Davidz90

Outstanding work! True. Thermal expansion results in about 3 seconds/day error per 1 degree C change of temperature. The current 35 seconds/day is extremely good for such a small and light pendulum. Out of curiosity, if it isn't too much of a trouble, could you record the sound of ticking (5-10 minutes is enough) and post it somewhere online? I could run some statistics on it, find the average period, standard deviation, difference between left and right swing time etc...

Sleek design! Waiting for stickers to fully understand how it works.

Very cool, and it seems really accurate too! Yes, given rather high-power mechanism, there should be no issues there.

Looks great! Love the color scheme.

Thanks! Yeah, crown gear is an underrated piece; so much more compact than the equivalent 12:36 with new bevels. Auto rewinder was surprisingly challenging - for some reason the on/off switch on the battery box needs an absurd amout of force, so a double lever arrangement was needed to make it manageable. In the future, I'll probably replace the current autorewinder with some sleeker 3rd party components.

Eureka! The antenna hits the locking piece at some speed, bounces back and then hits it again. The first hit is louder, but sometimes the second one was registered as the clock tick. The bounce takes 0.043 seconds.

Good question. I'll try just that - chop the sound trace into periods and overlap them. Slomo is an option too, although analyzing the footage in search of the anomalous period (about ~5% of them are anomalous) could be annoying and there's no telling if they can be recognized visually at all.

I finished fine-tuning the clock. Adding a single 1x1 tile to the balance wheel increases the period by about 0.01 seconds, so this was the smallest correction I could make. That being said, I got very lucky to get a period of 2.0012 s, which is an error of 0.062% or about 54 seconds/day. In other words, after a full day of working, the clock should be late by a little less than a minute. Here is how the individual periods look like The curious thing is the upper band around 2.04 s; there is some common occurence that causes the period to be 0.04 seconds longer than average. So far, I have no idea what this is. Here is the histogram, showning the mysterious upper band as well And here is clock error Started out extremely stable, but at ~12 minutes the period increased very slightly and clock started accumulating error at a rate of about 3 seconds/hour. Keeping that rate, after 24 hours the error would be about 1 minute. That being said, I don't expect the long term performance to be this good. Clock is heavily affected by thermal expansion - not only expansion of the balance wheel makes a bigger difference than identical expansion of a pendulum, but also the whip "spring" is very likely to become softer as the temperature rises.

Exactly! Amazing build and presentation. I love the "industrial sci-fi" style.

The design is more or less finalized. I went for gravity power and electric auto-rewinder to get long running time (probably ~2 days until batteries run dry, it isn't very efficient). What is unique is the overall layout of the clock: it consists of three separate pieces connected with a chain loop. On the top, escapement is located. In the middle, driving weight and auto-rewinder are hanging on the chain. At the bottom, clock face is also hanging on the chain. This is by far the most silly clock design I have made .

One of my favourite shows! Is it going to be removed? But back on topic, looks brilliant! First time I'm seeing chains used like this.

I finally have a general idea how the complete clock will look like. It will be gravity powered after all, but with something a little more interesting than just a weight on a string. More to come. But now, a little comparison of accuracy. Escapement makes a good job of maintaining pace despite large torque variation of spring power, but gravity power is still superior: (embedding images from flickr doesn't seem to work now?) Edit: Ok, Thanks!

Instead of planning the clock layout, I keep fiddling with the escapement design XD

No magic here, just a heavy weight applied almost directly to the escapement axle is just barely enough to keep it running. Also, isn't your mechanism running in opposite direction? In Brick Technology video, the escape wheel is pushing the anchor horizontally at the top and more vertically at the bottom. Or in other words, escape wheel runs counterclockwise with pendulum on the left side and yours runs clockwise. Details like this can matter a lot - I can see that the Brick Technology version runs substantially smoother. In general, with anchor escapement, the geometry needs to be optimized so that the escape wheel spends time mostly pushing the pendulum, not being locked by it. At any rate, look into the suggested escapement, either with knob wheel or in the original configuration with triangular connector (and appropriately changed gear ratio; shouldn't be too hard to replace 4 with 3 somewhere). It is very forgiving.

