aeh5040

Very interesting data! Can comparing the sound traces of the anomalous periods (perhaps averaged somehow) provide any clues? Or perhaps a slowmo video? (I assume you have ruled out the rewind). Also, very cool creation (of course)! The hanging face is very neat. Really enjoying the artistry as well as the cleverness.

Personally I kind of like the (humourous?) contrast that the whip (or wishbone) is the only organic looking part. That's what reminded me. Yours is indeed less creepy!

It's simpler than it looks. Basically 3 diffs sum the inputs to the 4 bottom drums and output the result to the top (day of week) one. The 4 inputs have very strong and precise detent mechanisms. Finally, the year drum also controls the sliding (alternating) leap year window on the month drum. It is crucial that 7 and 2 are coprime. The "offloading" is not really my idea. This is how the existing tabular day-of-week methods work. The devil is in the details. Currently the devil is finding plenty of details to occupy himself with.

Some progress. Getting the overall size and shape pinned down. MLCad is proving invaluable. The monster cylinder at the bottom is needed for the final two digits of the year, which will need up to 12 options per row.

In particular I am reminded of this: Steampunk colours: yes! I am thinking similarly for my calendar, and I think we are not the only ones. This may turn into a de facto steampunk contest!

Yes, studless building is a whole new game! It took me a long time to get used to it, but once you do it opens up many new possibilities! If you want to minimize parts it looks as though part of the base could be replaced with a 5x7 frame, and the gear mount could be replaced with the new 2-pin version of the steering part. Rubber inserts in the base as feet might be nice...

Amazing! It also has a sort of surrealist look to it, with everything centred on the whip piece! I hope it will retain its visual prominence in the final version.

I'm afraid I don't know more. I don't have it at my current location, and I'm not sure if I ever opened it (but it was very clear that this is what had been done). I would guess quite a lot and quite small, but I realise that's not very helpful! There are some pretty cheap ball bearing selections available (Amazon etc) - easiest may be to just get one and experiment...

These wheels look interesting

Long ago I bought a used 8480 and found that someone had added tiny ball bearings inside the turntable. It was quite effective at making it turn better.

Leap year - yes, exactly that! I did try this, but it ends up pretty bulky and heavy. The cylinder is lighter weight. Thank you for the encouragement! We'll see whether it actually works...

Some progress on the leap year selector mechanism: Seems a bit of a shame to put stickers on this drum, but trans blue was the most affordable colour!

The geometry is proving quite tricky!

Not more complicated. All this is accounted for, except that I will stick to the Gregorian calendar 1600-2400.

Wow, impressive analysis and results! Curious how the total error graph kind of looks like two line segments rather than, say, a parabola. Could be just my mental over-fitting though!

In case anyone is interested, I experimented with custom diffs like this to reduce the gear lash, but there was way too much friction, not to mention the bulkiness. The current approach seems the most promising.

I think most of those listed are really variations on the same one. Thanks for your interest!

The yellow one is bizarre! But potentially a useful tool for half spacing.

Thanks! Not exactly - the plan is for it to work similarly to the methods described at https://en.m.wikipedia.org/wiki/Determination_of_the_day_of_the_week. So there will be a month dial with 7 (or more likely 14) positions geared to the month knob, with the months suitably arranged (not in order) around it. Some of the slots will have more than one possible month in them. The dial for the last two digits of the year will have 14 positions, with leap and non leap years alternating, and again many different years in each position. In my head I have a fully functional design, although the practicalities won't be easy. Not sure whether I'll actually get it done by the deadline, but in any case it will provide good motivation! Handling the year 2000 (and 1600 and 2400) presents an extra challenge, since they are leap years, but other years ending in 00 are not. I think I have a solution to this that is actually a bit nicer than the calculating methods listed on Wikipedia. More to come...

Video of final model (see below for development thread) ---- Not quite sure what to call this thing... A machine to instantly calculate the day of the week for any calendar day of any year (e.g. birthdays, etc.) There will be four hand turned knobs to select the century, year, month and date, and the day of the week will then be displayed. The idea is to do it with differentials, and lots of 1:7 gear ratios. Leap years will be accounted for automatically by an extra lever that adjusts the selection for Jan and Feb (hopefully!) The biggest enemy is gear lash. However, I believe my current prototype shows that it can be done (just). I use a very precise detent mechanism for the knobs, and then gear up before the differentials and down at the end... Hopefully custom stickers for the dials are allowed... A long way to go!

Animals (real or imaginary) would be great!

Yes it does, although I don't immediately see as many applications...

Following on from the discussion of gear clearance above, I recently discovered that this works (perhaps it is already known to others?):

It's very hard for anyone to help with so little information to go on. I suggest you at least post some pictures or video.

This looks really useful! It would be nice to have a version that works with the in-grid train wheel bearing (if that's even doable with the small bananas).