fire station clock revival

[JackOfVA]

The telephone line synchronization mechanism is almost certainly a higher voltage / current limited arrangement and would be emulated with a constant current driver mechanism, instead of a low voltage supply where the voltage is set to deliver the rated current based on the coil resistance.

The reason for this is the current rise time. In a simple series resistor inductor arrangement, the time constant t = L/R where L is the inductance and R is the resistance. The time constant determines how fast the current rises in the series circuit when applied from a 0 point. As R increases, t decreases and hence the current builds up faster which is almost certainly a good thing in a mechanical device of this type.

If L = 1H, to put a number to the equation, and R=17 ohms, one time constant = 1/17= 58ms. At one time constant, the current is 63% of the steady state value.

In a typical telephone line system, the DC supply is -48V (in the US, at least. Some EU countries were -54V if I remember correctly.) To limit the current to the rated value, an additional series resistor is used. Plus, of course, the series resistance of the telephone line itself.

Taking our example with 14 ohms coil resistance and 220 mA to take the mid-point of the 170-270 mA values, the total resistance required with -48V DC source is R=48/0.220 = 218 ohms. This resistance would be the sum of the coil resistance, the phone line resistance and a current limiting resistor at the telephone central office.

But look what happens to the time constant; t = 1/218 = 4.5 ms, more about 12 times faster. Hence, the magnetic solenoid pull will happen much faster.

There are other reasons telephone line circuits are configured this way, including limiting short circuit current, and at least in the US, 220 mA would be considered definitely on the high side for maximum sourced current over a normal pair.

All this being said, it is possible to emulate this high voltage driving circuit with a lower voltage and a constant current driver device. The constant current driver can be as simple as a 3-pin fixed voltage regulator -- most data sheets for the 78xx series regulators include a circuit demonstrating how to make a constant current regulator.

[Simon]

surely the rise time will go down if there is less resistance in the circuit as that will just burn up current and limit it to the magnetic circuit. I could put a series resistor in to limit inrush current a bit and compensate for the slightly higher voltage, or I can just use 2 AA's instead but I'm not sure the mosfet will fire as that's putting the gate voltage a bit low, or I can tap off the second battery and use the 3rd only for the MCU.

I want to avoid any regulators as the quiescent current will kill off the battery life, with impulses of 100mS or less that gives me just over a month on 1A of battery.

Yes I could use counters etc but I'm not well versed on "discrete digital" and I'd like to have some practice on MCU's plus it will make it more flexible when I want to change the timings

[SeanB]

IBM made the original system ( before they did computers they did time clocks) and used a 6V battery made from large carbon cells ( the old PO local battery cells that were used in party line phones) along with a series resistor in each clock to keep the current corract. All clocks are in parallel, with a resistor for each one. You just do the same, and do not use a catch diode, but use a 24V zener across the transistor ( 400mW will do) to protect it, you need a fast collapse of the current in the coil. 100mS drive to the coil is fine, they normally had a pulse up to 1S as drive, but do work with smaller ones. Just add a series resistor with it to get the correct current with a 5V drive.

[MartinX]

When I saw this I thought exactly what grumpydoc said in his post and I thought of exactly the same circuits, and I would add one 4538 in the end so you could adjust the output pulse length to whatever the clock needed.

There is no way you could calculate what the clock needs just knowing the data of the coil, the mechanical movement involved needs more investigation. The best way to decide witch pulse length to use is to drive the clock with a pulse generator and a transistor, use the lowest battery voltage you are aiming for and see how short the pulse is when it stops working, then while looking at the mechanism working increase the pulse length until you are satisfied the clock works reliably, if you have no real data telling you what the clock needs this is the best you can do I think.

[SeanB]

Remember the master clock used a pendulum about 50cm long which took 30s to make a swing, and which drove a contact as the pawl engaged to drive the excapement. Not much more, and pulse width is not terribly critical, it just has to occur long enough that the coil reaches a steady state current. 100mS will be enough, and will not do much to the battery supply over a shorter pulse, You have a sub 1% duty cycle, and the battery will essentially have close to shelf life.

[PA0PBZ]

Remember the master clock used a pendulum about 50cm long which took 30s to make a swing

Isn't 50 cm more like 1.4 seconds?

[SeanB]

Mechanical divider( gears) ran the contacts, gives a smaller clock and you do not need a grandfather sized cabinet for the master clock. That also ran off the electrical pulse output, and had a switch to either add pulses or stop pulsing to compensate for errors, generally they would add or subtract one or 2 units a week when correctly adjusted ( invar pendulum which has temperature compensation in it) to do fine tuning. When set up correctly all keep perfect time.

