I've been thinkin' about an experiment...

[Raymond101]

Eric I found that beat sencing on slow tick clocks like torsion & brocot . I found just using a capacitive sensor on the fork and measure the pulse to my frequency counter set for "Per" gives very accurate results in microseconds. Or Hz easy enough to convert to bpm or bph . As 1 hz = sec or ms .
I measure from the furthest point of swing so it's half beat . Ok you will figure why and how . Very accurate on slow tick tocks . The pulse is just on/off so you can count on your arduino. X2 x60
Or what ever time you require.
Microphones are not good if there is ambient noise ie the other 100 clock in the house.
Plus it cheap & good 6 or more place of decimal.
I hope this has helped.

 

[Gyro Gearloose]

Well no joy sensing the escapement. I have a good running Schatz that I serviced some time ago set up and connected to a beat amplifier. The amplifier picks up the impulse and the drop, "tick-tick, tock-tock." Actually they all sound the same, more like "clunk-clunk, clunk-clunk." I'm not sure what I can do with that.

I looked up what I could find on clock and watch timers. Microset uses an optical sensor and a flag on the pendulum for torsion clocks, avoiding the escapement altogether. Not really what I was looking for but it was my original plan. Timetrax appears to use a sound sensor and an algorithm to figure out when the beats occur. Hmmm. Either way it looks like I'll need a third processor in the mix.

Eric

Eric,

The following probably has no bearing on your project, but some clarification of the Timetrax tool seems warranted:

Timetrax is my goto tool for the initial setup of 400 day clocks. I can tell you from experience that I am very familiar with the double "tick-tick | tock-tock" and the "clunk-clunk | clunk-clunk" of the amplified escapement. Any beat amplifier will pick up the impulse and the drop, however the amplitude of the impulse compared to the amplitude of the lock may seem equal to the human ear, but not to the Timetrax unit. Timetrax has a potentiometer that allows one to attenuate the signal to the point where the "tick-tick | tock-tock" becomes "tick-tock". If this were not possible the Timetrax unit could never accurately calculate the BPH.

In timer mode, the beat amplifier is mute, and the impulses from the sound transducer are shown on an LED light. The "tick-tick | tock-tock" appears as a double "blip" of the LED which the eye can easily discern compared to the ear (especially my old ears!). One simply attenuates the signal until only one "blip" of the light is seen at each end of the pendulum's swing. Simple! The BPH is calculated in minutes, and this offers a real fast track to initially regulating the pendulum in a very short time (10 or 15 min.) for the clock to keep good time.

Maybe a photo-cell located near the Timetrax LED could be used in your setup? Or even wire directly into the Timetrax LED circuit?

The Timetrax is a sophisticated instrument and can do much more than just calculate the BPH. If you want to learn more of its capabilities you can download the instruction manual here:

Timetrax Instruction Manual

And by the way, the transducer that clips onto the clock is a piezo electric microphone which detects the vibrations of the escapement, and is not sensitive to ambient sounds. I should know, there's a very noisy original wooden Mayenschein Idle-Tyme rolling ball clock a few feet away from where I work on one side, and a very old Tall Clock on the other side that has a fairly loud ticker.

~ Frank

.

 

[etmb61]

Eric,

The following probably has no bearing on your project, but some clarification of the Timetrax tool seems warranted:

Timetrax is my goto tool for the initial setup of 400 day clocks. I can tell you from experience that I am very familiar with the double "tick-tick | tock-tock" and the "clunk-clunk | clunk-clunk" of the amplified escapement. Any beat amplifier will pick up the impulse and the drop, however the amplitude of the impulse compared to the amplitude of the lock may seem equal to the human ear, but not to the Timetrax unit. Timetrax has a potentiometer that allows one to attenuate the signal to the point where the "tick-tick | tock-tock" becomes "tick-tock". If this were not possible the Timetrax unit could never accurately calculate the BPH.

In timer mode, the beat amplifier is mute, and the impulses from the sound transducer are shown on an LED light. The "tick-tick | tock-tock" appears as a double "blip" of the LED which the eye can easily discern compared to the ear (especially my old ears!). One simply attenuates the signal until only one "blip" of the light is seen at each end of the pendulum's swing. Simple! The BPH is calculated in minutes, and this offers a real fast track to initially regulating the pendulum in a very short time (10 or 15 min.) for the clock to keep good time.

Maybe a photo-cell located near the Timetrax LED could be used in your setup? Or even wire directly into the Timetrax LED circuit?

