Email: Password: Remember Me | Create Account (Free)

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
???
07/22/09 15:00
Read: times


 
#167699 - Think about crystal stability
Responding to: ???'s previous message
Just one thing more about precision and resolution.

You say you need a resolution of 100kHz. I take that to mean that you also want a precision of 100kHz - you want to be able to trust that the last digit is no more than one tick off.

If you have a processor crystal with 50ppm specification, the crystal may give a frequency error of 50Hz/1MHz. That means 125kHz at 2.5GHz.

That means that you should keep an eye on the selection of processor crystal. 50ppm is a quite common value for a processor crystal, and no good for you. You should try and look for a 20ppm or better crystal.

Another thing to think about is that the crystal has both a tolerance (the deviation from the nominal frequency when the crystal is new) and a stability - how much the frequency will change with age and how much it will change because of temperature.

You requirements are a bit marginal when it comes to stability. It can be a bit hard to find a good crystal that will age with grace. And depending on your temperature requirements, you may need to use a crystal oven to keep the crystal at a known temperature.

Your best bet is to go for a precision crystal oscillator - a complete temperature-compensated solution. You can get some of them for maybe $30, but instead get a stability of 1ppm. In the end, they are way easier to use than just a crystal.

List of 14 messages in thread
TopicAuthorDate
Frequency Counter using prescaler            01/01/70 00:00      
   Find prescaler. Then find time period. Then multiply count.            01/01/70 00:00      
      Think about crystal stability            01/01/70 00:00      
         Oscillator candidate            01/01/70 00:00      
         i wish that would become clear to the majority            01/01/70 00:00      
   0.1 %            01/01/70 00:00      
      I'm sorry?            01/01/70 00:00      
         i'm sorry            01/01/70 00:00      
            Clearer now            01/01/70 00:00      
   Dividers increase the error, but...            01/01/70 00:00      
      if extreme precision is required ...            01/01/70 00:00      
         or count more seconds            01/01/70 00:00      
         Remember resolution != precision :)            01/01/70 00:00      
   Excellent - responses thanks so much            01/01/70 00:00      

Back to Subject List