The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> Mixer Simulation under Matched Conditions
https://designers-guide.org/forum/YaBB.pl?num=1239997125

Message started by yannis on Apr 17^th, 2009, 12:38pm

Title: Mixer Simulation under Matched Conditions
Post by yannis on Apr 17^th, 2009, 12:38pm

Hello,
I was wondering if there is any way to simulate in spectre the power conversion gain (and possibly the NF and IP1dB) of a mixer, under the assumption that the RF and LO ports of the mixer are conjugently matched with the ports that drive them.

I understand that a solution would be to pump a lot of LO power (so that there is enough power to switch the mixing quads regardless of the mismatch) and calculate the large signals s-parameters through psp. Then, I can use the corresponding s-parameter formulas. I was thinking if there is a more elegant solution that can also be applied for IP1dB.

Thanks in advance,
Yannis

Title: Re: Mixer Simulation under Matched Conditions
Post by pancho_hideboo on Apr 18^th, 2009, 12:14am

yannis wrote on Apr 17^th, 2009, 12:38pm:

and calculate the large signals s-parameters through psp.

Not correct. S-parameters calculated by PSP of Cadence Spectre is Small Signal S-parameters.
As far as I know, PSP and QPSP of Cadence Spectre are not used so much.
So there are many bugs in PSP and QPSP.
See http://www.designers-guide.org/Forum/YaBB.pl?num=1208334972/21#21

yannis wrote on Apr 17^th, 2009, 12:38pm:

I was wondering if there is any way to simulate in spectre the power conversion gain (and possibly the NF and IP1dB) of a mixer, under the assumption that the RF and LO ports of the mixer are conjugently matched with the ports that drive them.

I can't understand what you mean by IP1dB. Is it Input referred 1dB Gain Compression Point ?

I have no answer for Cadence Spectre because I don't use Cadence Spectre for RF circuits design.
But you might be able to find some informations useful for you in the following.
http://www.cadence.com/Community/blogs/rf/archive/2009/02/09/shhhhh-spectrerf-tutorials-and-appnotes-one-of-our-best-kept-secrets.aspx

yannis wrote on Apr 17^th, 2009, 12:38pm:

I was thinking if there is a more elegant solution that can also be applied for IP1dB.

There is no elegant method in real RF measurent using actual instruments.
http://www.maurymw.com

Mixer is operated by non-linearity.
Higher harmonics are always generated in input currents due to non-linearity of DUT even if driving signal have only one frequency.
This is vey true for LO port.
Input impedances of DUT are different between all harmonics including fundamental.
Assume Zin(n*f0),n=1,2,3,4,5,....., generally they are all different.

Input impedances, Zin(n*f0) are affected by power level and fundamental frequency, f0.
Input impedances of RF port are also affected by LO conditions.
http://www.home.agilent.com/upload/cmc_upload/All/1_NVNA_and_X-parameters_in_ADS.pdf

I use Agilent ADS(RFDE) or GoldenGate for RF circuits design.
If you use Agilent ADS(RFDE) or GoldenGate, you can do simulation for your purpose, here you can set reactance for Port components.
I don't use LSSP(Large-Signal S-Parameter Simulation) for this purpose.

You heve to consider following issues.
http://www.designers-guide.org/Forum/YaBB.pl?num=1231854969/5#5
http://www.designers-guide.org/Forum/YaBB.pl?num=1231076499/8#8

You can set port impedances for fundamental frequency and it's harmonics separately.
Here you can add both fundamental frequency and power dependencies for this port impedance using DataAccsess Component of ADS(RFDE).
http://edocs.soco.agilent.com/display/ads2009/DataAccessComponent+%28Data+Access+Component%29
http://edocs.soco.agilent.com/display/ads2009/P+1Tone+%28Power+Source%2C+Single+Frequency%29
http://edocs.soco.agilent.com/display/ads2009/Load-Pull+Simulations
You can do Source-Pull Simulation although the above example is Load-Pull.

Also you can do automatically search and set conjugete matching impedance for Port without using DataAccsess Component of ADS(RFDE).

Title: Re: Mixer Simulation under Matched Conditions
Post by yannis on Apr 21^st, 2009, 11:26am

Hello,
Thank you for your reply. You were right that I didn't fully understand what PSP does and what the Large signal S-parameters are. I studied and looked at the background theory and I realized what I really need for proper mixer simulation. (X-parameter simulations)

I have been using Agilent Momentum for many years for simulation of various passive components. Now, i will give it a try and use the RFIC dynamic link :)

Yannis

P.S.: IP1dB = Input referred 1dB compression point.

Title: Re: Mixer Simulation under Matched Conditions
Post by pancho_hideboo on Apr 21^st, 2009, 6:19pm

yannis wrote on Apr 21^st, 2009, 11:26am:

You were right that I didn't fully understand what PSP does and what the Large signal S-parameters are.

Even if you use Cadence Spectre, you can do LSSP(Large-Signal S-Parameter Simulation) by "ADE>Tools>RF>Wizards>LSSP",
although I have not confirmed whether LSSP of Cadence is implemented correctly or not.

But LSSP is not suitable for your purpose.

yannis wrote on Apr 21^st, 2009, 11:26am:

I studied and looked at the background theory and I realized what I really need for proper mixer simulation. (X-parameter simulations)

I don't think X-parameters are needed for your purpose.

You can do simulation for your purpose, if you can set complex impedance for Port components which are valid even for large signal analysis.

You can not set complex impedance for Port components in large signal analysis of Cadence Spectre.
Again see http://www.designers-guide.org/Forum/YaBB.pl?num=1231076499

Title: Re: Mixer Simulation under Matched Conditions
Post by yannis on Apr 21^st, 2009, 7:36pm

Hello pancho_hideboo,
It's true that I don't necessarily have to use X-parameters but I think are very helpful to match the impedance of a large signal circuit.

The problem with large signal s-parameters is that they are not uniquely defined in the bibliography. I can cite several papers which define LSSP diffirently.... Most of them are even different to what Agilent ADS does for its LSSP simulation.

Thanks again,
Yannis