[Simon]

Why should I not use a back e.m.f. diode ? I'm using a 100V mosfet so maybe I can even do away with the zenner ?, I don't know what effect the inrush current to the coil will make but with 100mS pulses it takes (using ohms law on just the resistance) 0.8mA for the coil so for every mA of battery that is 40 days, not that long really, what capacity is a standard AA battery ?

[SeanB]

AA cell starts at 500mAH for a rechargeable cell, a typical AA alkaline is 2AH at best, so you will get a lot of operating time out of them, probably a year or more. The micro will use more in running the clock.

[Simon]

I thought a micro on a clock crystal used micro-amps, the coil thought will rack up current surely

[SeanB]

The clock takes most of the power in a sleeping microcontroller. If it is executing then the power consumed is higher. If you write the code so the micro is using a internal counter to count clock cycles and wakes on interrupt when overflow occurs then the power used will be roughly 2x the oscillator power ( cmos dividers generally use less power the slower the clock, they are basically acting as a switched capacitor filter simulating a resistor across the power rails) while sleeping. That will be the major factor in battery life, even though it is a milliwatt draw it adds up with time.

[Simon]

So I should keep the micro asleep as long as possible and the counter trip can wake it and produce the pulse - providing i use an interrupt.

[SeanB]

Yes. Unclocked and depowered silicon uses a lot less power than powered and switching silicon.

[Mr Smiley]

It's all those transitions from 0 to 1 and 1 to 0 that use the power.

The slower the clock, the less transitions, the less power.

 

[Simon]

that was the idea of a 32 KHz clock ?

[vk6zgo]

A friend of my dad's has asked me to knock up a circuit for him that generates a pulse every 30 seconds to tick the mechanism over. now there is some help on the back in the form of a label that states that the clock coil should be fed between 170 and 270 mA. But I have no idea how long the tick should take.

I'm currently running on 100mS, it's giving a healthy clonk every 30 seconds-ish (it will take hours if not days of running to notice any drift, luckily the instruction time of a pic at 32.768 KHz divides into 30s with a whole number so in theory I can get it bang on). The coil resistance is about 14 ohms that for 270mA dictates 3V but as I'm driving from 4.5-4.8V and the voltage at 170mA is 2.38 I guess i can use "PWM theory" and half my pulse time from the ideal at the voltage that it would otherwise run at.

I'm hoping to make it battery powered for him with 3 batteries, with 100mS pulse that is 0.8mA per hour of ticks just for the clock. I don't know the capacity of AA batteries but that is 40ish days per A.

Has anyone ever played with this sort of stuff before ?

(oh and if you want one - the driver circuit that is I may be marketing it - who knows who else has "acquired" an old fire station clock).

The idea was that the clocks of which there were many in the same station were all connected in series and a single battery supplied the pulse power that was pulsed by a pendant operated switch. I guess this was in the days when timing circuitry was very expensive so it made sense for the timing to be central

Are you sure about the pulse every 30 seconds?

It seems counter intuitive to do this,as you then have to use a spring mechanism or similar to run the clock during the other 29 seconds.
The Pulsynetic clocks they had at my old Radio Station produced one pulse per second,which drove the slave clocks directly.

The Master clock was a "Grandfather" type,where,if the pendulum swing dropped below a set distance,it operated an electromagnet to give the pendulum a "push"

[IanB]

Are you sure about the pulse every 30 seconds?

It seems counter intuitive to do this,as you then have to use a spring mechanism or similar to run the clock during the other 29 seconds.

The clock doesn't have to run for the other 29 seconds. It just sits there doing nothing. Each pulse advances the minute hand by half a division (and there is no second hand).

I've never seen or heard a master/slave clock in the UK with a 1 second pulse. If you listened to them in a quiet room they went "clonk" every now and then and you saw the minute hand step forwards.

[Simon]

yes, this is purely mechanical and the steps of the mechanism are in 30 seconds so it just needs a pulse to advance it, and yes it is one hell of a clonk compared to a modern clock, you'd not want that going off every second

[SeanB]

http://www-03.ibm.com/ibm/history/exhibits/cc/cc_room.html

A lot of info there......

[Simon]

This one was made by "Gents of Leicester"

[vk6zgo]

Are you sure about the pulse every 30 seconds?

It seems counter intuitive to do this,as you then have to use a spring mechanism or similar to run the clock during the other 29 seconds.

The clock doesn't have to run for the other 29 seconds. It just sits there doing nothing. Each pulse advances the minute hand by half a division (and there is no second hand).

I've never seen or heard a master/slave clock in the UK with a 1 second pulse. If you listened to them in a quiet room they went "clonk" every now and then and you saw the minute hand step forwards.

Yeah,thinking about it,you are right,it didn't have a second hand----ignore my comment.

I must have confused it with the ordinary electric clock at the other end of the transmitter hall.
The Grandfather clock made an even bigger "clonk" when it reset the pendulum.