Magnet Material for Free Pendulum Clock

[John PurdyParticipant @johnpurdy78347]

In all the free pendulum articles I have seen the material called for for the magnet assembly has been pure iron or mild steel with pure iron preferred. I know that pure iron has better magnetic properties in that when the external magnetic field is removed it decays quicker and doesn't retain a magnetic field. But in the case of the application in the free pendulum clock is there any significant advantage? We're not dealing with magnetic fields that have to switch polarity or turn off in milli seconds, (as in a transformer or solenoid ) the pendulum is slow moving (relatively ). The one disadvantage of mild steel I can see is the possibility of pole pieces and armature becoming slowly magnetized permanently over time which would tend to slow the pendulum down on each swing. What are peoples experiences?

John

[John PurdyParticipant @johnpurdy78347]

[Michael GilliganParticipant @michaelgilligan61133]

Two quick points, if I may, John

1. from your brief description, I suspect that “free pendulum” may be a misnomer … but I am happy to be wrong about that..
2. have a think about the accuracy, expressed as a percentage [or something like parts-per-million] required of a decent timekeeper. Then wonder if it’s worth the risk of using steel, when soft-iron has better behaviour.

MichaelG.

[Martin KyteParticipant @martinkyte99762]

Posted by John Purdy on 05/11/2022 01:29:29:

The one disadvantage of mild steel I can see is the possibility of pole pieces and armature becoming slowly magnetized permanently over time which would tend to slow the pendulum down on each swing. What are peoples experiences?

John

Not entirely sure why a permanent attraction of the magnet assembly and the pendulum would slow it down. On approach the pendulum would speed up and on departure slow down. The two effects should cancel. However if the swing is not exactly over the magnetic centre it will induce a wobble.

regards Martin

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Martin Kyte on 05/11/2022 08:53:02:

.

Not entirely sure why a permanent attraction of the magnet assembly and the pendulum would slow it down. On approach the pendulum would speed up and on departure slow down. The two effects should cancel. However if the swing is not exactly over the magnetic centre it will induce a wobble.

regards Martin

.

Fair comment, Martin … but, if the retained magnetism of the steel is changing slowly with time, isn’t that similar to a change in Gravity ?

MichaelG.

[John HaineParticipant @johnhaine32865]

The magnetic field will not decay any faster in pure iron. But the main problem with any magnetic material, especially if it can get slightly permanently magnetised, is that the attraction may vary as the pendulum swings even ignoring the intended impulsing by a coil. This will perturb the motion and affect timekeeping. With modern neodymium magnets though I don't think a core is needed at all. I have posted here details of a clock that has a 3x2mm neo magnet on the end of the pendulum that swings above a small air cored coil, which both senses position and impulses. My current build will have a slightly larger magnet, 13 x 3mm, with somewhat larger coils and higher current but position sensed by an opto. No need for any magnetic cores so the question is moot.

[BizibilderParticipant @bizibilder]

You can either use mild steel and treat it by heating to bright red and allowing to cool VERY slowly (the old way was to put it in the embers of a fire and leave it overnight to cool – you may have this facility?). Or to use soft iron which can be obtained fron school laboratory suppliers in bars 1/2" diameter and about 8" long quite cheaply. Soft iron flat bar is obtainable but with some difficulty – usually the suppliers are US based and you have to consider the postage etc.

[Swarf, Mostly!Participant @swarfmostly]

The yoke and core of the old 'Post Office relays' (used in electromechanical telephone exchanges) are a potential source of 'soft iron'.  The use of soft iron gives a lower 'reluctance' magnetic circuit, in turn giving more magnetic attraction for the same number of ampere turns of the coil.

Here's a web-site showing what they look(ed) like: P.O. 3000 Relay

I may have one somewhere and will post dimensions if I can find it.

There may be some still available on the surplus market (if the surplus market still exists! ).

Best regards,

Swarf, Mostly!

Edited By Swarf, Mostly! on 05/11/2022 10:41:00

Edited By Swarf, Mostly! on 05/11/2022 10:41:46

[Martin KyteParticipant @martinkyte99762]

Posted by Michael Gilligan on 05/11/2022 09:03:46:

Posted by Martin Kyte on 05/11/2022 08:53:02:

.

