
one day and one second the next, and nobody has noticed that the GPS
system is giving answers that are off by many thousands of miles.

[2] WWV, which is derived from the USNO master clock as well, is off
by 2 seconds one day and one second the next and only one of the many,
many people who listen to WWV has noticed this.

[3] Joerg is an alien invader and his interstellar fleet was two
light seconds away yesterday and one light-second away today, but
for some reason having to do with alien technology he can hear WWV
with no delay.

[4] Rich is pretending to be Joerg and is laughing at us as we
puzzle over this clever deception of his.

[5] There can be a 1-2 second delay between tycho.usno.navy.mil
updating it's webpage and Joerg's browser displaying the change,
and I and others were wrong about how big delays through the
Internet can get.

http://gpsinformation.net/main/gpstime.htm


--
Steve Sousa

 

[Joerg]

Hello Richard,

For a simple real-time check, try this one:
telnet time-a.timefreq.bldrdoc.gov 13
Info at http://tf.nist.gov/service/its.htm

Thanks. I had looked at that but it requires to install SW. Since it is
from NIST I trust the SW but I'll have to wait a couple weeks or so for
enough time to do it and test it.

Regards, Joerg

http://www.analogconsultants.com

 

[Joerg]

Hello John,

Port triggering, on the other hand, allows computers inside the firewall
to open a port for a limited time, so that outsiders (ntp servers) can
get through for the limited period.

That is exactly where my concerns start. One minute is plenty of time if
some kid were to hog traffic along the path and just sit and wait for
all the openings from gazillions of users that fly by. I'd rather not.

One minute is plenty of time access your PC and compromise it.

Regards, Joerg

http://www.analogconsultants.com

 

[Rich Grise]

On Tue, 07 Jun 2005 13:12:48 +0100, Terry Pinnell wrote:

Makes a change from those boring LED displays...
http://www.lares.dti.ne.jp/~yugo/storage/monocrafts_ver3/03/

...but, after about a minute, not a welcome change!

BTW, time is correct here (Timezone = GMT, currently plus 1 hour
Daylight Saving Time); how is it for others?

It's showing my system clock time, which is set to Pacific Daylight
Time.

I got lucky and got to see it change from 07:59:59 to 08:00:00.
Kinda busy!

Cheers!
Rich

 

[Rich Grise]

On Thu, 09 Jun 2005 00:46:32 +0000, Guy Macon wrote:
....

Rich is pretending to be Joerg and is laughing at us as we
puzzle over this clever deception of his.

"Ticking away the moments that make up a dull day
You fritter and waste the hours in an off hand way
Kicking around on a piece of ground in your home town
Waiting for someone or something to show you the way
...."
http://www.lyricsfreak.com/p/pink-floyd/108616.html

;-)

Cheers!
Rich

 

[Rich Grise]

On Wed, 08 Jun 2005 16:39:53 +0000, Joerg wrote:

Hello Mike,

NIST has link to info on firewalls on their time server page:

http://www.boulder.nist.gov/timefreq/service/its.htm

Thanks. I found the firewall hints. The gist is that it must be opened
for outbound connection. Nah, I'd rather stay with WWVB and the clock is
just a few feet from my desk. Keeping things simple ;-)

I figure, as long as it's close enough that I don't miss lunch, it's
good enough. ;-)

Cheers!
Rich

 

[Rich Grise]

On Thu, 09 Jun 2005 17:43:52 +0000, Joerg wrote:

Hello John,

Port triggering, on the other hand, allows computers inside the firewall
to open a port for a limited time, so that outsiders (ntp servers) can
get through for the limited period.

That is exactly where my concerns start. One minute is plenty of time if
some kid were to hog traffic along the path and just sit and wait for
all the openings from gazillions of users that fly by. I'd rather not.

One minute is plenty of time access your PC and compromise it.

