As I was sitting at my desk not writing the checks I should be, I was staring at my $2.00 battery powered LCD clock counting out the seconds and a thought crossed my mind.

How does this cheaply made device 'know' how long 1 second is?

Grandfather clocks have pendulums, some watches have quartz timing, the clock on the wall may be controlled by the AC cycles.

But what is it in a silicon chip that can tell time?

Bob

Digital watches use a quartz oscillator. A small rock that has a characteristc frequency it vibrates at and they use "divide by" circuitry to count the oscillations down to one second. A desk clock is just a watch with a bigger display

[This message has been edited by gfretwell (edited 03-19-2005).]

Yeah,
Good call Greg,
A clock uses a 38kHz crystal to provide the synchronisation.
The actual crystals themselves are what we would call Rochelle salts, they oscillate at a given frequency when applied with a bias voltage.
One thing I'm short of knowledge on, is how the oscillating frequency is set.
Is it the size, type of crystal, like the size of a capacitor plate vs dielectric?.

I believe the frequency is determined by the physical size of the crystals. When I was in two-way, we would order crystals "cut" to a specific frequency.

Bob,
A battery model has either a crystal oscillator as mentioned, or an integrated circuit that generates a timing signal (such as a digital 555 timer operating in an astable mode).
An a/c powered clock uses a synchronous motor. http://www.howstuffworks.com/digital-clock.htm http://clockhistory.com/telechron/

555s may be OK for something like a microwave timer but they drift too much for a decent clock. I am using one as a spa timer (along with a 14 bit counter) but that is not too critical.

Is there a way to stablize a 555, to be used as a "Tolerable Clock"? - like within 1% tolerance?

I have built a 555 based flasher (drives a 20mA LED), which is in close proximity of a "Kit-Bashed" AC powered Clock (Kit-Bashed being a normal alarm clock w/ digital LED segmented display, removed from original enclosure and placed inside an O-27 scale Caboose).

The normal "½ Second Flashing" of the "Colon" is directed to a Railroad Crossing Crossbuck - which controls alternating flashing LEDs via op-amp.

The 555 based flasher pulse cadence is at least 3x faster, yet does not keep a predictable sync. with the crossbuck.

Just curious if anyone has info.

Scott35

so what do you do if the clock on your computer gains 3 min. every day?

Replace the system board. Not an attractive option but the TOD clock is soldered on the board.

There is a chance this could be fixed with a BIOS flash but that would only be if lots of people have the same problem and it is some strange microcode problem. Check the manufacturer's web site. They will have the latest BIOS update (if any)

[This message has been edited by gfretwell (edited 04-07-2005).]

A few older motherboards had a trimmer capacitor to tweak the RTC oscillator frequency. You can either do it using a frequency counter (recommended), or turn it a bit at a time, and see if the clock gains or loses time. The usual oscillator frequency is 32.768 KHz.

If you have decent soldering skills , you could try replacing the RTC chip (may be socketed!) or oscillator crystal (if not integrated into the RTC).

A software workaround if you are running Windoze XP is to go into the time settings and enable automatic synchronization, and set it to do so every hour or 2, so the time drift becomes less noticeable.

FYI / BTW

Just in case anyone is wondering what the Acronym "RTC" means
(NJwirenut mentions in previous message),
it stands for "Real-Time Clock"

Scott35

And here all these long years I've been asking the opposite question..."where does all the time go?"

Now I'm wondering if it's recycled starting with evaporation when my eyes are closed as I sleep or worse, before my very eyes while I surf the 'net or something. Then it probably seeps into the ground and becomes mixed with the mineral deposits and over time begins to dry or solidify thus kicking off the crystaling process...

.... I know, I'm a litte odd when I think too much...

Bren;

LOL!!!

Nearly fell off the chair when the irony clicked in.

Thanks for the post and the laughs!

Scott35

Whatever you guys have, I want some.......

Email me a sample PLEASE.


LOL


Dnk...

I would say that I was 'dropped on my head as a child' but in all honesty, I jumped. True story. Out of the 2nd story balcony and down into the concrete basement incinerator stairwell and landed on my (14 month old) head - hence, I have a reason for warped thinking. Lack of sleep & having 3 small children who are also warped also probably contributes to my oddity.

As an aside, as I was driving I realized that the word I was struggling with was "crystallization" - doh!



[This message has been edited by BuggabooBren (edited 04-15-2005).]

I like electromechanical pendulum clocks, ( masterclocks FAVAG & Bürk ) which I use at home and have a 24 Volts 1 minute alternating polarity system set up which drives about 30 slaves in my house. accuracy is within ± 6 seconds / month. For the seconds I use a tacho generator which does 1 minute in 58.66 seconds, to keep them in step with the master.

As accurate reference clock I use an 4,194,304 Hz crystal oscillator, 22 divider to provide a very accurate time reference of 1 second, (drift less than 2 secs a month). A stable temperature of the crystal is helpful and a good quality trimming capacitor / reference cap.

