frequency measurement

[harryr]

how many cycles of a constant frequency sine wave are counted before one determines the frequency of the signal?
one cycle?
how about measuring the time it takes to go from 0 volts
to the first peak and multiplying the time by 4 to get the time for a complete cycle. the frequency is then the reciporical of that number.
for example: if you have the start of a 30 MHz signal and before the first peak the signal changes to 60 MHz and then changes back to 30 MHZ before it reaches its first peak.this repeats again and again and again.
What will the resultant signal look like and what is the
(measured) frequency.
Is it a 30MHZ signal followed by a 60 MHz signal followed be a 30 MHz signal followed by a 60MHz signal and on an on?

 

[VEBill]

If you measure the signal's period (or a reliably known ratio of the period, typically half), then you can calculate the frequency. This inverse technique is useful for very low frequencies where you don't want to sit around all day waiting for sufficient cycles to build up to have a count with sufficient resolution.

So to answer your first question, you could accurately determine the frequency in perhaps as little as one-half of one cycle provided that the signal's amplitude was accurately centered around zero (putting it simply). Any amplitude offset and you'd be off center and inaccurate.

One full cycle would be more accurate (regarding amplitude offset), but this still neglects noise and jitter and so on.

In answering your second question, if a "30 MHz" signal shifts to 60 MHz before the end of the first cycle, then it wasn't really a 30 MHz signal to begin with.

If you keep changing the input signal (at a lower rate), then the counter will keep changing the answer.

Most frequency counters use a zero-crossing detector (or similar) on the input to 'square-up' the signal so that the rest of the box knows nothing about the waveform of the incoming signal.

 

[IRstuff]

Are you describing a 60 MHz half sine running on a 30 MHz period?



TTFN

 

[bmnet]

I think you'll get a frequency reading that's slightly higher than 30MHz but less than 60MHz if you switch between 30MHZ and 60MHz. As VE1BL mentioned above, detecting the frequency normaly uses the zero point crossing, thus a small bit of the cycle will be going slowly (low freq) while the other bit is fast (higher freq)... Combine the overall time and you've got something in between...
An fourier transform operation should however determine the resultant frequency..