Crystal oscillators for MECHANICAL clocks : can they be bought ??

[Vince]

Hi gents,

Old man gave me his heater thermostat to repair. Brand : " Theben ". See pictures below.

It's an old fashion mechanical clock but the thermostat part of it is actually all electronic. The clock is powered by the mains but mains is not used as a time base. Instead, it powers a crystal oscillator that drives the solenoid inside the mechanical clock, and kicks it every second.

Old man complained that the clock would run irregularly : would speed up, slow down, seed up... weird.

I drew the schematic of the oscillator circuitry, below.

As simple as it can get :  a single diode to rectify the 240V mains. An 18K resistor in series with live, another on neutral, so all in all the total current in the clock circuitry is limited to 10mA or so. Voltage regulation is done with a.... a red LEd, forward biased. An electrolytic cap to filter the ripple, a smaller 100nF ceramic cap, and the back-up battery. LED regulates at about 1,7Volt. Battery is Lithium, 1,2Volt.

Battery was dead, 0,28V and leaky. Corroded a coupled traces, but not too bad, traces were still OK. Refurbished them.

Anyway. At first, I removed the battery, then the oscilator was working fine... so I thought OK, the dead battery was pulling the power supply too low to allow the oscillator to run properly, hence the irregular ticking. I didn't know any better, never worked on these things before.

So, I ordered a new battery, fitted it and... now the oscillator will NOT run at all... turns out it's now SHORTED, a solid SHORT ! 

Eh ?! So now the theory must be revisited... maybe it's more like it's the oscillator that killed the battery, not the other way around...

Anyway, this oscillator looks very weird, see pictures. Exotic looking package, with the crystal sitting on top of it.

I have NOT IDEA where to find / buy this kind of thing ! 

There is no marking on it, what soever. Well maybe underneath it, I don't know, have not removed it from the board just yet...

It's definitely a specialized bit of kit, made specifically to drive mechanical clocks. the output, two pins, drive directly the solenoid inside the clock mechanism. Well almost directly : there is an electrolytic cap in series with the solenoid. Electrolytic but NOT polarized.

Is there some specialized shop onlioen where one can buy this type of oscillator ? Or maybe one in a different package ? As long it does the job, I can try to make it fit !

since I drew the schematic, I know its pinout, and how the output connects to the clock. Just 2 pins for ground and  V+, and 2 output pins, and 2 more pins for the crystal that sits on top of the oscillator module/ package.

Also took some measurement. Crystal hasno markings on it, but I measured the frequency of the ticking. 1 Hz dead on. OK sounds logical, but well thanks to the gearing inside the clock, it could have been anything really ! But 1,000 Hz it is.

Power supply is ground and +1,7V.   Also must be able to operate at 1,2V because the backup battery is 1,2V....

So if I can'"t find an exact / drop in replacement, maybe I can find something that's electrically compatible, or can be made compatible with minor mods on the circuit board (slightly different supply voltage for example). I am will to go the extra mile if necessary !

So, anyone knows about this type of oscillator and can tell me more about them ? Where to get them ? How much they cost ?

Any help much appreciated, because I know NOTHING about these bloody things, yet I need to find replacement !   


Regards,


Vince

[Vince]

...

[fzabkar]

These are usually 32768 Hz "tuning fork" crystals. The circuitry divides the frequency by 32K to produce a 1Hz clock.

This chip can do it, but it requires a minimum supply of 2.0V:

https://assets.nexperia.com/documents/data-sheet/74AHC1G4215.pdf

[Vince]

But it's just a general purpose digital chip, not a fully integrated oscillator module able to drive the clock mechanism.

[Vince]

Oops my bad, reacted too quick sorry !  It's a 74 series thing but yeah incredible, it does embark an oscillator.

Still, output is standard digital  stuff, not made to drive a mechanical clock.. and of course that higher voltage level, though that could have been accommodated. Adding another diode drop in series with the LED for example, would regulate the supply to 2,4V instead of 1,7. Then as for the backup battery, I could have replaced the original one cell 1,2Volt unit by a two cell 2,4V one and that probably would have been fine.