The Timetrax is a sophisticated instrument and can do much more than just calculate the BPH. If you want to learn more of its capabilities you can download the instruction manual here:

Timetrax Instruction Manual

And BTW the transducer that clips onto the clock is a piezo electric microphone which detects the vibrations of the escapement, and is not sensitive to ambient sounds. I should know, there's a very noisy original wooden Mayenschein Idle-Tyme rolling ball clock a few feet away from where I work on one side, and a very old Tall Clock on the other side that has a fairly loud ticker.

~ Frank

Hi Frank,

Thanks for your explanation. I had read the manual for the Timetrax and believe I understand the tool. What happens inside Timetrax is an analog to digital conversion of the pickup signal. Then a processor discriminates the noise from the ticks using an algorithm based on the gain control. My original assessment of it is correct, but I don't own one. That means I have to make the equivalent of one.

My idea was to use a simple sensing of the escapement drop as a triggering event and to compute timing. I was hoping to reduce my programming load but it appears I need more signal processing just to figure out the ticks.

The pickup I have is a piezo microphone that is normally used with acoustic guitars and violins. It's insensitive to ambient noise but very sensitive to contact vibration.


Eric

 

[etmb61]

Eric I found that beat sencing on slow tick clocks like torsion & brocot . I found just using a capacitive sensor on the fork and measure the pulse to my frequency counter set for "Per" gives very accurate results in microseconds. Or Hz easy enough to convert to bpm or bph . As 1 hz = sec or ms .
I measure from the furthest point of swing so it's half beat . Ok you will figure why and how . Very accurate on slow tick tocks . The pulse is just on/off so you can count on your arduino. X2 x60
Or what ever time you require.
Microphones are not good if there is ambient noise ie the other 100 clock in the house.
Plus it cheap & good 6 or more place of decimal.
I hope this has helped.

Hi Raymond,

Could you elaborate on your sensor?

Eric

 

[Raymond101]

Hi Raymond,

Could you elaborate on your sensor?

Eric

Hi Eric if the pins are steel a simple proximity sensor.
I drawed a cad . Hope you understand my drawing.
Place a small sensor vertical to left or right of the fork pins the swing have the sensor set so it just sees the max swing..
Half beat . X2 your main results will be from your rotation sensor under the pendulum.
If the sensor is set in the middle it gives false results

 

[Raymond101]

Update on some clocks depending on room place the sensor horizontal. Also you can place a thin rod on the suspension rod plate clamp facing backwards out the back . It will not effect the timing as it's sort of count a balance to the fork then can add a sensor to middle position. Plus being external less likely to drop spanner into the works
The timex is great equipment if one can justify the cost . I get very good results with what equipment I have @ hand .also it's a hobby
And my prox sensor is giving me an accuracy of a a few seconds a week.

 

[Raymond101]

Hi Eric just a quick update. I was not so shore how to explain about the sensor pick up so I have added a photo.
The sensor has to be placed in such a way that the swing arm doesn't pass past the middle point of the pickup area .
The drop out has a small gating curve that you may have to allow for in programming.

 

[etmb61]

Test fixture mockup, just for show. The legs supporting the movement platform will need to be long enough to fit the pendulum with the position sensor below it. The winding arbor will be extended through the beam and supported on the near end in a bearing block. An arm will connect the beam to the load cell on the right. The length of the arm will be adjusted to disengage the winding click. That's the plan anyway. Time to start cutting aluminum.



The jury is still out on the escapement sensing issue.

Eric

 

[etmb61]

Here is the current state of my clock sensor array. All of the devices are now being handled by one 8 bit micro controller. Top left is the load cell. Below that is the rotary position Hall sensor with a rare earth magnet to its left. On the far top right is a micro SD card reader/writer. Below that is the micro controller with the environment sensors and real time clock.



I figured out a method of timing the pendulum without listening to the escapement. I can capture the pendulum's extremes with the rotary sensor and measure the time between them. That's the theory anyway.

This is the output that gets written to the SD card. I will lengthen the time between samples once I see how it actually works.


From left to right it shows date, time, the pendulum period in milliseconds, the current position of the magnet over the Hall sensor in degrees (just to show it's working), the difference between the pendulum extremes in degrees, the temperature of my basement, and the load cell reading in some portion of a gram.

My test subject is a good running Schatz 49 that I serviced a little while ago.


Now I'm ready to cut some metal and get it put together.

