help me with accurate phase control of LO

[AtticusWang]

There are two LO signals in my mixer design. One frequency is 2.5G, the other is 5G. Now I need to control the phase of of both LO signals:
when the 2.5GHz signal(sine wave) is crossing the zero point, the 5GHz signal is at its lowest point.
I have tried to use LO buffers(inverter chain)to set delay to the LO signals, but the result varies a lot due to different PVT.

Anyone who can give me ideals about how to accurately control the phase of LO? Thanks a lot!

 

[ucah]

One way that springs to mind is to phase-lock them to the same reference with an at least type II PLL, so that they are in-phase, and then feed the one that should be lagging by 90 degrees to a quadrature hybrid. From that you can get the 90 deg lag.
I am sure plenty other ways exist, like rat-race couplers etc. to give you the required 90 deg lag.
There exists the issue of bandwidth ofcourse, as these networks will stay true to the 90 deg lag over roughly a 10% bandwidth.
If the signals are monochromatic (fixed frequency) then a simple λ/4 line will also do.

 

[AtticusWang]

One way that springs to mind is to phase-lock them to the same reference with an at least type II PLL, so that they are in-phase, and then feed the one that should be lagging by 90 degrees to a quadrature hybrid. From that you can get the 90 deg lag.
I am sure plenty other ways exist, like rat-race couplers etc. to give you the required 90 deg lag.
There exists the issue of bandwidth ofcourse, as these networks will stay true to the 90 deg lag over roughly a 10% bandwidth.
If the signals are monochromatic (fixed frequency) then a simple λ/4 line will also do.

Thank you.
Actually these LOs are used in an active mixer, and LO buffer needed to drive the switch transistors with a parasitic cap of several hundreds fF. There would be phase shift because of the load cap.