IET
Decrease font size
Increase font size
Topic Title: Time to think
Topic Summary:
Created On: 15 November 2011 11:11 PM
Status: Post and Reply
Linear : Threading : Single : Branch
Search Topic Search Topic
Topic Tools Topic Tools
View similar topics View similar topics
View topic in raw text format. Print this topic.
 15 November 2011 11:11 PM
User is offline View Users Profile Print this message



jpchambers

Posts: 14
Joined: 18 January 2003

There seems to be a strong possibility that Coordinated Universal Time (UTC), the world standard time introduced in 1972 and now adopted by most civilized countries, will be signiificantly redefined with effect from 2017 without the knowledge, and certainly without the informed consent, of seven billion users.

Link to NPL 'The leap second debate'

Traditional time, based on dividing the mean solar day into 86,400 seconds, is still the basis of legal time in the UK, where it is known as Greenwich Mean Time (GMT). At any given time and place the sun appears in the same part of the sky year after year, convenient both for sundials and steered solar panels, as well as for navigation independent of GPS.

When UTC was defined it was cunningly reconciled with GMT by the introduction of leap seconds. The exceeding accurate atomic time scale (TAI) based on counting the 'world's best' (SI) seconds' was bound to drift apart from GMT, as there are no gear wheels between the Caesium atom and the Earth's rotation. UTC is a TAI offset by a whole number of SI seconds so that UTC is kept within a second of GMT. Currently TAI is running faster than GMT so this offset is increased from time to time, by the creation of a UTC minute containing 61 seconds, either at the end of a UTC year or at the end of June UTC.

The definition of UTC requires that broadcast time codes and time signals conform to UTC, and that's why the lenghened sixth (or very occasionally seventh) pip appeared in the 'Greenwich' Time Signal (GTS). The time signal no longer signals GMT except for brief periods between leap seconds when UTC coincides with GMT.

The majority of time users are only aware of the effect of leap seconds when they notice that their free-running clocks and watches suddenly become exactly a second further ahead (or less behind) at midnight UTC. Radio-controlled clocks will stay in step with the UTC seconds ticks but lose a second at or soon after midnight UTC.

The proposal is apparently to retain the name 'UTC' whilst abandoning leap seconds, allowing UTC to 'cast off' from GMT. This will make life easier for those designing and maintaining equipment which needs to implement leap seconds. Recently there was a period of several years between leap seconds, so there might have been a new generation who had never encountered a leap second in their career. Applications where precision timekeeping is essential, notably GPS, already avoid the problem by running on a version of TAI (with a fixed offest, and no leapseconds).

In 1972 the notices of leap seconds were distributed through the post, but now the offset between UTC and GMT could be distributed worldwide in real time.if necessary.

An alternative proposal

In my view the above proposal is a case of the 'tail wagging the dog'. I suggest that, instead of casting UTC off from GMT the world take the opportunity reconcile these two a thousandfold better by making the offset an integer number of milliseconds, with leap milliseconds at the end of a GMT hour when required. I suggest that these steps would be easily implemented in time codes (coherent with the 60 kHz MSF carrier) and time signals. Precision equipment would be able to infer the leap milliseconds without needing notice of them. Equipment for everday use, even stopwatches for sporting events, could be based on this 'new UTC' without problems.

It took hundreds of years to reform the calendar, how can we get a proper consensus on the reform of time?
 17 November 2011 12:16 AM
User is offline View Users Profile Print this message


Avatar for sfchew.
sfchew

Posts: 589
Joined: 10 December 2002

The ITU Assembly is to be held in January 2012. The member countries can be given the reason for the right choice by giving them the necessary information for the motion.

I am from Malaysia and is willing to try and influence our country's decision on the matter.

Regards
Chris Chew
 17 November 2011 11:28 PM
User is offline View Users Profile Print this message



jpchambers

Posts: 14
Joined: 18 January 2003

Thank you Mr Chew for your comment. Do you know whether anybody in your country is actively considering this subject? I expect that in most countries the problem is to decide (1) is there a problem which needs to be discussed? and (2) who then should discuss it? But there is no obvious person even to raise question (1),
which is what I am trying to do.

There was a meeting in the UK to discuss this issue

UTC for the 21st century

but I do not yet know what the outcome was.

I am worried that this is an issue which will affect everybody in the world for as long as there is anybody in this world but nobody discusses it except the group of people who work with precision timekeeping and who already have a solution to their problems (namely, International Atomic Time,TAI).

And I am also trying to say, 'if there is a minor problem with the time which we have now (UTC) let us now take the opportunity to make UTC better for the masses, rather than more convenient for the few who don't even use it. Let us abolish 'leap seconds' but replace them by 'leap milliseconds'
 18 November 2011 08:27 AM
User is offline View Users Profile Print this message



ectophile

Posts: 551
Joined: 17 September 2001

I have to say that I completely fail to see what leap milliseconds would achieve.

The average person on the street doesn't really care. Leap seconds are just a curiosity - most people's watches won't be any more accurate than to the nearest minute. They won't take any notice at all any leap milliseconds which change every day or so.

For anybody maintaining computers (or computerised systems that need the time), any changes to the clocks are a nuisance. It's hard enough to get leap years and daylight savings time right, without having to constantly adjust the clocks by a millisecond here or there.

And how does any of this affect stopwatches? They only measure the difference in time between two events. This should be measured in minites/seconds without any time "corrections" being applied.

