First Attempt at an Electronic Hipp Clock

My first attempt at making an electronic "Hipp" toggle clock. Definitely a proof of concept. Over the last 24 hours it has lost 1 1/2 minutes compared to the quartz clock in the workshop. I don't expect it to keep very accurate time as the pendulum rod is just a length of 1/4" 303SS and it is in the workshop whose temperature currently fluctuates from 6-7 to 20-22 deg. depending on whether I'm working in there or not. The bob is a 2" x 8" brass tube filled with lead and weighs 4.123 kgms. The electronics are similar to those of Carl Wilson (ME 4694 et al ) with a number of modifications and additions. Rather than optos I'm using Hall effect devices to sense pendulum position. Currently the pendulum is receiving an impulse every 3 min. 30 secs. plus or minus a few seconds. The clock is a standard quartz movement with the Lavet stepper isolated and driven by the pendulum clock electronics. Before anyone says " why didn't you use a micro computer ? (Arduino, BBC microbit )" its because I have a whole box of TTL and CMOS chips collected over the years and its been 35+ years since I did any "C" programming!

It's time now to start on Mk2, carbon fiber tube, revised upper suspension, micro processor controlled and to try John Haines Helmholtz coils idea for supplying the impulse for the pendulum.

John

Neat!

Nice to see it ticking John!

Another fan. Is there such a thing as pendulum porn? I've been ogling the pictures for a good 5 minutes!

Dave

Not in any way a criticism, you have without doubt produced an interesting timepiece but a Hipp clock?. I recognise you have titled it an electronic Hipp clock and omitted the toggle tag but I thought the whole essence of the Hipp Toggle clock was in fact the toggle mechanism of the same name. I realise you have to call it something so do we need a new name for electronically sensed, magnetically impulsed free pendulum timepieces.?

Maybe Pendulatrons ?

I’m sure someone has a better name but the variety of interesting new takes on free pendulum oscillators deserves a class name of their own.

regards Martin

Roger Jelbart called his a Chipp Toggle...

Rogerj.co.uk

Also the Hipp Toggle strikes me as a horrible mechanism which must perturb the pendulum every cycle just so it can detect when the amplitude has decayed enough to give an impulse. Electronics gets us away from that so maybe we can avoid "guilt by association"!

From the Ministry of UTLAs. How about Controlled Amplitude Timepiece, or CAT for short

Edited By duncan webster on 29/01/2023 16:22:42

The whole essence of the Hipp toggle isn't the mechanism itself, but what it is there for - which is to both detect the decaying swing amplitude of the pendulum and to trigger the electromagnetic impulse necessary to restore the swing amplitude, and therefore keep the swing amplitude within defined bounds. This clock uses sensors and logic to achieve the same end result, but with no mechanical interference to the pendulum - as did my own free pendulum clock, serialised in MEW last year (https://www.model-engineer.co.uk/forums/postings.asp?th=171748 ). I suspect that if Hipp had been able to use modern electronic devices in place of his toggle, he would have done so, but in his day, he had little option but to invent a mechanical solution. That being the case, describing this (and my clock for that matter) as "an Electronic Hipp Clock" seems pretty reasonable to me.

Fair comment Tony and I did acknowledge the removal of the word toggle and the inclusion of electronic. However to me the name Hipp toggle does seem fairly wedded to the impulsing mechanism rather than the rest of the clock so I wondered if we could do better. All these clocks are essentially free pendulum timepieces and the similarity in my mind to either the Hipp or the Synchronome is somewhat tenuous. Just an opinion.

regards Martin

An update. After 48 hours it is now 1:48 minutes slow, so in the last 24 hours it has only lost 17 seconds (had been 1:31 after 24 ) so seems to have stabilized. The pendulum is currently swinging through an arc of 5.67 deg. right after it has been given an impulse. Just from measurements the pulse of current to the magnets lasts 96ms. At the moment I'm not about to lug my scope into the workshop to confirm that! The coils for the magnets measure 6.82 and 6.69 ohms and when connected in series draw ~370ma @ 5 volts.Too keep things simple for a start I decided to drive the coils with 5 volts from the logic board power supply. I removed the 5 volt jumper on the L298 module and connected the 5 volts from the logic board to both the 12 and 5 volt input and ground on the L298 module. I wasn't sure if this would be enough to give sufficient impulse to the pendulum as the output from the L298 to the coils is probably in the neighborhood of 4 volts, but it seems to be fine as it is impulsing only once every 3 1/2 minutes to keep the pendulum swinging.

Here's a couple of pics of the logic boards.

A few comments if I may:

• My clock data shows the pendulum takes a while to stabilise from starting, so I let it run for a few minutes before setting the time. The error isn't huge, but it's detectable,
• There's a good chance drift is the dominant error. That is the pendulum isn't adjusted spot on, and runs slightly slow or slightly fast. If the clock gains or loses at a constant rate over several days, the mechanical construction and electronics are probably both good, The problem is how to adjust the period of the pendulum accurately. It's not easy, especially when the pendulum is close to correct, because the effect of temperature and air pressure changes become significant and confuse the issue. The early regulator clocks used in Observatories to calibrate time didn't attempt to get their pendulums spot-on. Instead, they put considerable effort into finding the rate of the clock relative to time from star observations. When the rate was known, they read the dial, and corrected the reading from a rate table. So the dial in the observatory clock might be significantly 'wrong' compared with the ordinary clocks and chronometers it was setting accurately. So let the thing run, and write down how it compares with an accurate time over several days. If the rate of drift is constant, award self a gold star. If the rate varies inconsistently, look for mechanical, electronic, sensor or environmental problems. How good should it be? In their day, when they were best in the world, Shortt clocks were good to about 1 second per year.
• In the great hardware vs software debate we're exact opposites. I'm a fairly competent programmer, but it's 25 years since I last did any serious digital circuitry, and 20 since I made a PCB with Ferric Chloride. To poor old out-of-date me physical circuits are hard work, to be avoided if possible, especially since components became sub-miniature. But hardware logic has some significant advantages, notably speed, and should always be on the table. Suspicion is growing that my clock's input electronics are a step too over-simplified!
• If anything misbehaves erratically on your clock my first suspect would be powering the magnets directly off the logic board. The magnets could easily disrupt the logic by injecting spikes and brown-outs into the chips through the power rail. (You can guess how I know!)