Eric

 

[Raymond101]

Starting to look really interesting .



Your comment. "The jury is still out on the pendulum". Reminds me of the old film. "The Pit & Pendulum"
Update
Eric I don't know what equipment you have at hand you might be able to fit 2 photo optics into the inspection holes on the back to measure beats.

 

[etmb61]

Test fixture starting to take shape. This may not look like much but the hours are adding up on this pile of aluminum.



Eric

 

[Raymond101]

Looking good I do hope that clock is going to be well fixed down so it does try to flip when the torsion bar is connected.

 

[TQ60]

For the escapement the audio can be used.

Your sine wave image shows the position of the pendulum.

Assuming it is plotted based on time.

Having the sounds plot in same space would allow one to see where in the cycle the escapement action happens then observe how long before re ersal of motion.

One could precisely measure where everything is happening and show beat.

Could add overlay to show exact position in degrees.

 

[MartinM]

If you're trying to get a reading on the actual escapement, how about contact-cementing a paper or aluminum foil flag to the tippy-top of the anchor pin and use optical sensors on the left and rightmost excursions of the pin?

 

[Raymond101]

Eric well how's it going.. ? Must be close to the finishing touches

Raymond

 

[etmb61]

Eric well how's it going.. ? Must be close to the finishing touches

Raymond

Not finishing touches but I am making progress. I have a movement mounted to get the spacing for the parts to go on the winding arbor. Next is the winding extension and the bearing to support it.



I don't own a power cutoff saw so I chewed through the stock for the bearing with a hacksaw, milled it square, and then took a long nap!


I'll get there.

Eric

 

[etmb61]

Here is the winding extension. The beam that presses on the load cell will clamp onto this. Time for another nap.



Eric

 

[etmb61]

And here is the bearing support. I've resisted the urge to make it fancy.



Eric

 

[Raymond101]

And here is the bearing support. I've resisted the urge to make it fancy.

View attachment 751287

Eric

I'm impressed beautiful work. Your are putting a lot of effort in to this project.
When finished, you will have to make a nice wood & glass cabinet for the finished product.
Raymond.

 

[Richard Cedar]

A few years ago, I tried to do something similar, although not quite as ambitious, in an attempt to understand the impact of temperature and humidity on the accuracy of one of my large wooden torsional pendulum clocks.



I used a Raspberry Pi processor attached to temperature and humidity sensors to record the time and ambient conditions. To measure the time between ticks and tocks, I clamped a microphone to the wooden frame close to the escapement. I wrote some very crude signal processing code to detect the sound of the escapement and filter out the background noise. This worked reasonably well but would occasionally either miss a beat or add an erroneous "beat," probably when someone slammed a door. The data was written to a file that I could plot and identify apparent issues that I edited out.

I recorded data for days and even turned the heating vents in the room on and off to force temperature changes. There were definite trends in the clock speed, but I could not find a statistically significant correlation between the clock speed and either temperature or humidity. This surprised me as I use spring steel (straightened pocket watch main springs) as torsional pendulums.

I did not think about trying to correlate the results with the phase of the moon (gravity). I need to see if I still have the data files and, if so, determine if there is enough information to see if I could detect a correlation with the moon phase.

Richard Cedar.
Cedar Clocks
www.CedarClocks.com

 

[Raymond101]

Wow Richard that is a beautiful clock.
I have been messing with a masters nisshin 400 day I got for $20 . It's got more plastic & aluminum than brass . The rotating pendulum has a mass of 257 grams lead balls . Just been timing the swings . But one thing I did notice over the last few days it's very sensitive to seismic vibrations .
My German torsion clock uses brass weights and is not changing.
Maybe different metals /materials react to the earth's changing moods.
I do know that spinning disc's will always produce centrifugal forces so the changing of direction has a breaking force. Over swing.
I have a feeling this may be a problem that will be hard to pin down as mother nature doesn't disclosure her secrets.

 

[etmb61]

Update: Here is the balance beam in place and the load cell for position. The link that will eventually connect the beam to the cell will hang from the beam and press on the cell. The weight of the link will be compensated on the other end of the beam so the cell will only sense the force from the mainspring. (The brass cylinder atop the load cell is just to keep it from falling over while everything is still loose.)



Eric

 

[etmb61]

Update: I'm almost ready to do some real testing. Here the load cell is mounted and calibrated. The readout is in grams. I went with a simple push rod to connect the beam to the cell. The rod has a 90 degree point on each end and the beam and cell each have a 108 degree pointed socket to receive the ends of the push rod. The extra brass bit is to keep it from falling apart. Once the mainspring is wound the push rod will clamped between the beam and load cell.



