2023-09-05

Last modified: 2023-09-28 19:07:54

• Roofing
• Escapement prototype
• Meshmill

Roofing

First job: flashing. Went OK. I couldn't use the slit I had cut in the mortar because it would have made the flashing go "downhill" where it meets the house, so I cut a new slit in the bricks instead. Went OK.

I have run out of clear silicone, screw head caps, and barge boards, so waiting on those before proceeding.

Glad to have got it done before it gets hot this afternoon.

Screw caps and silicone are coming tomorrow. Fascia boards on Tuesday. Then we want to look at getting insulation & plasterboard, and maybe some battens to hold up the plasterboard.

Escapement prototype

The parts for the escapement prototype have finished printing, let's try and put it together...

I am having some trouble straightening out the 1.75mm stainless wire. Maybe it doesn't matter.

Great, I managed to snap the tip off the ball end mill. And again. This is a time-consuming technique, even in plastic. Probably I want the holes to be closer to final size before I start reaming.

Well I got it together and it works pretty well.

Here's a recording with some weights and springs added: https://www.youtube.com/watch?v=RApdxL61nmA

Natural frequency of balance wheel: 5.9 Hz

Frequency with more torque (start of run): 5.2 Hz

Frequency with less torque (end of run): 5.2 Hz

Amplitude: 8.2°

Both of the "5.2 Hz" readings were 18 ticks long, and both took exactly the same number of frames of video, which was more precise than I expected. I estimate the torque at the start is maybe double what it was at the end, so that is very promising.

Let's do the same measurement, but this time with no springs or weights. How far out of isochronous is it then? https://www.youtube.com/watch?v=EpGQYWlIT5E

Conclusion: much worse.

Frequency: 3.9 Hz to 4.4 Hz (+12.8%)

Amplitude: 12.8°

Frequency with more weight: 4.5 Hz to 4.7 Hz (+4.4%)

Amplitude with more weight: 14.5°

So, with either type of weight (and with less variation in torque than before), the frequency spans a quite large range instead of being constant. If we want to run within 15 mins/day then we need to stay within about 1%, and this already blows our inaccuracy budget before we've even started dealing with uneven gears, varying power delivery from the spring, etc.

So I think we definitely have to have a spring on the balance wheel.

The amplitude is markedly increased compared to when the balance wheel is resonant, but this is evidently no use. Lesson learnt: making the balance wheel resonant is more important than having high amplitude.

Meshmill

Ruari wants to do some rotary carving tomorrow, so I better make Meshmill support helical paths.

Wrote the code, did a release: https://github.com/jes/meshmill/releases/tag/v0.1.2

It has bugs but might be useful.

I've still never actually ran any rotary gcode generated by meshmill, so will be interesting to see whether it works. It looks believable though. I checked and Grbl does support G93 (inverse time mode), so that is lucky.