It will definitely be good, especially if you manage to implement some better pendulum suspension. Yes, thread links should work better than strings. Ok. Seconds can be problematic if the seconds hand is heavy and/or not balanced well, so that needs attention. Minutes and hours are usually non-issue. Ball chime may or may not need its own power source, but lifting a ball doesn't seem to take that much power. At any rate, this part could be attached directly to the electric rewinder, especially if the said rewinder activates often. This puzzles me. On the video, I see close to one second period (swing left to right and back), which advances the escape wheel by one tooth (=1/4 rotation)?

Hahaha :D Yeah, I'm pretty predictable, although I chose balance wheel to have some novelty; simple pendulum would be lazy. Frankly, I don't think that I can compete with other entries as well; the musical machine, lathe, calendar, Berthil's clock are all extraordinary (so far, music machine is my favourite). My humble clock probably won't be much more than already shown and I don't have that much time to work on it.

Yeah, single pin escapement cannot be changed into half rotation easily. Is it half, not quarter? The one on your video has four "teeth". How about Galileo escapement with propeller piece? Going from 4 to 8, You could reduce the gear ratio by 2 somewhere. The striking mechanism has its own weight and a low friction trigger. World time rotates only once per 24 hours and calendar only advances by one day every 24 hours, so both of these are very slow and connected to the mechanism through huge gear ratio. In general, for optimal performance, it is best to use largest gear reduction immediately next to escapement (so one or few 8:40 ratios) and then go from that. Your pendulum seems to have a period of about 0.8 seconds. So 4-toothed escape wheel would make one rotation in 3.2 seconds. For minute hand, 1:1125 ratio is needed. 1125=5*5*5*9, so three 8:40 in a row and then two 8:24. Also, kind of obvious but avoid adding seconds hand if possible; it uses up a lot of energy. Forgot to add: the first escapement in the video I linked is very robust and with little tweaking, it can be converted to a knob wheel to have 4-toothed escape wheel.

Yeah, multiple mechanisms stealing energy from the escapement is a common and annoying problem. One solution is to add a small power source that powers only escapement, which is then rewinded at regular intervals by the larger power source. Or, in your case, all functions could be attached to electric auto-rewinder instead of the escapement. Finally, below are some simple escapements that are not as fiddly as the really efficient ones:

Whoa, what a beast! I'm trying to deduce the exact mechanism from the photo, but I think I'll wait for physical model and some videos. Offloading most of the calculations to some specific pattern on the drums is a very clever solution.

Whoa, this is a little creepy but very cool. Maybe I'll try to incorporate more moving parts with organic feel, to make it more like a kinetic sculpture. So many things to consider, fortunately there's a lot of time left. Yes, indeed I have. Turned out to be way easier than anticipated, I had no idea that balance wheel can be made so stable so easily; I expected it to be a complete garbage compared to a pendulum, but it's only ~10 times worse. Funnily, this is similar to real wristwatches, which are about 10 times worse than grandfather clocks.

I hope not - the optional motorization complicates the drivetrain to the point that it barely works in my alfa romeo, making all sorts of terrible clicks. At least the new car has a reasonably sized engine.

Haha thanks, the whip will certainly remain the centerpiece. I'll try to make the dial as another ring going around the whip. I think the clock would look cool with more steampunk color scheme with brown and gold, orange wheel should go well with such combination.

Another update. General structure is more or less finalized; studded beams are used where extra rigidity is needed. No matter what I tried, studless construction was inferior; technic pins have some play, while plates dont. Given the weak power source, I need to reduce every little bit of friction. It runs for 10 minutes on a single rewind, so it is more of an egg timer than a full clock after all. I plan to add a dial counting down to 0 and some chime/bell that rings when it is reached. Now, there are attachment points for plates on the wheel. By adding mass, one increases the inertia and slows the clock down. This method of regulation proved to be surprisingly good, with very linear relation between added mass and period change. I managed to get very close to 2 seconds period, so the gear ratio to the dial will be based on that value.

Sure! One more thing worth testing is some better pendulum suspension than just an axle. This is also a big source of friction. Just replacing the axle with bar/lightsaber resting on the bottom of technic hole reduces the friction by a factor of 10 or so.

Wow, this is a huge project! Really curious how well will the bearings perform.