Next I need to make an adapter to hold the magnet on the bottom of the pendulum and install the test movement and electronics.

Eric

 

[KurtinSA]

What have you done with the click and/or spring? Once wound, do you remove the click so the beam transfers the load/torsion?

Kurt

 

[etmb61]

What have you done with the click and/or spring? Once wound, do you remove the click so the beam transfers the load/torsion?

Kurt

I had thought of removing the click at first. I don't have room to get a screwdriver in there now. The current plan is to just move the tip of the spring off of the click or remove the spring entirely if that's not possible. Then once it's all assembled just move the click clear of the ratchet wheel. I think that will be safe.

Eric

 

[etmb61]

Here I have the click disengaged.


And the clock is running!


Eric

 

[etmb61]

First test with a live pendulum and the Hall effect position sensor failed. The pendulum with just the magnet attached runs fine. With the sensor sitting below it puts on the brakes and the clock eventually stops (at about 327 seconds on the plot below). So the magnetic field is interacting with the electronic sensor enough to put a drag on the pendulum.



Hmmm...

I will try again with a less powerful magnet moved farther away from the sensor and see what happens. If that fails I'll have to make an optical encoder.

Eric

 

[Raymond101]

Eric the pins you used in the board are not copper . Replace the Chinese connector pins with short copper wire .
And that will probably solve your drag problems.
Even the brass colored pins are steel.
Run a magnet over the board and you will see everything is magnetic. Solder the sensor panel directly onto pcb with out pins or socket. .
Also some of those cheap China pcb are magnetic. . Veroboard is all copper

 

[etmb61]

Eric the pins you used in the board are not copper . Replace the Chinese connector pins with short copper wire .
And that will probably solve your drag problems.
Even the brass colored pins are steel.
Run a magnet over the board and you will see everything is magnetic. Solder the sensor panel directly onto pcb with out pins or socket. .
Also some of those cheap China pcb are magnetic. . Veroboard is all copper

The decoupling capacitors and their leads are magnetic. The rest is not.

Eric

 

[Mk2]

Eric, OMG, just discovered this thread!

I've been planning something along almost the exact same lines as yourself the last few years. All thought experiments so far, on how to combine everything. The original design was going to be a perpetual radio beacon, which evolved into a timepiece. Think ET phone home...

The biggest problem I have been trying to find a "compact" or miniature solution to, is the energy source, or pressure, or potential energy. The "stuff" that winds up the main spring. I don't want to use wind or solar. From reading this thread top to bottom, seems like the atmospheric or ambient experiments you've been doing confirm my thinking exactly; it's all about consistency or a reliable source of the "stuff".

People I know who've got the atmos style movements tell me that air pressure changes alone are not enough in the equitorial latitudes, and around the "doldrums". Same goes for temperature variation in humid tropical zones.

The thought experiment solution I have so far is that the energy source has to be huge and the motor or engine (movement) has to be tiny. The numbers I have come up with are that a 1kg weight moving (descending) about 275mm is enough to run a tiny ladies movement like an ETA 2365- where the arbour of the barrel is extended out into a pivot that connects via additional clock style wheels to the weight. The clever bit that needs more thought is how to raise the weight, from changes in ambient conditions. One friend suggested a 'huge' barrel as used in automatic movements instead of a descending mass...

And this all started for me with the energy given out by a cigarette lighter piezo crystal high voltage spark generator, connected via an ignition transformer in reverse to charge a capacitor...

Rgds -frank

 

[Mk2]

I failed to add that it has to run for 400 days with the 1kg weight holding the potential energy. Once it gets to the bottom and can't go any lower, it'll stop. In theory, a 2.75m (9ft drop) would run an ETA2360 variant for ten years. Only problem is mechanical movements need servicing every 5 years or so.

Just as a side thought for your timing measurements- I know already mentioned above, but I've used a laser (pointer) rigged up with a photo transistor for "very" precise edge movement detection in the past. The only thing you need to look out for is diffraction of the laser light on the edge, which can be measured in nano seconds (think school Young's slits experiment). Microscopic dust particles floating in the air become a nuisance as they add errors. The transistor output fed into a schmitt trigger, which then is fed into the counter. And then all you need is an oven controlled crystal timing gate to get the number of counts over a time period, just like a time grapher. You'll be able to achieve excellent averaging over long gate times.

