There's also the fact that a pendulum only obeys an *approximation* to simple harmonic motion, and this gets progressively worse as the angle of swing increases. I'm not sure whether the period actually alters if the swing is too wide.
But I agree, for normal amounts of swing, a fundamental defining characteristic of a pendulum's usefulness ought to be that it is independent of swing.
NY was thinking very hard :
The PC reads the RTC as it boots, then the PC maintains it own separate version of time until next boot. What you see when you check the time on your PC is the PC's version. Adjusting the time, writes to both clocks.
Sod's law of course then means that I had to forget to add "for small amplitudes that is"
sorry
to make up for it, I've dug out one of the formulae for the true period. That suggests the pendulum would need to increase its amplitude by about 15 per cent to make a difference of 10 seconds a day
thanks - though it'd have been even more helpful if you'd nudged me on that before I posted ;)
The Natural Philosopher formulated on Friday :
My recollection is that it used(s) the CPU's own crystall oscillator, rather than a free running oscillator. Still rubbish at keeping time though.
In the pre NTP days, I used to have a TSR, which would frequently adjust both clocks, based on time running and time off.
It is not.
It is approximately true for small angles of swing
yes. It does. Imagine if it ALMOST goes full circle and hesitates at the top...and falls back down.
In reality it is given just enough impetus to have a CONNSTANT amplitude of swing.
Ok?
Ok.
Ok, so that could be an equally plausible reason. ;-)
Cheers, T i m
Fine weather = high air pressure = denser = more resistance. Maybe it's to do with humidity.
On older versions of Windows (certainly up to and including W7) there's a key which sets the number of seconds between polls of the NTP server). I wonder if W10 doesn't use it. The key is
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\services\W32Time\TimeProviders\NtpClient\SpecialPollInterval
That's SPECIALPOLLINTERVAL (a lower-case L and and upper-case i look identical in some fonts).
On a default installation of Windows, the value is 604800 (7 days) but I've set it to 86400 (1 day).
You need to restart the W32TIME service for it to pick up changes in the key.
My only Win10 PC is difficult to get at right now, otherwise I'd try changing the key and seeing whether W10 honours it - eg set it to 1 minute, restart W32TIME, manually set the time to an almost-correct value and then check that it magically changes to the correct time within a minute.
NY explained :
I know about the key and I found it in W10, setting it seemed to make no difference at all that I could tell. I have used the same key before in previous versions and had it work fine.
harry wrote on 03/08/2018 :
Air is denser in cold weather.
A common problem with pendulum clocks. I believe the problem was solved by virtue of the Notwen Elppa damping system. One of these little units will ensure that no pendulum goes into overdrive.
Operating without same can tip the balance and induce positive feedback into the pendelum thus having a destabilising effect on the space time continuum.
AB
Also, the amplitude of the pendulum will (almost) be controlled by the escapement. There will be a second order effect though.
But I don't think it is the density which controls the resistance, but the viscosity, which goes in the other direction.
Checked my desktop's clock just now against Time.is
"Your time is exact! The difference from Time.is was -0.003 seconds (±0.044 seconds)."
IIRC, I got a similar result the last time I tried this some months or years ago.
I've been running Linux Mint 17.1 KDE64 for just over 3 years now so have no idea of just how s**te the MoBo's RTC actually is[1]. The quartz crystal oscillators are stable enough, the problem is the manufacturers simply can't be arsed to calibrate them[2].
The later versions of MS windows are pretty shoddy with regard to syncing to NTP. That business of leaving the System Time to go unchecked/ corrected for a week at a time is yet one more minor (in the overall scheme of MS badness) thing not to like about windows ten.
Just refreshed the page...
"Your time is exact! The difference from Time.is was +0.001 seconds (±0.043 seconds)."
and a minute later...
"Your time is exact! The difference from Time.is was +0.000 seconds (±0.044 seconds)."
That's via Opera's VPN. I'll turn off the VPN and try again.
"Your time is exact! The difference from Time.is was +0.012 seconds (±0.026 seconds).
That's about what I'd expect having eliminated the extra latency of Opera's VPN service. It also confirms that my 13 year old Casio Data Bank
360 watch is still one second ahead since the last time I had to rescue it from January the first 2000[3] a week or so back.Ever since I disabled the VPN option, subsequent refreshes with time.is have been giving differences in the +0.012 and +0.011 range (±0.025 seconds). The last time I checked the time.is site under win2k, ISTR seeing variations of 2 or 3 seconds between ntp sync events (but that was with the previous MoBo which may or may not have had a less inaccurate RTC).