The last time I tried to install Windows 2000 with an open connection,
I got hacked before I even had time to run "windows update" - i.e.,
it seemed like a matter of seconds. I think the hackers have bots
going 24/7 - you can poll an awful lot of IPs in a minute!

Cheers!
Rich

 

[Joerg]

Hello Rich,

One minute is plenty of time access your PC and compromise it.

The last time I tried to install Windows 2000 with an open connection,
I got hacked before I even had time to run "windows update" - i.e.,
it seemed like a matter of seconds. I think the hackers have bots
going 24/7 - you can poll an awful lot of IPs in a minute!

And crooks can install an awful lot of spyware in that time. If the port
is opened regularly for 1-minute intervals that seems to be comparable
to leaving your house keys on top of the mail box. Well, not in our case
because then there are two large dogs to negotiate entry with but they
won't protect the Internet link.

Regards, Joerg

http://www.analogconsultants.com

 

[Richard H.]

Joerg wrote:

Hello Richard,

For a simple real-time check, try this one:
telnet time-a.timefreq.bldrdoc.gov 13
Info at http://tf.nist.gov/service/its.htm


Thanks. I had looked at that but it requires to install SW. Since it is
from NIST I trust the SW but I'll have to wait a couple weeks or so for
enough time to do it and test it.

Regards, Joerg

http://www.analogconsultants.com

Ummm... this uses telnet...

Telnet to port 13 on telnet time-a.timefreq.bldrdoc.gov... you'll get back:
53532 05-06-11 04:10:01 50 0 0 533.0 UTC(NIST) *

The command-line syntax is commonly:
telnet time-a.timefreq.bldrdoc.gov 13

Format info at the URL above.

Cheers,
Richard

 

[Richard H.]

Joerg wrote:

One minute is plenty of time access your PC and compromise it.

The last time I tried to install Windows 2000 with an open connection,
I got hacked before I even had time to run "windows update" - i.e.,
it seemed like a matter of seconds. I think the hackers have bots
going 24/7 - you can poll an awful lot of IPs in a minute!

And crooks can install an awful lot of spyware in that time.

A couple years back, someone did a study of the probe traffic and
estimated the average time to live for an new / unpatched / unprotected
Windows system on the Internet was 7 minutes.

FYI, most NAT systems allow incoming traffic only for the same IP+port
source+destination pair as the outgoing UDP traffic. So, you're only
exposed if the guy you connected to wants to (and could even do) nasty
stuff through that return channel. I.e., it gives UDP similar abilities
as TCP in terms of dynamic rules.

Richard

 

[Richard H.]

Guy Macon wrote:

Any idea why the rendering delay might be 2 seconds one day
and 1 second the next?

Well, I'd normally chalk it up to browser bloat, etc., but further
testing has been interesting.

telnet://time-a.timefreq.bldrdoc.gov:13 (NIST in Boulder, CO)
versus
telnet://tock.usno.navy.mil:13 (USNO in Maryland)

Yields different times, consistent with what's being discussed here.
Tock seems to be in sync with the web page, but NIST is ahead by 1-2
seconds. Everything I can find says they're supposed to be in sync,
with no allowance for leap seconds that isn't in the other.

My next best guess is that USNO is being overrun by traffic, or is
seeing errors on their links, either of which would cause
retransmissions for TCP traffic and shift the time display as TCP
re-sends. This is consistent with the behavior seen when telnetting to
USNO vs. NIST above.

NTP should be more resilient because it's UDP-based (i.e., won't
retransmit a stale packet) and has mechanisms for detecting network
latency. Clearly, TCP-based connections to USNO are returning stale
time data.

For sites that are really concerned about accurate time sources, the
better approach is multiple GPS receivers with NTP or ACTS (dial-up) as
backup. http://www.truetime.com makes some good units - I like the
NTS-200. Curiously, they only provide for the lightning rod^D^D^D GPS
antenna to be connected via coax; no fiber extenders available.

Richard