 

[Raymond101]

Mk2 , That sounds like a Frankenstein experiment surely have you tried using levers & pulley this would increase your mass Old school .
Is your ET phone home referring to a GPSDO . I do have 1 for calibration but really over Kill for a torsion clock
as there would be no need for 10 to 14th accuracy

I think optics were mentioned in an earlier post.
The way Eric is doing it now, is working really well . By using the rotation sensor .
The main gain of Eric's setup is to study what are the effecting unknown forces that influence all torsion clocks .

 

[Mk2]

Just had to chuckle at that comment... 10 to 14th accuracy!!! LOL. Very good!

Yeah, imagine using a GPS reference clock source and attempting to adjust your mechanical clock. I think after five minutes I'd just give up, as the teeny tiny variations would become huge when measuring with such a high frequency.

Since a mechanical clock is a device for us mortals to know the time, I'd say close to a second is good enough. Maybe even half a second if you're german and run the railways...

Yeah, Eric's measuring experiments sound like a challenge. Anything magnetic, metallic and in motion usually sets up eddy currents which drag on the moving component. Even the weak magnetic field of our planet causes drag on metal things moving (Aircraft, satellites, bullets, ships...).

Hey, just thought- has anyone ever considered the coriolis forces on torsion clocks? I'm sure it'd affect beat error if moving from say Canada to Argentina.

 

[Raymond101]

O yes moving a clock half way across the planet does foe sure effect a clock any clock . As all material become changed over time in the same zone .but I'm not sure that's will effect Eric's tests .
I have got a torsion clock from the USA and it had been fully service and was sent to me 3000 miles away and it was running way out fast.
Anyway welcome to the party .
We are waiting on Eric to show some new progress.
O that thing about the weight . 1kg to wind your clock had me laughing as well .I could just visualize a clock on a high shelf and a bucket with flowers attached to 2mtr rope . & telling the wife it was green energy

 

[Mk2]

Yes yes, but if no one is allowed to touch the clock, ever, how does it get wound up? That's what I'm trying to solve. Step in Atmos... But different. 90milli bars of pressure change really isn't much.

Energy storage and how to harvest/capture it, then turn it into either spring force or raise a weight. No solar. No wind, so it'll work inside a centrally heated house.

 

[Raymond101]

Yes yes, but if no one is allowed to touch the clock, ever, how does it get wound up? That's what I'm trying to solve. Step in Atmos... But different. 90milli bars of pressure change really isn't much.

Energy storage and how to harvest/capture it, then turn it into either spring force or raise a weight. No solar. No wind, so it'll work inside a centrally heated house.

I suggest you watch

It's a simple idea & 90 millibar is more than adequate. At least watch the video.
Raymond.

 

[Mk2]

Yeah, I've seen that before.

I'm trying to find the thread on the UK forum where problems with it were discussed in detail a few years ago. Works great for europe or N america where the weather is very cyclical.

What I mentioned above, where if you live in the tropics or in the zones around 30 degrees latitude, it just doesn't work. The variation in air pressure for MONTHS on end is minimal, like perhaps only 5-15mbars weekly variation (nothing really). You get more room pressure from closing a door. And places like the canary islands (Tenerife) where I've stayed for a long time years ago, we had a good, accurate garden barometer, weather station thing. I used to think it was broken. Both temp and pressure hardly moved for months. "Tap tap tap" every morning- no change yet again...? Think climate rather than weather.

 

[Raymond101]

Yes I live in Israel and my barometer does not move more than 10 millibar plus/minus max. we are only about 800 miles from the equator. But I don't own an Atmos. All my pendulum clocks I had to change the weight & length.
Both my torsion clocks now working within seconds a month were both modded as they ran mad fast if set to the book.
Your location description says your in Reading UK.
Also the magnetic north has changed by over 45 mile a yrs plus the earth's rotation has increased. The core has also changed. Not sure how this effects every thing but it does . Mother nature is playing with time .

 

[Mk2]

Yeah, i reckon that the Atmos idea was dreamt up one day by a bored swiss clock/watchmaker, in mid winter storms on a dark evening. He probably noticed the pressure was rising and falling by a huge amount (as it does in northern latitudes), and thought "i could harness that energy change...".

Had clock makers originated in more tropical zones, we'd probably not have such an old invention today! Or maybe it'd be powered by something else instead of changes in air pressure.