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.
Unless you use an R/C oscillator at that speed you will definitely be precise enough. That will give 1MIPS, so every 1,000,000 instructions will be one second.
If you are trying to time something within a second (really really slow) you may consider using an external 32.767Khz oscillator like a watch crystal on your TMR1 and have it interrupt every time it overflows. Or, use a RTC (look up the MAX1337) and have it assert one of its interrupt pins every second on an alarm.
Trying to count 1,000,000 cycles accurately with C or ASM is difficult because C introduces extra code you didn't intend to be there and ASM requires that you figure out how to do 24-bit or 32-bit math using 8-bit registers.
Crsytals give better accuracy. If you are using any crystal, try to use more accurate as possible with the values for better accuracy. Mean if you use 4 mhz, use a crystal with 4.000mhz marked on it. Good luck
one second over what period of time? 1 second over an hour / 24 hours is nothing. 1 second over a month is doable. 1 second over a year is pretty much out of the possibilities for most of us.
Quartz crystals provide u d best accuracy
Check out the ppm rating while choosing ur crystal
Lowest ppm offer better stability
I have a small question.
Suppose, i have a external 32.768KHz crystal connected.How can i maintain accurate timing if i want to run a normal clock?
I havn't worked much with this crystal so i don't have any idea behind using this crystal for RTC.
1 second over a month equates to less than 1ppm, and I don't think you will find any affordable crystals at that kind of precision.
1 minute over a month is about 20ppm and that's quite doable with regular crystals and no thermal compensation.
if you really need 1 second / month type precision, the least expensive solution is likely to get a high precision watch. Citizen I think used to produce a watch that they claim is good for less than 12 seconds a year.
I don't think you have any choice to use achieve high precision timing accuracy (let say 1ppm) without calibration and without temperature compensation if the device is to be exposed to large temperature variations.
I once saw a design using a silab DDS that uses a to220 transistor mounted to the oscillator as a heater to maintain temperature. Pretty cleaver if you ask me.
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.
