question about simulation of mixer in spectreRF

pancho_hideboo

I have been executing noise figure simulation of a mixer after the tutorial in the page **broken link removed** ,

there are two ways to simulating NF . one is PSS + PSP , resulting a plot of NF vs frequency; another is PSS+Pnoise ,resulting a plot of NF vs magnitude of LO. I got 2 different NF , what is the matter ? which one is better ?

the conversion gain from the PSS+PSP simulation is also different from results of other type simulation. what is the matter ?

thanks!!!!

difference between pss and pac

Hi.

Definitions of Voltage Gain are different between PSS/PAC and PSS/PSP.
S21 from PSS/PSP will be same as Vgain from PSS/PAC for your example,
but these will be different from Vgain from PSS/PSP.

As you know, there are many gain definition.
What definition of gain do you need ?

Results of NF can be same value for both PSS/Pnoise and PSS/PSP unless you don't misunderstand NF, NFieee, NFdsb.
In your example. PSS/PAC sweep local power for fixed frequency,
on the other hand PSS/PSP sweep frequncy for fixed local power.

The followings are conclusions.
For NF, there is no remarkable difference between PSS/PAC and PSS/PSP.
For Gain, it depends on your needs.

If you require S-parameters accounting frequency translation effects, you shoud use PSS/PSP or QPSS/QPSP.

spectre rf pxf tutorial

the "conversion gain" i refered is the voltage conversion gain, that is "The voltage conversion gain is the ratio of the RMS voltages of the IF and RF signals" .
Is there any else definition of " voltage conversion gain " ?
which simulation is meant this voltage conversion gain ? PSS+PAC or PSS + PXF or PSS+PSP? what is the difference of them ?

whati is NFieee? is it NFssb ?

i have got a NF of 60dB, maybe there is something wrong in my circuit. working ....

however,thanks!!!

passive mixer input matching

Voltage Gain of PAC and PXF is
VCG=Vout/(Erms/2)

Transfer Function of Pnoise is
VCG=Vout/(Erms/2)

Gain of PSP is
Gain=Vout/Vin

If Rload=Rsource, S21=VCG.

Many people are embarssed with this issue.
https://www.designers-guide.org/Forum/YaBB.pl?num=1122952471/5#5

In Spectre, NF is convetional SSB NF(NFssb).
NFieee is IEEE definition SSB NF.

ADS don't give NFieee.
And NFdsb defintions between ADS and Spectre are different.

See https://www.designers-guide.org/Forum/YaBB.pl?num=1162399761

But NFssb results could be almost same between ADS and Spectre unless simulation setting is wrong.

mixer pxf simulation

thanks for your information!

under the direction of the tutorial of h**p://www.ek.isy.liu.se/courses/tsek26/LAB2_Mixer_Tutorial.pdf, i paid no attention to the impedenc match between the input and output port. i think it is the root of the problem.

but, i'm wondering, the mos is capacitive, what should i do to realize the impedence match ? how much should i set the resistance of the port to ?

having a try ...

spectrerf mixer noise simulation

Output matching is not matter.

In your case, input signal is port(Rs=50ohm).
Although input port is terminated 50ohm directly, impedance which input port sees is not 50ohm since mos gate input is visible through balun.

Your problem is that you use direct plot function of cadence without understanding their definition.
If you calculate gain of your own definition without using direct plot function, you can get same results regardless of PSS/PAC, PSS/PXF, PSS/PSP and even Transient Analysis.

In your example, input matching is broad band matching by 50ohm resistance.
If you want to realize different matching, remove 50ohm resistance and insert reactive matching network instead.

spectrerf mixer

hi， thanks again for your information.
as you have said that ,

pancho_hideboo said:

Voltage Gain of PAC and PXF is
VCG=Vout/(Erms/2)

Transfer Function of Pnoise is
VCG=Vout/(Erms/2)

Gain of PSP is
Gain=Vout/Vin

If Rload=Rsource, S21=VCG.

.

Click to expand...

i thought if i could realize the impedence match between the source port and the input port of the mixer, then i would get S21=Vcg. now i get an approximate result. NF is 22.5dB from PSS/Pnoise, and 25.0dB from PSP; gain is 14.3dB from PSS/pxf, and 11.1dB from PSP. the reason of differenc, i think, is that some mos go out of saturation region. is it right?

Today another problem emerged. when i executed a trasient simulation, i find that the dc value of IFP and of IFN are different, so do the magnitude of small signal of IFP and IFN. The mixer is symmetry, the input signal is also summertry, what causes the IF output unsymmetry?

thanks!

cadence pxf vs pss

In above LAB_2 tutorial, they put 50 Ohm resistor parallel with PORT1 to get the correct power from PORT1. (Otherwise you will get half). But the noise will also raise due to temperature noise from that resistor. I am worried that we don't get the correct NF from that simulation setup. What do you think ?

mixer input impedance simulation

I’m going to make a fully differential passive mixer (ring mixer architecture) . Now I must make a simulation about conversion Gain (in the passive mixer this ideal value is -4 db). I setup the PSS/PAC analysis follow the SpectreRF Application Note and a “LAB2_Mixer_Tutorial”. The problem is the reliability of the results.
I saw in the previous post that the CG is the ratio about Vout and Erms/2.
Now, if I match the RF Port and LO Port to 100 Ohm because there are a differential Port. (I check this in a IEEE article) The port are connected with the ideal_balun. So, the result can be correct? This issue born because if I simulate the net without the matching network I find a CG value upper than ideality value of -4 dB. This is impossible!