WTF with my computer clock?

Live or recorded, it is perfectly possible for broadcasters to maintain program timing to the nearest second; we used to do it back in the sixties, when nationwide network switching was synchronized by people watching Western Union clocks on the walls of broadcast stations all over the country. What has happened is that broadcasters either don't care any more, or there is some commercial advantage to playing fast and loose with the timing. My bet is on the latter.

Isaac

Yes, they do. Timing is based on the number of frames in the entire show, and the frame rate is very, very, accurately controlled by major broadcasters -- figure on something better than one part in a hundred million from any major network. You can use the frame rate, line rate, or color subcarrier frequency as at least a stratum two timebase if you refer to any major network's signals.

That just makes the switching times more precise -- IF the operator cares to be...

Isaac

Wonder if it's because a wrist watch is kept at a pretty constant temperature via the skin?

I know the length won't vary as transmitted, but all one hour progs etc ain't *exactly* the same length.

That's my feeling too. It definitely used to be much better here in the UK, than it is now. If a programme was billed to start at 8pm, then it pretty much did. Now, it is often several minutes late, after they have finished showing genuine commercials, and then long trailers for forthcoming programmes. Even the BBC is now poor, and they only have their own trailers to factor into the equation. I really don't think that they care too much these days, as the 'networks' are no longer formed from independant regional stations, each with their own control centre, which had to synchronise, and jump on and off the network, as the programming and commercial breaks dictated. It probably is just a combination of 'no need', someone's smart-arsed thinking about channel surfing, and the general 'don't really care' attitude that's pervading everything we do now ...

Arfa

Pretty much sums it up. But in those days few had dead accurate clocks which are so common now.

If you'd said 9 o'clock you'd have been right - that was one data point for the network, the 9 o'clock news.

Uh $$$ is my guess.

A computer not on a network? That's blasphemy!

That's ok. Nobody ever agrees with me. I'm used to it.

Yep. However, the IBM PC was designed in 1981. At that time, there were 10 Phase 1 GPS birds, incomplete coverage, and $5,000 receivers. There were overpriced WWV and WWVB receivers, and no internet. The best you could do was something synced to the color burst frequency of a local TV station, assuming they were on 24 hours per day. Your brilliant hindight is totally correct for a 21st century design, but would be impossibly expensive in 1981.

My point was that in 1981 IBM had a reasonably accurate clock inside the IBM PC using a fairly large 14.31818MHz xtal which could have easily been temperature compensated. There were other computers at the time that had a seperate stabilized RTC that did it the right way. However, the IBM PC was originally designed as a home computer, not a laboratory instrument. The casette tape interface should be a clue. Using it for industrial, scientific or navigation applications was probably never considered by the original architects. We're living with the results today.

You might want to look into:

For NTP I use: us.pool.ntp.org

Or, if you prefer something stand-alone which will give you a good time reference if your network connection is down: use GPS.

It's not hard to find a GPS receiver which has a decent "pulse per second" output on its serial port, as well as standard NMEA sentences. Software packages are available which will monitor the NMEA output and the PPS signal, and synchronize your PC's clock very accurately.

The better receivers (those specifically intended for timing purposes) synchronize the PPS pulse edge with the start-of-second to within a small number of nanoseconds. The limiting factor in your PC's clock accuracy is likely to be the speed at which it can respond to the change in the PPS signal (which typically requires taking an interrupt).

A good timekeeping package should be able to compare an averaged PPS timing (over a period of some minutes) with your system's inherent clock drift, and figure out how to jiggle the internal clock so that the drift averages down to zero. You should end up with time accuracy good to within a few milliseconds.

On Linux, you can do this by running "gpsd" to monitor the GPS, and having it feed timing information into the NTP daemon. In effect, your GPS then serves as a new time source to the local NTP timing pool... it's very accurate in the long term but somewhat prone to short-term jitter. You can, if you wish, configure the ntp daemon to use both the local GPS time source, and one or more network time servers... this will give you redundency in both directions.

Correct. Use "pool.ntp.org", or one of the regional subdomains thereof (e.g. "us.pool.ntp.org"). These domains point to a list of well-connected, relatively-high-stratrum servers which have volunteered to serve as public NTP resources.