[RoGeorge]

Yes, they can be bought, search for 32768 Quartz at Mouser or Digikey.

Much simpler to search for a quartz in a defective electronic wristwatch, or to buy any cheap $1 alarm clock from the nearest shop and took its quartz.  The source doesn't mater much.  They are all the same frequency, 32768 Hz.  They can tolerate only very low power, thought I never seen a defective 32768Hz resonator yet.

[fzabkar]

You could probably cannibalise a cheap watch or wall clock and buffer the output with a transistor, if necessary.

[Vince]

Yes, they can be bought, search for 32768 Quartz at Mouser or Digikey.

Much simpler to search for a quartz in a defective electronic wristwatch, or to buy any cheap $1 alarm clock from the nearest shop and took its quartz.  The source doesn't mater much.  They are all the same frequency, 32768 Hz.  They can tolerate only very low power, thought I never seen a defective 32768Hz resonator yet.

I have not reason (at this point at least) to believe that the Crystal is bad. The problem is the oscillator not the crystal. Oscillator is shorting the power supply, so need a new oscillator.. with an output designed to drive a mechanical clock.

[shakalnokturn]

Considering the supply voltages involved the simplest may still be to cannibalise a cheap clock.
They run from 1.5V the output is 1Hz they may even be directly electrically compatible with the 2 pole motor (?) used in your timer.

[Vince]

Considering the supply voltages involved the simplest may still be to cannibalise a cheap clock.
They run from 1.5V the output is 1Hz they may even be directly electrically compatible with the 2 pole motor (?) used in your timer.

How disappointing ! Yeah salvaging an oscillator from another clock might fix the problem but that would not be fun. Was hoping actually UNDERSTAND how these oscillator work !  The output driver I mean, of course. To me it's just so obscure. Was hoping to find a website that sells them, download manufacturer datasheets to see details of how the driver is put together, how to interface it to the clock.. and what's in the clock to begin with ! A solenoid ? A motor ? What kind ? Stepper ? what kind ? Specs ? How to drive it exactly ? etc etc.... I wanted to lean stuff... I learn nothing :-(   If I knew in detail how it worked at least I could maybe DIY my own driver so I can interface a logic level oscillator, all we can find it appears, to a mechanical clock. But no, no info. Hopeless   

I can't believe you can't buy these things ! Makes no sense ! How can people make mechanical clocks if you can't even buy the bloody oscillators from somewhere     Makes - no.- sense 



Anyway.. you seem to imply that the clocks are run by a motor ? I thought maybe it was just a solenoid that would kick the gearing once a second to get it to the next step.
So if it's a motor well at least I learned something...
Stepper motor ? That would make sense, I guess...

The one in my thermostat has only 2 wires. Both are black so I guess somehow it's not polarized ?!

What intrigues me is that as you can see in the schematic, they put a big electrolytic cap in series with the motor, if motor it is ? I don't get it... since when do you put a cap in series with a motor winding to drive it...   It's all so mysterious and confusing...  I hate when I can't understand stuff, so frustrating !!   

We are in 2021, internet rules, you can buy anything, yet you can't even buy a bloody watch oscillator, nor find detailed info on how to drive them, how they work in detail. It's silly. 

[magic]

Specialized components for low cost mass produced items are obscure stuff that often lacks publicly available documentation.

Are you absolutely sure it's the oscillator that is shorted and not the PCB or some other component?
If so, do you remember what its output was? Square or sine, 32768Hz or 1Hz?

As for the motor, I have a vague recollection of taking apart a mechanical quartz clock and finding that it was driven by a solenoid which caused some cog to rotate by one tooth on each activation.

[fzabkar]

Digikey has plenty of 1Hz MEMS oscillators that run off 1.2V sources and have a tiny current draw, but they are BGA devices.

https://www.digikey.com/en/products/filter/crystals-oscillators-resonators/172?s=N4IgbCBcoJYCZRAJgIwFYAMIA0IAOALogKoB2MBA8gGYCyApgIYDOArgE704gECeeXSCBYBjEAF9xQA

[Vince]

Sussed it I think  !