=========================================================================
[NOTES] [1] The OS upgrade from win2k to Linux Mint being the result of a major hardware upgrade that had sidelined win2k (lack of driver support and more importantly, the waste of a 64 bit CPU on a 32 bit OS).The consequence being that I've been shielded from the RTC's shortcomings as a cheap watch by Linux Mint's NTP daemon service which not only takes care of syncing with ntp services, it also applies a correction factor to the system clock to keep drift to a minimum, unlike MS windows' system clock's time keeping.
With win2k at least, the user could choose their own internet time synchronising schedule. AFAICR, I set mine on a 3 hour 59 minute schedule to reduce clashes with requests from "The Whole World and their Dog" as a compromise between overloading the servers and the clock error getting out of hand.
I'd have to shut the PC down and unplug from the LAN, leave it say 12 hours and make sure to go straight into the cmos setup when I next power up to see the RTC time before Linux gets its chance to 'correct the time' using its calibrated correction factor when it's unable to access the internet, if I want to see just how piss poor the RTC's accuracy is for the lack of any attempt at calibration by the MoBo manufacturer.
[2] Not even to the humble timekeeping standards of your typical quartz watch (a lousy +/- 30 seconds a month when, with a bit of care, they can readily be set to stay within +/-1 second a month from one year's end to the next before needing another tweak - assuming access to the padding trimmer, not a normal option in the typical on-board RTC oscillator). [3] It has the best thought out digital watch display I've *ever* seen but after 9 years of faithful service, just as it had finally settled down to an accuracy of +/- 1 second per annum, it took to blanking out/ resetting itself to the 1st of January 2000 every few months or so.The last time I had to rescue it from its brief sojourn in the past, the 'shock' of the 'rescue' was so severe, the pin holding the retaining toggle on the adjustable clasp of the stainless steel bracelet jumped free. Luckily, I was able to find it and reassemble the clasp.
I really aught to replace it but the only digital watch that comes close to the display ergonomics of my quirky Casio is another DB360 and the only sensibly priced sources seem to be Ebay traders so I've played my strong suite and procrastinated like crazy - it's what I do best.
The theory states that the period of oscillation is independent of the amplitude if the centre of gravity of the pendulum describes a *cycloid* which is the locus of a point on the circumference of a circle, rather than an arc of a circle. (A cycloid looks like a circular arc which is turned up at the ends.) This was worked out by Christiaan Huygens, inventor of the pendulum clock.
Some pendulum clocks achieve this by means of cycloidal jaws constraining the suspension spring.
More usually the amplitude is kept small, where a circular arc approximates to a cycloid, or the amplitude is kept as nearly constant as possible.
Some get their time from GPS if they want greater precision.
It only *needs* to be that accurate but there is real clockmakers kudos in having something that drifts by only a miniscule amount over the period where it needs to be rewound. 200 day clocks for instance.
Marine chronometers and observatory clocks at the beginning of the last century represented the peak performance of classic mechanical clocks. They also included cunning mechanisms that prevent rewinding interfering with the accuracy and amplitude of the pendulum beat.
The Shortt-Synchronome was the ultimate in precision mechanical clocks - good enough to detect seasonal variations in the Earth's rotation back in the 1920's - long before atomic clocks.
That is mainly because the designers CBA to load and trim the crystal properly unlike the ones in cheap watches which are usually done right.
The original PC design didn't have an RTC either - you had to manually set time and date every time you booted it (pre internet).
Domestic "atomic clocks" usually do have a local RTC that is disciplined by the incoming signal to trim it when free running to fractions of a ppm. Simpler ones just free run at whatever the crystal inside happens to oscillate at.
My guess at the design of the pendulum that is running fast would be an invar main pendulum bar with an iron weight that is about 1/10 its length supported from below. This is approximately tempco = 0ppm. Trouble is invar used to vary from batch to batch.
My Windows XP's clock updated this morning and is next scheduled to update the same time (+-) next week. I run SymmTime which is set to update every 60mins from ntp0.pipex.net in Cambridge. Disappointingly:
Time.is Your time is exact! The difference from Time.is was +0.011 seconds (?0.056 seconds)
That would make the difference between finishing on pole and being on the second row.
We've had a lot of high pressure systems, there is a barrometric effect on the pendulum too.
HomeOwnersHub website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.