Sound sync on London News Year Fireworks

[NY]

I'm just watching the BBC, ITV and Sky News reports of the News Year
fireworks in London. On all the reports, there is a lag of about 3-5 seconds
between the audience chanting the countdown and the bongs of Big Ben - the
first hour-bong was when chants were at about 4 seconds before midnight.

At first I thought there was a sound sync problem between the video and
audio feeds, but when it happened on Sky News and BBC as well as ITV,
although to different extents, I became suspicious. Was there a sync problem
between Big Ben (assumed to be adjusted so it is pretty accurate to GPS
time) and the countdown screens and aerial drone displays that the audience
were using for their chant? Is it a casualty of digital codec delays between
the source and the graphic screens/displays?

I'm always amazed that the clock in the Elizabeth Tower which drives Big Ben
can be made so accurate. My experience with clockwork clocks (eg chiming
granddaughter clocks) is that each one gains or loses a variable number of
minutes each day - each one is not (for example) consistently five minutes
slow 24 hours after being set correctly, but the error varies day by day.
And that's for clocks which have been repaired within the last year or so. I
imagine that the pendulum of "Big Ben" (shorthand for "the clock that drives
Big Ben") is well compensated for variations in temperature and therefore
length. I know they used to (maybe still do) add/subtract coins on the
pendulum weight to make subtle changes to its length. I imagine for
important events like New Year is is tweaked *very* accurately!

With modern digital timing and a light beam which is broken by the pendulum
rod, they can probably time the swing of the pendulum very accurately and
so, even averaging over just one swing, they can determine very accurately
how fast or slow it is ticking and therefore how much correction needs to be
made to the period. A bit different to our clocks which have a pendulum
which is only a couple of inches long and which has a simple knurled nut
which screws up and down the shaft to make the bob rise or fall. The very
act of moving the clock to get at the pendulum, and then unhooking the
pendulum to adjust it and then re-hooking it, probably introduces changes in
addition to the adjustment of the knurled nut. We're fighting a losing
battle! We've got used to tweaking the time of our clocks every morning so
they are at least reasonably correct for that day.

[John Williamson]

On 02/01/2024 10:17, NY wrote:
<Big Ben accuracy>

The movement of the clock that operates the chimes in the Queen
Elizabeth Tower is consistent to within less than a second per day, and
the actual timing is adjusted by adding or subtracting pre-decimal
pennies which rest on the top of the weight. Adding or removing a penny
alters the speed by 0.4 seconds per day.

It is still wound by hand three times a week, which means need never be
more than about a second out,as someon vists the movement every two days
or so. If you are standing on the South side of the river, that is less
than the speed of sound delay for that distance.

When it was made in 1854, it was the most accurate publicly visible
clock in the UK, and for many years, there were people who made a living
by setting their watch by it, and setting clocks in offices and stations
by their watch.

[JMB99]

Post by John Williamson
<Big Ben accuracy>

Big Ben: New Year's Eve marks 100 years of live bongs on radio

https://www.bbc.co.uk/news/entertainment-arts-67632538


hey do extra checks for New Year's Eve.



And with people across the country waiting for the sound of 12 chimes
signalling the start of 2024, it's a big day for Big Ben's clock
mechanics like Andrew Strangeway,

"We'll be here on the day, making sure the clock is running precisely to
time," Mr Strangeway said.

"We'll be taking time checks throughout the day and making small
adjustments to the pendulum, just to make sure that midnight happens
exactly when everyone is expecting it to."

[J. P. Gilliver]

In message <un11fa$2n27b$***@dont-email.me> at Tue, 2 Jan 2024 13:01:31,
JMB99 <***@nospam.net> writes
[]

to time," Mr Strangeway said.
"We'll be taking time checks throughout the day and making small
adjustments to the pendulum, just to make sure that midnight happens
exactly when everyone is expecting it to."

