Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
Much easier and more accurate is to use external Real-Time ICs and just read the Time (and/or Date) from its registers. These ICs (for examples see MAXIM_IC site) generate periodic interrupts to inform the CPU that next period of time (1s or other) has elapsed..
Your 12 MHz oscillator will have some frequency error depending on crystal quality, oscillator circuit design, temperature, aging, etc. That frequency error will cause an equivalent speed-up or slow-down of your clock (if you don't apply some sort of software correction). For example, if your oscillator is actually 12.00012 MHz (that's 0.001% too high, or +10 ppm) then your clock will run 0.001% too fast. Since one day has 86400 seconds, your clock would gain 0.864 seconds per day.
Or maybe you knew all that basic stuff, and were wondering how to compute the actual frequency error of your specific board? Your microprocessor data sheet may say something like this, "if you use this brand crystal and this circuit layout, then the frequency accuracy will be (whatever)".
I think this clock accuracy is important if you are very depedent for critical applications involving timers in the long run. That is if application use synchronization. Again if it is in long-run.
I have already tried building a digital clock using 8051 @ 11.0592MHz. But it use to loose 3 seconds per day. I had so many tasks to do in the ISR like update the time variables, display time on 7-segments, check for key-presses etc. All this plus interrupt overhead may be causing error which makes "RTC" to "non-RTC".
Or it may be due to crystal frequency variation. I had used standard crystal (the one which comes in rounded rectangular steel tube).
A TCXO will improve the accuracy about ten times, but it is also more costly.
If your clock is connected to the power grid you can use the mains frequency
as a reference - provided that the country you live in adjusts the frequency on
a daily basis. During high load hours the frequency drops a little, so during the
night-time it is increased to make up for lost "time".
If you have e.g. a clock radio at home that you never need to set, then your
mains frequency is corrected.
You should use the crystal for short time stabillity and correlate it, in software I
presume, to the mains frequency for long term stability.
How to go about doing it, I leave up to you... :wink:
CMOS, losing 3 seconds per day is -35 ppm. That is typical accuracy of an inexpensive crystal oscillator (XO).
For example: https://www.ecliptek.com/crystals/ec/
For more money you can add temperature compensation (TCXO), or put the oscillator inside a temperature-controlled chamber (OCXO), or discipline the oscillator using a more accurate frequency reference (Rubidium, Cesium, Hydrogen Maser, etc). Performance and cost increase with each step.
Less expensive methods include synchronizing to somebody else's clock, such as the 50/60 Hz AC mains, WWVB, GPS, or a network time (NTP) server.
Your 12 MHz oscillator will have some frequency error depending on crystal quality, oscillator circuit design, temperature, aging, etc. That frequency error will cause an equivalent speed-up or slow-down of your clock (if you don't apply some sort of software correction). For example, if your oscillator is actually 12.00012 MHz (that's 0.001% too high, or +10 ppm) then your clock will run 0.001% too fast. Since one day hasan 86400 seconds, your clock would gain 0.864 seconds per day.
Or maybe you knew all that basic stuff, and were wondering how to compute the actual frequency error of your specific board? Your microprocessor data sheet may say something like this, "if you use this brand crystal and this circuit layout, then the frequency accuracy will be (whatever)".
Though mcu clock with external quartz could be very inaccurate , nothing prevents you to use external oscillator with good stability (if mcu allows to use such) .
But there is other concern - how properly you can prevent your mcu design from malfunction ? Any single power failure or firmware fault will force you to
adjust clock time again .
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