-------------------------
S P Barker BSc PhD MIET
 18 November 2011 02:21 PM
User is offline View Users Profile Print this message


Avatar for sfchew.
sfchew

Posts: 589
Joined: 10 December 2002

Thank you Mr Chew for your comment. Do you know whether anybody in your country is actively considering this subject? I expect that in most countries the problem is to decide (1) is there a problem which needs to be discussed? and (2) who then should discuss it? But there is no obvious person even to raise question (1), which is what I am trying to do.


As far as I know our country will be represented by government ministry officials. I can only inform them of the industry's preference if I can convince them on the reason behind the choice. I appreciate strong reason to back the millisecond correction.

Regards
Chris Chew
 18 November 2011 08:35 PM
User is offline View Users Profile Print this message



jpchambers

Posts: 14
Joined: 18 January 2003

Thank you, Dr Barker, for asking what leap milliseconds would achieve.
This has helped me to develop my thoughts as follows -

Why leap milliseconds?

An atomic time scale without any discontinuities (leaps) already exists as International Atomic Time (TAI) together with offset versions of this such at GPS time. So there are already time scales available for those who need a precise timescale without complications.

The leap second puts a systematic 'jitter' of 1 second peak-to-peak in the UTC time scale in order to approximate mean solar time. The leap second occurs so infrequently that people forget, or do not learn, how to handle it. Normal timepieces appear suddenly to gain a second if the leapsecond is forgotten. It is a nuisance for some people, notably those maintaining pendulum-controlled public clocks, including the Parliamentary clock 'Big Ben'. I have a recording of 'Big Ben' striking on the fifth pip of the seven-pip time signal when a leap second was added. The timings of the leap seconds need to be widely distributed several months in advance to allow for preparation, and further widely alerts are needed immediately before the event to avoid confusion.

Leap milliseconds would reduce this peak-to-peak jitter from 1 s to 1 ms when approximating mean solar time. This jitter would not be noticed by the majority of users of time signals, time codes, and network time servers. Most of them would not even need to know when the leap milliseconds are implemented, perhaps once, twice, or even more per day. The providers of the time signals would, of course, need to know when to implement the millisecond leaps.

Some might argue that the only proper way to provide mean solar time is to have, in effect, somebody continually steering a master oscillator in accordance with smoothed astronomical data. But if anyone actually required such a time scale, for example, for steering telescopes, they would almost certainly be able to do to job themselves to their own preferred specification. The object of the leap millisecond proposal is to make available a version of mean solar time similar to that which we enjoyed before 1972 but which is precisely related to International Atomic Time, so that data collected using one system can be readily and precisely expressed in the other system.

Anyone maintaining a system controlled by a master clock with a frequency being a multiple of 1 kHz would have no trouble in locking the clock to seconds pulses containing leap milliseconds. All but one in 3600 of these seconds pulses would always be exactly one SI second apart from the previous pulse, and in the event that a particular second at the end of an hour happened to be 1.001 s (or possibly 0.999 s) long this step would be coherent with the master clock and so obvious that it did not require signalling.

There would be none of the complication in the 1960s when 'precision offset' standard frequencies were broadcast with time codes incoherent with the carrier frequency. All standard frequencies would continue to relate exactly to the atomic time scale(s). For those needing to relate their local time with a particular version of atomic time, such as TAI, the current offset in milliseconds would be available on the internet together with details of the next following leap millisecond(s). Anyone operating a system based on an atomic time scale (such as GPS) could make the new time scale available by arranging to receive and apply the appropriate offset. For compatibility reasons, and only if necessary, an arrangement could be devised whereby 'legacy' leap seconds are declared as appropriate and at appropriate times in order to maintain the correct functioning of mass-produced non-reprogrammable equipment which now depends on the leap seonds.

I hope that this helps to explain the thinking behind the proposal for 'leap milliseconds.

John Chambers FIET
 18 November 2011 10:46 PM
User is offline View Users Profile Print this message


Avatar for sfchew.
sfchew

Posts: 589
Joined: 10 December 2002

With the current technology available to us we can certainly maintain the atomic and solar time in harmony with the necessary adjustment. I support the 'leap millisecond'. On the other hand it may be necessary to evaluate the reasons for opposition from the industry quarters. I think we are deciding between the ease of practice and the law of nature.

Regards
Chris Chew

Edited: 03 December 2011 at 03:28 PM by sfchew
 01 December 2011 04:19 PM
User is offline View Users Profile Print this message



jpchambers

Posts: 14
Joined: 18 January 2003

Thank you, Mr Chew, for your support for the 'leap millisecond'. If we can double the number of supporters another 32 times we will have the whole world on our side!

It was not intended to be the only solution offering a UTC for people who wish to retain UT (mean solar time), or who simply wish to go back to something close to the pre-1972 UTC. It was simply an example to show that it can be done. But I suspect that nobody is yet asking for this because they don't yet know that they need to ask.

Concerning stopwatches (Dr Barker) a stopwatch based on differences of leapmillisecond UTC would give the result correct to within one millisecond in several hours. If better accuracy is required the stopwatch would use a stable internal timebase (e.g. ovened or temperature-corrected crystal oscillator). This might be free-running, or locked to a standard frequency (which would always offer 'real'seconds'), or locked to the leapmillisecond UTC where its lock would not be disturbed by millisecond jumps in the seconds markers, particularly if the internal clock is a multiple of 1 kHz.

I have raised this issue in a letter "Time is running away with itself" today (December 1) in The Times (London) in the hope of alerting a wider audience.
Statistics

See Also:



FuseTalk Standard Edition v3.2 - © 1999-2014 FuseTalk Inc. All rights reserved.