I like the implication that midnight happens when the bong happens,
rather than that the bong happens at midnight (-:

[NY]

Post by John Williamson
<Big Ben accuracy>
The movement of the clock that operates the chimes in the Queen
Elizabeth Tower is consistent to within less than a second per day

I'd be fairly pleased if the clock on my PC (assuming it wasn't synced
with a master time source on the internet) was accurate to +/-5 seconds
a day. For a clock made in the 1850s to achieve that accuracy is
phenomenal, though if pennies are being added and subtracted, maybe that
can be thought of as being resynced fairly frequently or at least as and
when a discrepancy is noticed.

It wasn't until read about Harrison's "Longitude" clocks that I even
realised that clockwork could ever be made that accurate. There is
something wrong if modern consumer clockwork (watches, clocks) is worse
that high-precision from 300 years ago - you tend to assume that what
was precision a long time ago will become commonplace and consumer-grade
a few centuries later.

Certainly my Window 7 PC which doesn't phone home (*) as often as my
Win10 and Linux ones do, can be a couple of minutes adrift if I happen
to sync it manually.


(*) I think by default it's as infrequent as every 7 days, which is
either very optimistic or very stupid. Win10 and Linux I think are every
24 hours by default.

[John Williamson]

Post by NY
It wasn't until read about Harrison's "Longitude" clocks that I even
realised that clockwork could ever be made that accurate. There is
something wrong if modern consumer clockwork (watches, clocks) is worse
that high-precision from 300 years ago - you tend to assume that what
was precision a long time ago will become commonplace and consumer-grade
a few centuries later.

It is all down to cost. Expensive devices can justify using things like
bimetallic balance wheels which use differential expansion to keep the
effective radius constant, and even to compensate for the change in
elasticity of the spring with temperature. You can still buy clockwork
chronograph quality watches, but they can't be made as cheaply as, say,
the Timex of childhood. Cheap watches of all types now, analogue or
digital, are only made well enough to avoid warranty claims, as they
were in the 19th Century.

Post by NY
Certainly my Window 7 PC which doesn't phone home (*) as often as my
Win10 and Linux ones do, can be a couple of minutes adrift if I happen
to sync it manually.

Probably uses a ceramic resonator for the system clock, not a crystal.
Even a cheap quartz crystal can do better than +- 50ppm. (A fraction of
a second per week) A ceramic resonator can be as much as 0.5% out and is
nowhere near as stable.

[J. P. Gilliver]

Post by John Williamson

Post by NY
It wasn't until read about Harrison's "Longitude" clocks that I even
realised that clockwork could ever be made that accurate. There is
something wrong if modern consumer clockwork (watches, clocks) is worse
that high-precision from 300 years ago - you tend to assume that what
was precision a long time ago will become commonplace and consumer-grade
a few centuries later.

It is all down to cost. Expensive devices can justify using things like
bimetallic balance wheels which use differential expansion to keep the
effective radius constant, and even to compensate for the change in
elasticity of the spring with temperature. You can still buy clockwork
chronograph quality watches, but they can't be made as cheaply as, say,
the Timex of childhood. Cheap watches of all types now, analogue or
digital, are only made well enough to avoid warranty claims, as they
were in the 19th Century.

The development of electronic timepieces did for the cheap mechanical
watch of our childhood.

Post by John Williamson

Post by NY
Certainly my Window 7 PC which doesn't phone home (*) as often as my
Win10 and Linux ones do, can be a couple of minutes adrift if I happen
to sync it manually.

Probably uses a ceramic resonator for the system clock, not a crystal.
Even a cheap quartz crystal can do better than +- 50ppm. (A fraction of
a second per week) A ceramic resonator can be as much as 0.5% out and
is nowhere near as stable.