Nice work,

Dave

Edited By SillyOldDuffer on 30/01/2023 11:24:02

Dave
Thanks for your reply.

My thoughts are along your lines in that I plan on just logging the time difference over a number of days and see how consistent it is. The large difference between day one and two might be that on day one I had the heater on for about 2 hours bringing the temp up from its normal ~8deg to 22 for about 2 hours whereas on day 2 there was no heat on. When logging the daily difference I'll have to note whether the heat has been on to see what difference that makes.

As far as driving the magnets from the logic power supply I haven't noticed any problems, there are Schottky snubbing diodes on the output of the L298 module that should take care of any back EMF from the coils.

Your comment about being opposites in the software/hardware debate is quite correct. The last programming I did was in the '80s, and that was in Fortran and Assembler and a small amount of "C" ( still have Kernighan and Ritchies book on programming in "C", they were the ones that developed "C" at Bell Labs for the UNIX operating system ). I've been making PCB since the late '50s, first ones with model airplane dope and a paint brush! Got more sophisticated over the years graduating in the '70s to artwork/cameras,litho film/photo resist. Currently it's software/laser printer/iron on resist, much less hassle. Gave up on ferric chloride quite a few years ago, one time use, disposal problems and corrodes everything in sight. Now use cupric chloride. While not as fast as ferric can be regenerated indefinitely by  bubbling air through it and periodically adjusting the pH and SG with HCl and water . I started out with 500 ml, now have over 2L.
John

Edited By John Purdy on 30/01/2023 19:15:12

Edited By John Purdy on 30/01/2023 19:16:59

Very nice!

I'm working on something similar. A question: I note that you have positioned the coil drive pointing vertically. I've seen that elsewhere at least once, too. I wonder why that is? Wouldn't it be more efficient pointing horizontally? Also, does it have an iron core or is it open?

Thanks.

SK
Thanks. The coils are 1.875" between cheeks and consist of 12 layers, 100 turns per layer of .45mm (.0177 ) wire wound on a soft iron core .460" dia. ending up just under 1"dia. The two coils are spaced 1.000" apart bolted to a soft iron yoke at the bottom, and spaced out from the backboard so that the centre of the two coils is on the centre line of the bob when at rest and spaced off to the left so that the left edge of the 3/4" wide armature on the bottom of the rod is on the centre line of the coils. The coils are activated as the pendulum is 11mm from the vert position in its swing right to left and are deactivated as it reaches the vertical postion . The two coils are connected in series with the start of each winding connected together and the end windings connected to the voltage source. This gives the max mag flux.
As far as positioning them horiz or vert, all the pics I've seen have them where I have them. If horiz they would have to be out to one side beyond the max swing with the impulse given just before max swing rather than at the centre of the swing. How this might affect the timing I don't know. It would have one positive effect though, it would eliminate the downward pull on the pendulum every time they are actuated that exist with them underneath as I have them. The effect is quite obvious in mine as I don't even have to look at the indicator LED on the logic board to know when the coils are driven as there is a very audible "sprong" (for lack of a better description ) sound from the pendulum rod. I put it down to one of two possible things, the rod, suddenly being put in tension vibrates like it has been plucked, or the rod has a very slight bend it it and when put under tension straitens slightly again vibrating like it has been plucked.
John

Mine has a single coil off a 24vAC relay, with mild steel pole pieces to make up the circuit. Triggered about 4mm before centre and off just before centre (timed by the Arduino). The coils are driven off the 5v through a potentiometer, resistance increased until it misses only the occasional pulse. It pulses both ways, and can only miss one way. If I reduce the resistance enough I can make it swing wildly, only being impulse on alternate swings. It is completely silent.

Thank you for the detailed explanation. The second coil was almost hidden in the head-on photo, so thanks for mentioning that too.

I think a horizontal arrangement should be OK in principle even when activated at vertical, except that the amount of energy needed would be higher. One pendulum I saw (from memory) included symmetrical rod "horns" extending towards the coil, and partially encompassed by the coreless coil, so it was operating essentially as a solenoid.

Do you have any comments about the use of the Hall effect sensors? I looked up the sensor you are using, and it looks there's a non-trivial temperature effect on the "operate" and "release" of the sensor about room temperature. I have never used them, but I've been presuming that optical would be superior.

In my view one should avoid any magnetic material in the fixed part of the coil system. There clearly has to be some somewhere to be attracted by the coil, usually on the pendulum. But you want to avoid any spurious forces from remanent magnetism. With any reasonable Q the force needed for impulse is so tiny that a core should not be needed.

I would have thought 5v power to the coils would not be enough to move a long pendulum and weight.

My recolection of the master clocks in telephone exchanges is that 50v was used.

5V is actually plenty for these coils - the issue isn't voltage but amp-turns, which basically determines the magnetic strength. I suspect the Post Office coils used many more turns of much finer wire. Telephone exchanges were based on 50V DC electrics so it was probably easier to make coils to suit the voltage than to mess around with a lower supply voltage. I believe I am currently running mine on 12V but that is massive overkill.

The John Wilding 3/4 second Pendulum Clock which was the inspiration for my design (and in turn, for this one) used a 3V supply (2x "D" cells in series).