hi manfred,

my comments are:

1. Choosing Q is dependant on your design criteria (bandwidth, gain over band, return loss, etc..). So if your band is very narrow, you can afford to have a higher Q so long as you tolerance for component variation. Offchip components do have variation, which tends to be less with cost! You can readily get 1% off-chip components if you bill-of-material cost is not important... cheaper parts might be 5%. Also your mixer input impedance will vary with process, temperture (and maybe voltage). For an L match the Q is determined by the source & load (mixer Zin) impedances, so to change Q you'll need to modifiy your mixer input for a given source (50R ?)

2. agreed, substantial voltage gain can be achieved in the input match. This might not be a good idea unless you confirm it is toleranced to account for component/device variations. Also linearity may be harder to acheive because of larger voltage swings on the mixer inputs if Q is very high.

cheers
aw

Hi,
As ACWWong say Q depends on your design criteria, so it's better to choose a Q as Q=f0/BW and take into account the component variations, the L-match doesn't give you enough degrees of freedom to design, maybe it's useful to study a pi or T matching network, because having to redesign the transconductance stage to fullfill matching criteria it's not very optimum. For the intrinsic gain of your mixer it depends really on your specs(are you worried about power consumption?,which specs do you have in NF ,linearity,gain?), I find it is a low gain for an active mixer but all goes to your specs and power budget so judge if it is a successful design or not it's up to you, the only thing I can say is that if you are designing as a low noise Mixer to eliminate the LNA take into account Friis, the noise it's determined by the first stage noise sure but it  needs Gain to reduce subsequent stages contribution.
Hope it helps,

Hi manfred,
Sorry for the delay in the answer, I had data of a merged LNA/Mixer in 0.35um CMOS for 2.45GHz ISM band designed by a friend for her masters thesis. She focused in low power consumption, and all the components were on-chip except the balun(differential input):
1.Gain(dBV)=14.95
2.NF_SSB(dB)=16.7
3.IIP3(dBm)=-6.95
4.S11<=-29.9 dB in the whole band
4.Power consumption=0.43mA@3.3 V
If you are focusing in NF then you have achieved a good NF(its SSB or DSB?), I can only recommend you to check the s11 across the band(you didn't mention your operation frequency and your Bandwidth,Q=f0/BW), as a rule of thumb s11 should be less than -10dB in your whole band(check this rule of thumb with other people,this is just one that I use other people maybe disagree with it).
Hope it helps,

Hi manfred,
For me  your numbers look good, but if it's part of a whole GPS receiver better check with system level. About the merged LNA/MIXER of my friend probably call it LNA in the name was a little bit generous, but her objective was completely different than yours she was working in sensor networks that need low-power consumption and NF was not critical(rather it was a system limited by interferences), the weakest number of her block was the gain. I could sent you a copy of her thesis(it analyzes several combos of LNA,LNA+Mixer and 3 different Mixers all analytically and with simulation, in fact she ended with near 20 different possible front-ends) but I think it will be rather useless if you don't understand Spanish(all thesis must be submitted in Spanish and/or Catalan-a regional language-), sent me a PM if you are interested.
Hope it helps,