Not entirely sure why a permanent attraction of the magnet assembly and the pendulum would slow it down. On approach the pendulum would speed up and on departure slow down. The two effects should cancel. However if the swing is not exactly over the magnetic centre it will induce a wobble.

regards Martin

.

Fair comment, Martin … but, if the retained magnetism of the steel is changing slowly with time, isn’t that similar to a change in Gravity ?

MichaelG.

Yes, which would shorten the period not lengthen it. In practice I feel the higher magnetic hysteresis of mild steel compared to soft iron would lead to a residual magnetic polarisation of the solenoid core which would quickly approach a stable value. After that point there should be no change in period.
However as the residual magnetism is going to be small compared to the active impulse which will change the amplitude and introduce a small circular variation I suspect it can be neglected.

regards Martin

[John HaineParticipant @johnhaine32865]

Sorry to labour the issue but I can't see the point of using a core. The energy required per impulse is tiny – for a clock I'm familiar with it is ~30 microjoules per swing which translates to a very small force. Once you add a ferromagnetic core it can have remanence and unless very carefully designed will have some "salience" which will generate a small varying force on the pendulum that is hard to predict. An air-cored coil which is kept well away from anything magnetic except a small neo magnet avoids this completely.

If you do need a core then IIRC Bateman's clock uses a core made from transformer laminations. This are made to have a very low remanence and being thin they minimise eddy currents that slow down magnetisation and demagnetisation. Impulse is applied through a magnetic blade that swings into an air gap in the core – the blade could be a single lamination.

I did buy a foot of 1.5" swedish iron for making cores once, it is a complete pig to work, like pure aluminium but harder and tougher.

[SillyOldDufferModerator @sillyoldduffer]

For what it's worth, which may not be much, the data collected from my experimental pendulum didn't contain a trend that might have been caused by its' bob becoming magnetised.

The pendulum was a short length of carbon-fibre rod with a mild-steel bob. The bob was energised unconventionally at the far end of the swing by an electromagnet from a dismantled relay, the idea being that the bob would be gently attracted towards a timed electromagnet in hope of reducing impulse shock and not 'dominating the clock'.

I assumed the relay core was built to minimise residual magnetism, but recognised the bob itself might become magnetised.

The pendulum was energised by an Arduino connected to various sensors and arranged to measure short times accurately:

• Second pulses from a GPS unit were used to calibrate the Arduino's crystal oscillator: the Arduino measured how long it thought a GPS pulse was, and the difference revealed the actual frequency of the computer's crystal, including shifts due to temperature, so the Arduino's report of pendulum swing time were dynamically tweaked accurate to within about 10 microseconds.
• Once every 'n' swings the Arduino also measured Temperature, Barometric, and Humidity
• The need for an impulse was detected by measuring the time taken for the pendulum to pass between two adjacent infrared sensors. Pass time is proportional to amplitude, and – to minimise shock – an impulse was only generated when the amplitude fell below a certain value.
• All this data was transmitted after each swing to a RaspberryPi. The RaspberryPi was set up mainly to log the data over long periods (years if necessary), but it was also programmed to time-stamp log entries with accurate internet time, which is always correct to within about 15mS. Thus the log contained time stamps capable of detecting microsecond variations per swing and time-stamps capable of measuring changes of rate over long periods.

The log was transferred to a PC for analysis and graphing. The data showed the pendulum wasn't a good performer, mainly because the Q is low and – worse – my suspension allowed the bob to swing in an ellipse, which it often did! The pendulum was also markedly sensitive to vibration, including cars passing 20 metres away.

The analysis included a search for correlations, which showed my pendulum wasn't effected by temperature, was slightly effected by changing air-pressure, and was distinctly sensitive to humidity. Although carbon-fibre is stable, it seems the matrix that forms it into a rod isn't!

Anyway, my data shows no sign of the bob being effected by being progressively magnetised. The results don't prove residual magnetism is never a problem! My small bob was designed to be impulsed unconventionally by tiny intermittent pulses; maybe magnetism would show up on a bigger clock pulsed conventionally.

