Last modified: 2023-09-25 02:12:58
< 2023-09-23 2023-09-25 >Yesterday I said that as the drive torque increases, the tick rate drops. And I also said that as the balance wheel gains amplitude, the tick rate drops. And yet I did not make the connection between these two effects.
Maybe the only reason the tick rate drops as the drive torque increases is because the amplitude increases!
So let's trying replacing the copper hairspring with a steel one and see if either or both of those effects is improved.
Alleyway is mostly done:
Just a bit more to do along under the gutter.
Plasterboard is all done:
And I did the bit under the gutter:
Now just need a man to plaster the plasterboard and Emma to paint everything.
I noticed this a few days ago, and took a picture today, pretty interesting:
There is a big concrete circle buried underneath the dirt that the grass is growing in, and obviously the ground there doesn't retain moisture as well, or something, so there is a big circle where the grass doesn't grow as well!
I've bent up a steel hairspring, using the "APICO" 0.7mm stainless steel wire. Let's see how it ticks...
So it still ticks faster when it has less amplitude (i.e. after it's done more oscillations).
I wish I still had the old data (rather than just the chart) so I could plot them both on the same axes.
I could overlay them in GIMP and see what they look like...?
So, yes, the line is a lot flatter with a steel spring. Good. Also pretty obvious, I should have done this sooner.
So if I repeat the experiment of letting it run 2 revolutions of the escape wheel with 2..14 nuts, how is our isochronicity now? Last time it ranged from about 2.050 to 2.225 Hz (+8.5%).
I'm having some trouble repeating the experiment, because now it has too much amplitude! At first I had to remove some weights which were smacking the balance wheel shaft, and now it is swinging so far that the pin hits the end of the slot in the lever.
"Too much amplitude" is a good problem to have, though! Need to check if putting the slot in the wheel can work, and if so whether it can allow more amplitude.
Also, I noticed that it looks like the escape wheel pivot has not been properly in the frame, which will have increased the friction.
But let's check the tick rate for the data I was able to gather, just to get an idea of a comparison, and then go about either trying to allow more amplitude, or using maybe 1..7 nuts instead of 2..14.
And maybe 1..7 nuts?
It doesn't run with one nut. So 2..7 then? And I've put the extra 2 weight bolts back so that it is a more direct comparison.
Whoa! 9 of the 15 measurements all had exactly the same number of video frames (it looks like only 6 on the plot because 3 are on top of each other; and "4 nuts" had 3 measurements).
Once again the natural frequency of the balance wheel is higher than anything we measured while running.
The R^2 of the fit line is rubbish this time, but the value for k is much better! 0.016 instead of 0.091. 5.6x better.
https://www.youtube.com/watch?v=pF3ZOXC_8ng
If we ignore the data point with only 2 nuts (on the basis that it might not be running properly), then the range from 3..8 nuts can be compared between the copper hairspring and the steel one (albeit we need to interpolate to get the "3" reading from the copper one).
Copper: 2.090 to 2.175 Hz (+4.1%)
Steel: 1.734 to 1.745 Hz (+0.6%)
Much better now! I think this is going to work. We can skip the ideas about experimenting with lever length and tooth geometry, because it is already good enough. I do want to try swapping the pin and the slot, though, just to see if more amplitude is attainable without having to switch to the "guard dart" system.
So that I'm not fooling myself, what if we include the "2 nuts" reading:
Copper: 2.090 to 2.230 Hz (+6.7%)
Steel: 1.734 to 1.765 Hz (+1.8%)
Still loads better. And even if the tick rate varies by 1.8% over a factor of 4 variation in drive torque, I think that is probably acceptable, because the drive torque variation will be a lot less than that.
In this video, Roger W. Smith shows us how the "basket-weave" pattern is applied to a dial using the straight line engine: https://www.youtube.com/watch?v=LBiiDpexmA8
This makes me want to pick up CNC guilloche again. Just make a toolpath that does that pattern, and run it with the drag engraver on some random bit of metal.
The "Watch and Clock Making and Repairing" book includes a drawing of a "Douzieme Gauge", this is a spring-loaded caliper where the pivot is close to the object you're measuring, and the needle is on the end of a long lever, pointing at a scale far away, so that the scale is 12x ("douzieme") as big as the object.
Like this:
I should make one.