The cheap crystals in watches (often 32768 Hz) have the advantage that
they're at a fairly constant temperature, since they're strapped to your
wrist! When I left electronics, un-ovened crystal oscillators came from
about 50 ppm at best, to about 100 or 500 at cheapest, by the time you'd
taken ageing, drift, and temperature variations into consideration.
I don't think the ones in PCs use other than crystals: the ones that
keep the time when the power is off are basically a watch mechanism, and
may well use a watch crystal, but without the wrist temperature aspect.
But when the PC is actually running, it's implemented in software from
the system clock (copying the time from the hardware one at boot), and
what you actually do with the PC can make it run slow. That, rather than
the actual crystal, is the major cause of drift. (The system clock
usually _is_ a crystal rather than a resonator, but isn't chosen for
precision. [The ones in the original PCs I'm pretty sure used NTSC
colour crystals.])

[John Williamson]

Post by NY
I'm just watching the BBC, ITV and Sky News reports of the News Year
fireworks in London. On all the reports, there is a lag of about 3-5
seconds between the audience chanting the countdown and the bongs of Big
Ben - the first hour-bong was when chants were at about 4 seconds before
midnight.

The crowd would have heard the bongs a couple of seconds late, due to
speed of sound delays. The transmitted bongs come via a microphone a few
feet from the bell.

There is also a noticeable delay in the transmission chain on all
digital transmissions. Round here, the pips and bongs on DAB arrive
anything up to two seconds after the same things on FM. On Freeview, the
delay is even worse.

[J. P. Gilliver]

In message <un0nte$2kgeg$***@dont-email.me> at Tue, 2 Jan 2024 10:17:12,
NY <***@privacy.invalid> writes
[]

Post by NY
I'm always amazed that the clock in the Elizabeth Tower which drives
Big Ben can be made so accurate. My experience with clockwork clocks
(eg chiming granddaughter clocks) is that each one gains or loses a
variable number of minutes each day - each one is not (for example)
consistently five minutes slow 24 hours after being set correctly, but
the error varies day by day. And that's for clocks which have been
repaired within the last year or so. I imagine that the pendulum of
"Big Ben" (shorthand for "the clock that drives Big Ben") is well
compensated for variations in temperature and therefore length. I know

It was part of the original specification for the mechanism that it be
very accurate - it was pushing the limits for such clocks at the time,
but was achieved.

Post by NY
they used to (maybe still do) add/subtract coins on the pendulum weight
to make subtle changes to its length. I imagine for important events
like New Year is is tweaked *very* accurately!

Last time I saw any documentary on it, they were indeed still doing that
- using old pennies, and logging what they were using - I think last one
I saw they were logging it on a computer, so fairly complex calculations
_could_ be made.

Post by NY
With modern digital timing and a light beam which is broken by the
pendulum rod, they can probably time the swing of the pendulum very
accurately and so, even averaging over just one swing, they can
determine very accurately how fast or slow it is ticking and therefore
how much correction needs to be made to the period. A bit different to

I don't remember any mention of anything like that. I think basically
for the purposes it is used, they don't need to be _that_ precise.

Post by NY
our clocks which have a pendulum which is only a couple of inches long
and which has a simple knurled nut which screws up and down the shaft
to make the bob rise or fall. The very act of moving the clock to get
at the pendulum, and then unhooking the pendulum to adjust it and then
re-hooking it, probably introduces changes in addition to the
adjustment of the knurled nut. We're fighting a losing battle! We've
got used to tweaking the time of our clocks every morning so they are
at least reasonably correct for that day.

One aspect you don't have to deal with is bird loading: at certain times
of year, starlings on the minute hand cause an appreciable slow-down or
speed-up effect on the mechanism!

Something that has occurred to be about software imitations (and the
real thing): I use ClockSmith Lite, but I'm sure there are many others.
These usually work such that the first of the 16 quarter chimes occur on
the instant; in the real clock, it's the first bong that is the denoter.
So the chimes must actually set off in advance of the hour.

(Another thing I hadn't appreciated until I saw it explained: there are
_five_ four-note sequences. You'd think that, for the four quarters,
you'd get sequence A at quarter past, AB at half, ABC at quarter to, and
ABCD on the hour, but for mechanical reasons, that's not the case: you
get A, then BC, then DEA, then BCDE, ready to be back at A [my arbitrary
lettering].)