mixer buffer why to use LO bufers?

mixer buffer

I have a signal coming out of a frequency divider. it's a square wave going from rail to rail. Normally would we stick this is directly into the mixer LO? it doesn't seem necessary since it is already rail to rail, but why do I see all these LO buffers before a mixer in bunch of papers?

Let's say that I make the LO buffers(I need this anyways). It will be a differential pair with LC tank load(I need this to be a clock resonant buffer, meaning resonating at the LO frequency). So if it's WCDMA transmitter with center frequency of around 1.9GHz, I'll need something like an inductor of 1.1nh and a capacitor of 5.7p. do I need to stick a resistor in parallel? or would just an LC be suffice? this really comes from my lack of experience. Also, how would I control the common mode voltage and and differential swing? I just did a quick simulation and the out goes to vdd.

sorry, i posted this on the AnalogIC section before but got no help, so I'm putting it up here in the RF section again. Thanks.

Re: mixer buffer

The main reason of LO buffer is not only isolation from previous stage, but to provide a proper impedance at the LO mixer input (which indirectly affects mixer isolation).
LC matching network in that place it will work, but is narrow band by definition.

A lot of design engineers use there a pad attenuator to provide good wideband resistive impedance at mixer input. You lose some power through this pad, but generally this is compensated by the high LO power.

Most dividers output insufficent power to drive the LO of say a diode based mixer. many frequency dividers will output say 1 Vpp, while a burly mixer may require anywhere from 7-20 dBm of LO drive. The reverse isolation can keep your IF and RF signals from getting back into your system as well.

As stated a broadband match on the LO port of the mixer (and RF, and IF) really helps to keep the mixer "happy". A common diode based mixer creats a lot of odd harmonics (3rd, 5th, etc). If these reflect back into the mixer it can result in changes to the conversion gain. Similarly the match at the image frequency at the RF port can affect the conversion gain. So especially in broadband applications (i.e. swept LO) it is of paramount importance to either use pads on all ports, provide termination at the image frequency, or use isolators if you can't afford the loss of a pad. Doing so will greatly reduce the amplitude of ripples in both conversion gain, and ripples in your IP3.

To prevent from "Pulling effect"...Because input impedance of the mixer is changed by signal amplitude and that will cause an instantaneous frequency shifting ( e.g. Phase Noise) at VCO..