Seamless Switching of Two Clock Signals

Found this....

http://www.astrium.eads.net/media/document/astrium_ens2_mdef.pdf

SolarEagle:

Yes, they are parallel clock systems and there is a monitoring device which takes care of the drift in multiple systems to keep them on time. You can see that they have used multiplexer at the input and this is where I am talking about. I think they may average phase change for many cycles and then correct it but problem for me is phase synchronization for the event when we switch the frequency. These people use one more clock source that takes care of the error correction and their frequencies are very close to each other, just having some minor phase error.
I am thinking to vary clock right from 1mHz to 24MHz range anywhere in between.

Have you considered controlling the clock frequency with a VCO? It would never glitch while changing frequency.

WoodwardDL:

Yes, VCO and OVCO are often very good to generate a wide range of frequencies.

SolarEagle also suggested the same thing. In that design external frequency sources are selected or switched and frequency is converted into a proportional DC voltage and then DC voltage is fed to VCO to regenerate the frequency on continuous bases. This scheme is very practical type and very nice one. However cost is very high due to high stability VCO design.

For example if I have excellent two very close packed frequency sources to be switched in a manner FSK signal is generated then there will be little drift in the base or central frequency but VCO need to be very stable to achieve that kind of stability as base frequency likely to shift into another nearby channel.

I think for simple applications VCO can be tried very well.

How about a pair of retrigerable one shots triggered by some appropriate logic combination of your two clock signals. The first one shot would be responsible for the ON time of your clock signal and the other would be responsible for the OFF time.

Pulses that arrive too soon or are too short will be ignored by the one shot combination.

The problem seems to be that there is really insufficient information in the description of the problem. If your two clock source are only fairly close, you will generate a beat frequency that represents the difference. If they are far apart, the slower one can be surpressed by the presence of the faster clock.

You might also consider a different microcontroller that includes a low power sleep state.