Re: Group Focus

Neville Michie

Hi Neil,
I have often thought about the problem of fixing a glass rod in a pendulum.
Some of the adhesives are very good, but there is always the possibility of creep
and the effects of moisture absorption.
One scheme that addresses fragility and creep is to use a glass tube.
Grind and lap the ends flat, and make end pieces to mate with the flats.
Use a steel (or invar)rod inside the tube to tension the caps against the ends.
Glass is very strong in compression and if the tension in the steel rod
is greater than the weight of the bob then the glass will always be in compression.
The steel will respond to temperature changes with a small change in the tension,
and the glass will respond to the change in tension according to its own modulus,
but a very robust and predictable pendulum will be the result.
Neville Michie

On 9 Dec 2020, at 07:40, neil <> wrote:

Hi Bob,
when I said rod, in fact I am using tubing - there is a company here in NZ that make electric heaters and they were able to supply me with 12mm quartz tube that they use, for a reasonable cost. I ordered 4 lengths and 1 got broken in transit so i'm using that one to drill holes in & experiment with. I did try Robert Matthys soldered sleeve type suspension (chapter 25 of his book) but was not happy with that, so I now have a 4mm dowel pin through a hole in a scrap piece with a big bag of lead shot hanging off it. If it survives a week or two the I'll go down that route.
On 9/12/2020 09:10 am, Bob Holmstrom wrote:

For what its worth, glass or quartz tubing is far less likely to break than rod. This is from experience designing quartz ware for the semiconductor industry. Diffusion furnace pull rods made from rod would break on the storage shelf and the same made from tubing were strong enough that we joked that you could fence with them.

They can be easily drilled with diamond burr drills from a lapidary shop - drill from both sides toward the center ignorer to avoid visible chipping - using a pin to located the tubing when flipped to drill the other side.


On Dec 8, 2020, at 11:57 AMDecember8, neil <>

Hi Bepi,
I have that very experiment going on at the moment. My EM clock is controlled by a homebrew Atmega processor board, and I can very easily switch between pulsing it every 30sec ( or any other interval for that matter) vs impulsing it when the amplitude drops to a preset level determined by an opto.
I have a Mumford Microset connected to it to measure the rate, but got side tracked recently - replacing the invar pendulum with a glass rod, and that is still a work in progress. I'm busy boring holes in pieces of scrap glass rod with a diamond burr in order to find the perfect attachment for the suspension and the bob. Maybe over Xmas I will make some real progress.

On 9/12/2020 05:55 am, Bepi wrote:

Thanks John, my intention was to pose an in principle question: how should one choose the amplitude at clock design level. More relevant with mechanical clocks, so much more power hungry and inefficient than the EM ones and where it's also a lot harder to investigate a subject like this experimentally. I was right today discussing of how to build a mechanical escapement with an easy way to control the number of periods per pulse. If anybody is aware of such a design please let me know.
In a sealed case my synchronome works very reliably at exceptionally low amplitudes and its noise level grows a lot compared to normal, I just never stopped to investigate why.
It could be an interesting field of research for a "real" electric clock like yours (or mine) where changing amplitude and pulse rate independently is super easy and the escapement error virtually absent, just like the subject of how often to impulse. Has intermittent impulsing, supposedly superior than pulse by pulse for accuracy, ever been investigated experimentally? p.-

On Dec 8, 2020, at 16:05, John Haine <>

Pepi, I have to say that the amplitude of this clock is determined by its history! 2 or 3 years back I started to make a clock with a seconds pendulum (0.5 Hz) and a target amplitude of less than 1 degree. This was set by the spacing between the drive coil and a hall effect sensor in the base. I never got that going, I was trying to make an over-complex discrete electronics driver and could never debug it. When I started to make the new clock I just used the same mounting block with the coil and HED but with the shorter pendulum the amplitude is 4 x bigger! To begin with I was really just trying out a very simple Arduino pendulum driver I found on the web
using a crude lash-up on a bit of MDF, but it worked rather well so I decided to make it into a clock. So its amplitude is much bigger than is probably ideal, but it is at least controlled, though with what precision I don't know. This time I haven't got round to setting up any precision timing measurement.


Join to automatically receive all group messages.