Another thought, my clock was tested on a table where the pendulum was aligned North-South with the earth's magnetic field. Would residual magnetism make a difference if the bob were swinging East-West?

I keep meaning to have another go, applying 'lessons learnt'. One way and another I've never got round to it.

Dave

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Martin Kyte on 05/11/2022 10:46:06:

Posted by Michael Gilligan on 05/11/2022 09:03:46:

Posted by Martin Kyte on 05/11/2022 08:53:02:

.

Not entirely sure why a permanent attraction of the magnet assembly and the pendulum would slow it down. […]

.

Fair comment, Martin … but, if the retained magnetism of the steel is changing slowly with time, isn’t that similar to a change in Gravity ?

Yes, which would shorten the period not lengthen it. […]

.

Agreed … but I skipped lightly over that, because the important point I was trying to make is that the polarisation of a steel core is unpredictable, and takes place over a long period of time [which is I regret, an opinion contrary to yours]

John’s approach seems far more elegant, provided it is appropriate to the other specifics of the mechanism.

If John Purdy could identify the clock he is interested in building, then we might have more to go on.

MichaelG.

.

Edit: __ Perhaps illogically, I would presume that someone building a “Free Pendulum” clock would be aiming for timekeeping better than a minute per year.

Edited By Michael Gilligan on 05/11/2022 11:15:02

[Martin KyteParticipant @martinkyte99762]

I’m happy to be shown to be wrong Michael but why is it unpredictable and takes place over a long period?

regards Martin

Edited By Martin Kyte on 05/11/2022 11:19:29

[Michael GilliganParticipant @michaelgilligan61133]

Timothy Taylor expresses my line of thinking more succinctly than I probably could:

[quote]

A piece of a ferromagnetic grade of sreel exposed to/immersed in the magnetic field of a current carrying coil ( electromagnet/ solenoid) varies in response to shutting off the current, depending on its chemical makeup. It can lose all magnetism, or the magnetism can last “ forever”, depending on the tendency of the small ferromagnetic zones/domains ( similar to grains but smaller, within grains) to stay aligned with the imposed field ( magnetically hard) or revert to random orientations ( magnetically soft).

[/quote]

MichaelG.

.

Ref. __ https://www.quora.com/How-long-does-steel-retain-its-magnetism-after-the-current-is-stopped

Edited By Michael Gilligan on 05/11/2022 11:27:11

[BazyleParticipant @bazyle]

Another source of pure iron is wrought iron, Not the modern kind that is just mild steel but the genuine 19th century stuff used in railings and gates. You might find an 'architectural recycling' yard has some and has a broken bit spare. You want the bar part not the cast iron decoration.

[Martin KyteParticipant @martinkyte99762]

Posted by Michael Gilligan on 05/11/2022 11:26:00:

Timothy Taylor expresses my line of thinking more succinctly than I probably could:

[quote]

A piece of a ferromagnetic grade of sreel exposed to/immersed in the magnetic field of a current carrying coil ( electromagnet/ solenoid) varies in response to shutting off the current, depending on its chemical makeup. It can lose all magnetism, or the magnetism can last “ forever”, depending on the tendency of the small ferromagnetic zones/domains ( similar to grains but smaller, within grains) to stay aligned with the imposed field ( magnetically hard) or revert to random orientations ( magnetically soft).

[/quote]

MichaelG.

.

Ref. __ **LINK**

Edited By Michael Gilligan on 05/11/2022 11:27:11

Does that not just say that different steels have their own BH curves. Our core will have it’s own specific curve and will be magnetised accordingly in a predictable way. I don’t really see that there is going to be much change to the residual magnetic field after the first time round the loop?

regards Martin

[Michael GilliganParticipant @michaelgilligan61133]

Sorry, Martin

I can’t see the validity of the final four words here:

“Our core will have it’s own specific curve and will be magnetised accordingly in a predictable way”

… We are at risk of chasing this forever, so let’s just agree to both keep open minds.

MichaelG.

[Martin KyteParticipant @martinkyte99762]

The validity is that any specific piece of iron will have a unique BH curve which will show how it behaves in a changing immersed field. Thus it’s behaviour is predictable. Your Timothy Taylor quote just says that different steels have magnetic properties that vary according to their chemical and crystalline make up.
regards Martin

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Michael Gilligan on 05/11/2022 11:50:00:

… We are at risk of chasing this forever, so let’s just agree to both keep open minds.