1)

Magic yes you are right ! Short was NOT due to the oscillator or any COMPONENT ! It was due to operator dumbness !
I removed the oscillator in a vague hope to discover markings underneath it.. no luck. But.. I realized that even with the thing off of the board, I STILL had a short, and that the oscillator, stand alone, was NOT shorted at all !    Pulled the filter cap again... STILL shorted. Pulled the red LED / voltage regulator.. STILL shorted. Then I was left with nothing on the board but the battery ?! Then.. OH MY GOD !!!!   
The tracks under the battery, that got damaged by the old leaky / dead battery. Refurbished them but forgot to insulate them afterwards ! Battery cell is all metal, no insulation around it whatsoever. The bottom of the battery, in contact with the PCB, was the positive side. The refurbished / bare track that ran under the battery was.. GROUND ! 
Cleaned that area, put a bit of sticky label under the battery, problem solved, oscillator / clock runs just fine now ! 

How stupid, can't believe it ! Well I don't mind, I learned something, in the human factor department rather than the EE Dept but useful all the same ! 

Anyway, glad I finally sussed it, I don't like mysteries 


2)

Then I probed the output of one of the two outputs versus ground, to see what it looked like.  It looks like a .... boring square wave, 50% duty cycle, excursion from ground to 1,2V the supply voltage. Hmmm.... so very much like these chips fzabkar linked to, which are readily available...
But what's with this series capacitor !!!  So, I probed again, this time probed the two outputs directly. Thermostat is powered only by the battery so it's floating, I can probe however I want with the scope... So put the ground lead of the scope probe on one output, and the tip of the probe on the other output. See result below. Yes, I can't conceive a repair thread without at least a scope screen capture, that's how I am 

See ? We got sharp pulses, 1second apart obviously. One positive going, next one negative going  etc etc. From ground up to about 1,75V, so higher than the 1,2 supply, not sure what the deal is.
But at least I now understand a bit better what this series cap is doing !  The oscillator output is just a mundane square wave, and the capacitor is there to act as a  " differentiator " to convert the square wave into pulses. The size of the cap is determined I guess so as to limit current / power consumption : stores juuust enough energy to be able to give a pat in the back of the solenoid in the clock, and nothing more.

OK, so I managed to learn something anyway ! 


Thanks for stopping by !

[shakalnokturn]

I think some clocks use solenoids - an electromagnet only pulling a ferrous piece - their driving pulse always has the same polarity.

Because of the capacitor between the timing circuit and coil in this case as in other clock designs I'd expect a very basic motor design:

Both poles of the electromagnet are used to flip a circular magnet, a mechanical arrangement ensures that the magnet will only rotate in the wanted direction.
The driving pulse must change polarity. The series capacitor avoids the use of a H bridge output and allows for similar voltage swing.

[magic]

Digikey has plenty of 1Hz MEMS oscillators that run off 1.2V sources and have a tiny current draw, but they are BGA devices.

Umh, have they caught up with quartz in terms of frequency accuracy and stability?

Reminder: 1ppm frequency error means half minute off per year.

[Vince]

Because of the capacitor between the timing circuit and coil in this case as in other clock designs I'd expect a very basic motor design:
Both poles of the electromagnet are used to flip a circular magnet, a mechanical arrangement ensures that the magnet will only rotate in the wanted direction.
The driving pulse must change polarity. The series capacitor avoids the use of a H bridge output and allows for similar voltage swing.

I am impressed !   

Yes.. had a close look at the clock internals.. all the gears are stuffed in a clear plastic housing, quite handy.
Can't take  picture to show you as it's too small and obscured partially by gears but... yes, you are right, it's not a solenoid it's a motor !
I can see it clearly. Rotor is about 10mm in diameter and similar height. dull dark grey in appearance... looks like ferrite. Permanent magnet I guess. I can see it rotate half a turn with every beat of the oscillator.