On 8/16/2009 10:12 PM isw spake thus:

I don't care about zero. One minute a month is plenty accurate enough for me.

You're admonishing me not to use NIST? Why?

After all, they offer this service to me. See

The NIST Internet Time Service (ITS) allows users to synchronize computer clocks via the Internet. The time information provided by the service is directly traceable to UTC(NIST). The service responds to time requests from any Internet client in several formats including the DAYTIME, TIME, and NTP protocols.

So why shouldn't I use them?

Keep in mind that I use this service *at most* 3 or 4 times a *year*.

....

David, it depends upon how you use it. If you use Windows' or MacOS's automatic time sync or *NIX's NTPDATE, you only access it occasionaly. Windows and Mac access it once a week, NTPDATE does it whenever it is invoked, usually when you boot your computer.

If you are runnin *NIX NTP deamon (including MacOS's) or a third party Windows time sync program, your computer is in frequent contact with the time server. In that case, it would be a good idea not to use those servers as they are heavily loaded down.

For once in a week sync of one computer, you can use just about any server without worry about it being overloaded or adding any additional load.

If you have multiple computers networked together, that is a different story.

Geoff.

I use 8 hours, which guarantees at least one update during working hours.

You're fine, but there are NTP abuse problems among application writers and device vendors:

Since there doesn't seem to be a way to stop someone from flooding the NTP servers with excessive requests, the SNTP group came up with the "kiss of death" packet, which tells the sender to shut up:

Why bother with an internet solution? It won't work for laptops, PDA's, stand alone PC devices, and such. A WWVB receiver is cheap enough that it's included inside weather stations, alarm clocks, wrist watches, and yes.... computahs:

etc... The only problem I can see with building one into a PC is the RF noise generated by the PC will probably trash the receiver. That's what long extension cords and external antennas are good for.

You can also sync to the local CDMA cellular provider, although the prices are close to astronomical:

Got $10.70? Build your own:

On 8/17/2009 6:44 PM Jeff Liebermann spake thus:

Interesting.

But once you've picked up WWV, what do you do with that signal to derive a time base from it? (I guess you gots to know something about the signal, which I don't.) Pretty simple?

Actually,they were very close.

Western Union clocks all over the country were almost always all synched to within a second or so. The technique was to use clocks (those big things with the red sweep hand you may have seen in a broadcast studio) that were basically pretty good, and to synch them to a remote timebase from time to time.

The clocks were pendulum timed and electrically wound (couple of big dry cells inside), and every one of them had a leased-line connection to the nearest WU office, and from there to a national site.

Every 12 hours (AFAIR), Western Union sent a pulse down the wire that "jammed" the sweep hands of all those clocks to 12 (and illuminated a little red light behind the clock face so you could see that your time was being corrected). I don't think the minute and hour hands were controlled. It was up to the engineering personnel in each station to twiddle their clocks' pendulums so the clocks could run within a second or two in 12 hours -- not at all difficult for a good pendulum clock.

So as long as the accounting department paid the WU bill, you could join your network or insert a local commercial with almost perfect accuracy.

Isaac

If you need stratum one precision, NIST is not a good choice unless you live near them; you should choose a stratum one server near to you. If you don't need that precision but just want your computer's clock to be decently close all the time, why put an unnecessary load on any of the stratum one (i.e. high precision) servers? Leave them alone to serve folks who *do* need that accuracy.

There's a decent number of lower stratum servers that sync to NIST and some of the other high precision servers. They are specifically intended to be used by the folks who don't need their computer to be within a fraction of a microsecond of "actual time".

Lots more info here:

Isaac

Except I don't see how you can call something that *must* receive radio signals all the time "stand alone" (I guess you could say that it's on a "wireless network"). And of course, GPS time is ultimately synchronized to that same aggregation of stratum 0 clocks as all the other servers we've been talking about.

That said, GPS is a very good way to distribute precision time. You can see 4-inch mushroom-shaped things on top of buildings at the base of cellphone towers all over the place; they're GPS receivers (often two for redundancy) used for getting the signals timed properly before injecting them into the telco network.

Isaac

Bingo!

Isaac