.

O.K. one last try, and then I am sticking with that suggestion ^^^

If any specific piece has its own unique BH curve then its behaviour surely cannot be predictable, except by measuring it and assuming that it does not have any ability to suffer change.

MichaelG.

[Martin KyteParticipant @martinkyte99762]

Posted by Michael Gilligan on 05/11/2022 13:00:37:

Posted by Michael Gilligan on 05/11/2022 11:50:00:

… We are at risk of chasing this forever, so let’s just agree to both keep open minds.

.

O.K. one last try, and then I am sticking with that suggestion ^^^

If any specific piece has its own unique BH curve then its behaviour surely cannot be predictable, except by measuring it and assuming that it does not have any ability to suffer change.

MichaelG.

Your Taylor quote says that the properties are dependent on composition. The composition of whatever piece is chosen for the clock will not change and thus the magnetic properties will be fixed. Knowledge of the exact form of the BH curve is not necessary merely knowing that it will be unique to that piece and remain constant is enough to be confident that the magnetic behaviour will behave in a predictable or repeatable way rather than in a random fashion.

By predictable I mean it will behave in the same way from the same initial conditions and exposure to the same immersed field rather than it being calculable beforehand. Maybe that is where the misunderstanding is occurring.

regards Martin

[duncan webster 1Participant @duncanwebster1]

My clock has worked fine with mild steel on pendulum and mild steel core in electromagnet. There is no evidence of residual magnetism if I wave a bit of shim steel at it. If you really want soft iron how about the laminations from an old transformer

[Michael GilliganParticipant @michaelgilligan61133]

For the benefit of anyone who didn’t bother to do the calculation:

It is widely agreed that there are 31,557, 600 seconds in an average sort of year:

So if we assume that a ‘good’ timekeeper can manage a maximum error of +/- one minute per year,

that’s +/- one part in 525,960 … which is considerably better than most of our ‘engineering’

Just a thought worth pondering

MichaelG.

[SillyOldDufferModerator @sillyoldduffer]

I feel an experiment coming on. Would this work?

Set up a pendulum with a mild-steel bob swinging across one of John's air-cored coils. If the bob is magnetised, it will generate an voltage in the coil that can be measured and noted. This voltage is proportional to the degree of magnetism in the steel, whatever it happens to be.

Then the air-coil can be energised to impulse the bob and the pendulum left running funder power or few days.

After the power is cut, immediately remeasure the voltage generated by the bob swinging across the coil. If the bob's residual magnetism has altered, the second voltage should be different, probably higher.

The voltmeter needs to be high-impedance, perhaps an oscilloscope.

Dave

[John HaineParticipant @johnhaine32865]

Experiment is always good. However even with a neo magnet very close to the coil the voltage is pretty small, I suspect it would be very hard to measure anything. Better perhaps to use a hall probe to see if there's any field then pulse the coil a few times and see if the field has changed.

[peak4Participant @peak4]

For those with an interest in electronic pendulum clocks, Light Straw hosts a page/chapter containing all sorts of interesting stuff
https://www.lightstraw.uk/gpo/clocksystems/pages/docindex.html
Every line on this document index page takes you to the relevant TI. (Technical Instruction)

Light Straw replaced Battleship Grey as the paint colour for many items in old GPO and later BT telephone exchanges.

The common Type 36 pendulum clock, as used in Strowger exchanges etc, was expected to be accurate to less than 8 seconds per week.
https://www.lightstraw.uk/gpo/clocksystems/pages/b1001.html 
In practice, the one I had hanging in my previous house, still to be installed here, was probably good to a minute or two a year.
I almost only ever altered the time on it for BST changes.

There is also documentation on 3000 type relays, and their identification and adjustment. B5144
I never did do the relay bashing course, as it was just before my time as a construction engineer, rather than maintenance.
https://www.lightstraw.uk/gpo/clocksystems/pages/b5144.html

Bill

Edited By peak4 on 05/11/2022 22:01:54

[Author]

Posts