The other rather nice thing is that the tapping reducing pendulum amplitude increases going. This seems exactly in accordance with pedulum theory [link] so where does this leave the theory about amplitude increase being casued by the sensor being off centre?
14 Jun 2016 [22:00]
OK, so I went up the tower again tonight (with Paul Ashley to watch the St John's fireworks) and gave the pendulum hanger a second tapping. Yet another drop in amplitude (see below). If the theory is correct then as the temperature stabilises the amplitude ought to drop again and tapping should give upward steps. This should satisfy John Hodgson - I don't think economists are very good at making predictions! [link]
14 Jun 2016 [22:15]
changing the suspension would count, in my view, as irreversible, because if it works then who would want to put back the old one? And there's not much point in going to the trouble (and it would be a lot of work) if there's no intention of using it. The barometric compensator is fine because it can be removed in under 5 minutes and the clock is back to its original state. Likewise the teflon tape and graphite powder. The only thing I don't like about the vibrator idea is that it needs a battery power supply. I'm wondering about energy harvesting from the bell flies to charge up a battery...,
14 Jun 2016 [22:15]
When I tap it I tap the shaft horizontally back and forth the 0.5mm a couple of times. PTFE is convenient. When I first put it in about a year ago (look at diary comments April to June 2015) it seemed to make a difference. Perhaps the graphite powder too. Not sure which! And for mods like knife edges, etc, they'd have to support the 100kg weight of pendulum. A contact area of 20mm^2 leads to stress of 50MPa so the "knife edges" would be quite thick. I suspect the 19c clock makers would have done it this way if it was a better solution. I also wonder about creating a third lightly damped degree of freedom. Torsion is already coupling with E-W swing, and adding N-S swing at a close frequency will be complex. I'm also reluctant to make irreversible changes to this heritage clock. The idea of a vibrator has crossed my mind, something I'll try if lubrication isn't enough.
14 Jun 2016 [22:15]
Some new information: I tapped the pendulum hanger and the pendulum amplitude suddenly reduced. Aha! I think the tower tilt theory is looking stronger. The optic sensor that measures amplitude is mounted on one side of the pendulum shaft (see photo [link] ) It's a top down view, so the sensor is at the bottom of the photo. Note that the pendulum suspension is in vee blocks so that if the tower tilts then it ought to remain swinging in a vertical plane. But friction (which I've tried to reduce with teflon tape and graphite powder) prevents the pendulum from being perfectly vertical. But if it swings out of verticla it also twists. Now, the sensor is off centre so twist registers as a change in amplitiude. So knocking the hanger has made the swing vertical and twist is reduced. In fact from this it is possible to decuce direction of tower tilt, and it's consistent with warming on the sunny side. Getting there!
14 Jun 2016 [22:15]
Thermal diffusivity is a useful parameter to describe heat flow through the stone walls. It is about 1e-6 m^2/s. This means that a step change in temperature will take 1 second to reach a depth of 1mm, and 3 days to penetrate half a metre. This is why I'm thinking that a gradual rate of change of temperature (over days) can cause tower tilt.
14 Jun 2016 [22:15]
The clock itself is an engineering marvel, accurate to better than 1 second per month. Most of the knots have been combed out, see here for example [link] and what remains are curious glitches that wouldn't even be noticed unless we'd bothered to look. What is great about this is the opportunity to revisit and admire the fabulous achievements of 100s of years of engineers and scientists and to see what little we can contribute ourselves.
14 Jun 2016 [22:15]
tapping the pendulum hanger to check if the twist effect is still happenning - and yes it is: [link]
14 Jun 2016 [13:30]
winding with Sir Jonathan Ive and family and with Roberto Cippolla. Also tapping the pendulum hanger to see if there is a tower tilt effect.
09 Jun 2016 [07:55]
Winding with Anna Basaj
06 Jun 2016 [14:00]
I have been thinking of sticking of the zinc-steep compensation in the pendulum tube. There are several reasons for NOT thinking it is a sticking zinc tube. 1. It is not sudden - it happened over 8 hours from 10am to 6pm <[link]> The clock normally responds very quickly to even the tiniest of mechanical step changes, eg <[link]> which was an adjustment on 1 May. 2. If the zinc is sticking then we would expect under-compensation, ie slower going as temperature rises. It seems to be faster, so it is the wrong way round. 3. It is a small effect, perhaps 200ms/day fast at the moment. Given that an uncompensated pendulum would be -600ms/day/C then if it were sticking we would expect something bigger 4. stress = E * alpha * dT and for zinc E=100GPa and alpha=30e-6/C , say. Then for a 1C temp rise stress = 3MPa, and with a cross section are of about 1000mm^2 then that is an axial force of 3000N, or 300kg. I do not think there is enough friction to sustain this. When we removed the zinc tube a while back it came out relatively easily. 5. There is a small cyclic stress pulse every 1.5 seconds due to centrifugal force. It s about 2N for a 100kg pendulum bob. This ought to be helping to dislodge sticking. But what can it be???? One clue, perhaps, is that the change occurred during a maximum of "rate of change" of temperature. But why?
06 Jun 2016 [14:00]
To give you an idea of what is "normal" then here is data since November. [link] Note that the red and blue arrows indicate intentional adjustments, but that the recent glitch is really one of many, perhaps the most dramatic in recent months. You can change the date back a few years if you want. Also you can look at air pressure, humidity, swing amplitude, etc. Simon mentioned spiders - search for that in comments too. The spider incident was, I think, when a web was spun in the pendulum chamber. the pendulum amplitude reduced which speeds the clock up. But this time the amplitude has not changed, so it is not spiders again - I think.
06 Jun 2016 [14:00]
Things like pigeons standing on the hands, wind, snow etc do not have an effect - that is the beauty of the gravity escapement. Likewise lubrication of gears etc. The effect of opening doors - do a search for "the bedmaker effect" - is understood too, after a fashion. The clock dial faces south and it is definitely possible that sunshine has an effect, but not as one would expect. I have a student (from St Johns) this year exploring the effect of warming of the south-facing wall, thermal expansion and therefore tower tilting. The pendulum develops a noticable twist from time to time and this I think is due to tower tilt. It would be good to see if tower tilt, pendulum twist and changes in going are correlated
06 Jun 2016 [14:00]
so what has caused this dramatic change in going? [link] Temperature compensation is good. Pressure also (air density affects buoyancy). Humidity is a small effect, humidity in this unheated room is fairly constant. There are step changes, like on 6 June. Still stron is the theory that sunshine on south-facing wall causes thermal expansion and tower tilt. I have a final-year project student (from St Johns !) to measure this effect, strarting October.
05 Jun 2016 [12:08]
clock tour for the winner of the Rag Auction: Andrew Sellek, Finn Kristensen, Anthony Wong, Noah Porcelli, Jess Young, Jess Callaghan, Daniel Thompson, Yuen Ng, Toby Henley Smith
03 Jun 2016 [08:04]
Winding by Tit hall engineers Isaac Uden, Micheala Chan, Nathan Clark and Joe Myescough from Windfall films!
![[link]](http://www2.eng.cam.ac.uk/~hemh/clock/photos/top_view_of_pendulum.jpg){kind=link}