Can someone give me a rule of thumb of choosing Oscilloscope with appropriate bandwidth for the Jitter analysis of clock signals? What device is highly preferable: spectrum analyzers, frequency counters, real time scopes Or Sampling scope with clock recovery?? Which is highly preferable of getting accurate jitters measurements? I need to measure between 10 Mhz - 10 Ghz signals?
I'd consider investing my time and money in a phase locked loop (PLL) and VCO circuit - use a long time constant and build a very steady circuit - not hard to achieve but will save you a lot of money. Input the signal you wish to measure jitter on and look at the control signal to the VCO before it gets filtered - if there is a lot of jitter it will show up here and there is enough "information" in that signal to calculate jitter.
I hate spending lots of money on one-off equipment if at all possible - that's the motive for my answer.
To choose an Oscilloscope you would need the highest frequency you will encounter in your measurements and the you can apply the Nyquist criteria (fs > 2 x fmax ). Jitter measurements aren't really affected by the scope, since jitter isn't an increase in frequency it's a movement of the signal in time.
Since you need to measure signals up to 10 GHz you would need a scope of 20Ghz which is a very high bandwidth scope and will probably be a very expensive scope, I'm currently working with a 15GHz one and that is already a "good" cost. What signal are you trying to analyze ? RF, Serial Data, ... ?
Some very usefull Application notes:
Don't know if needed but here: Jitter—Understanding it, Measuring It, Eliminating It Part 1: Jitter Fundamentals": http://www.highfrequencyelectronics.com/Archives/Apr04/HFE0404_Hancock.pdf
"Evaluating Oscilloscope Bandwidths for Your Application": http://cp.literature.agilent.com/litweb/pdf/5989-5733EN.pdf