Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
It is the project that have to drive the choice of a mixer with a certain linearity carachteristic.
When i say project i mean the whole project and the evaluation includes cost, how easy you can find it, if it is best to design it or to buy it...
there are some known-good topologies and some guidelines could be given in general.
Diode mixers are very linear and simple to build. The trade-offs with diode mixers are in chosing a diode for the application. Low barrier Schotky diodes require less LO pumping power to achive low conversion loss, but they are most nonlinear. High barier diods are most linear but LO pumping power is higher. Medium barier are somwhere in between.
FET resistive mixer could achieve almost linear operation, the transistor is operating in so called triode - linear region. Conversion loss is in line with diode mixers.
FET gate mixers could have significant conversion gain, but they will be most nonlinear comparing with other two.
Hope this helps. Much more details are available from the book Microwave Mixers from Steve Mass
Typical requirements of ETS300-086 Standard for a VHF or UHF mobile radio's receiver:
(1) sensitivity of -113dBm (0dBuV EMF) at 10dB S/N;
(2) Third order Intermodulation Distortion of -70dB min ( with 2 interfering signals detuned +25kHz and +50kHz (and/or -25 and -50kHz) from the receiver's nominal frequency Fin.)
With these interfering signals the frequency of one of the worst case IM3 products is, Fim3= 2*(Fin+25)-(Fin+50)=Fin, exactly on the receiver nominal frequency.
Let suppose we have an ideal (lossless and linear) RF filter before the mixer and no RF amplifier. Let say also, the sensitivity of the receiver is at the required Pmin=-113dBm, the receiver tract after the mixer is ideal too (noiseless and absolutely linear). So, we need our mixer alone to meet requirement above.
Let see, what the requirement of ETS300-086 means:
Our mixer, whith two interfering signals applied, each 70dB larger than nominal sensitivity (or Pin= -43dBm), must generate IM3 prodicts at the noise level of the receiver, which is Pn= Pmin - SN= -113dBm - 10dB = -123dBm. Or IM3's of the mixer must be at least IM3= Pin-Pn= -43 -(-123)=80dB, to meet the requirement. It means the required minimum input IP3 of our mixer must be equal to:
IP3=0.5*IM3+Pin= 0.5*80+(-43)=-3dBm. (A)
As you can see in the equation (A) , the noise level is not present. It means, the IM3 and IP3 can be measured without knowing the noise factor. We need to apply two signals with equal (reasonably low) amplitude and look at the level of generated IM3 products and calculate then IP3.
Of cource, in a real receiver, you probably will have an RF amplifier and a filter before the mixer, so you must know the gain from the antenna to the mixer input, the noise factor of the receiver, taken on the mixer input and the noise factor of cascaded RF amplifier and filter to make proper calculations.
In any case, everything is system dependand. Standard requirements and published data of the products, similar to yours design, are a good start point. There is no universal receipt, you must make some compromises in every project - an oversized in terms of IP3 system will be more expensive and probably more power hungry. The initial system level design is very important - it must be made carefully. Errors or wrong decisions made on the initial system level design usually are not easy to be corrected in later stages of the design.
You may downoad Agilent AppCAD and play a litle bit the NoiseCalc inside.