I'm still working on an automatic synchronising system which opts for the highest accuracy, OK I admit my hobby is a little out of control and there are not always enough hours in a day to get it all sorted out yet, but I prefer brass gears instead of plastic fantastic.

Scott35, I guess I'm a little confused about what you want to do with your 555 crossbuck ckt. One thing that I tend to do with my 555 ckts,(that I don't believe they show in the data sheets), is put a 1N914 or equiv. in parallel with my discharge resistor(2,6 cathode,7 anode) so that my effective charging resistance can be <= my discharge R for a <= 50% duty cycle. The stability is largely dependent on external components but I never really put 555 and accurate in the same sentence. Of course, this one in my pacemaker is arrrrrrgh.......

Trying to answer iwire’s question “How does this cheaply made device 'know' how long 1 second is?” I want to say something previous to inventing of timer ICs (like the 555 you are talking about) more electrical than electronics (maybe not all the electricians in this forum know about flip-flops and comparators, but I’m sure they are more familiarized with resistors and capacitors):

The “time constant” of an R-C circuit (resistor and capacitor) is the time, in seconds, needed to charge that capacitor through that resistor at a value approximately equal to the 63% of the voltage applied to the circuit. So, for a circuit “to know” how long is 1 second (or any time period) is just a mater of doing the proper calculations at design time; no magic on it. Charging and discharging the capacitor is the way of doing an electrical pendulum (asymmetrical, by the way).

There is a nice little article that may help at http://www.tpub.com/neets/book2/3d.htm; it even has some graphs.


Joe.-

Just looking through the thread again...
You will tend to see a lot of R/C & L/C stuff at lower freqs and Crystal oscillators at the high end. Don't forget the ceramic resonators that are very typical in the low-high KHz. range. They often look like little yellow to orange monolithic caps or like small tubular caps. Any remote control in your house is likely to have them.

If you folks want to play with a versatile timing chip, check into LSI/CSI devices LS7210/11/12. Radio Shack at least used to sell the 7210. I designed a 150Y Device calibrator around one, allowing me to step between 12mS to 372mS in 12mS steps. They have quite a few timer,lock, & PIR chips that would be fun to play with.

As for keeping accurate time, my watch cheats. This Wave Ceptor locks into WWVB at 1AM local time to sync. The bummer is when it hasn't captured the signal for 3 days and displays 12 Noon on Mar.1.
Joe

so what do you do if the clock on your computer gains 3 min. every day?


get some ntp software like automachron for windows or ntpd for linux and just
have your computer's clock sync off of a government or university time server...
It will bypass the bios clock and keep your computer within milliseconds of the
atomic clock in colorado..Windows has its own native time service as well but I have never managed to get it working and assume it is of the same low quality as most of MS's other software

Aside from Xtals and other sources mentioned one has not been dicussed (actually more than one), and that would be your plain ole 60 Hz power supplied by your POCO. Many of the older digital clocks took a sample via a capcitor to a frequency divider and simply divided it by 60 naturally.

dereckbc:

Long term this 50 or 60 Hz will be very accurate, Depending on loading on the overall grid the frequency can vary by up to quite a few seconds a day if not tens of seconds a day.

During peak demand the main frequency may be 59.687 Hz for example, for say a couple of hours when all generators in the system are heavily loaded and it is not warranted to start up extra generating capacity. Generally at night when the grid is lighter loaded the frequency will be raised a little like 60.206 Hz for a period of time to make up for lost cycles generated so that over a 24 hour period a "nil" error is in generated cycles.
60 Hz system . 24h*3600*60 = 5,184,000 cycles
50 Hz system . 24h*3600*50 = 4,320,000 cycles
The mains operated clock usually has a hysteresis motor, or synchronous motor which runs exactly in step with the mains frequency, so some seconds are lost during the day and made up for at night. Over a long period these clocks are very accurate.

Depending upon whereabouts in the world you're located, there's always the option of synchronizing with one of the broadcast atomic clock standards.

Here in the U.K. we have the National Physical Labs. broadcast from Rugby:
http://www.npl.co.uk/time/

Software is available to interface a suitable receiver, for example:
http://www.galsys.co.uk/atomic-clock/msf-rugby-atomic-clock.htm


In the U.S., there is station WWV, Fort Collins, CO, which provides a similar service on several HF transmitters:
http://tf.nist.gov/timefreq/stations/wwv.html


It's also possible to synchronize by internet:

http://tf.nist.gov/timefreq/service/time-computer.html

Or failing that, you can just call (303) 499-7111 and set the clock manually every day!



[This message has been edited by pauluk (edited 01-08-2006).]

I don't think that they normally let the grid frequency drift that much. Under normal conditions the grid operates at plus or minus 0.01 hertz. When time correction is needed for cumulative error correction the system is run at 60.02 or 59.98 hertz.
Don

In the U.K. the grid is specified as 50Hz +/-1%, so 49.5 to 50.5Hz allowable. They make tiny adjustments to the frequency to that in the long term it always averages out to exactly 50Hz to keep synchronous clocks accurate.