IIP3 for a mixer: shooting vs. hb

Please see this Article on Cadence Online Support:
11454704 (COS) Difficulties simulating IP3 on a switched-FET mixer

And see this reference: LINFET: A BSIM Class FET Model with Smooth Derivatives at Vds=0

by Lawrence F. Wagner, C. Michael Olsen. IBM Systems & Technology Group, Global Engineering Solutions. 

http://www.nsti.org/procs/Nanotech2007v3/7/W79.604

This is a known limitation with BSIM models.

It is strongly recommended that you use the PSP, BSIM6, or HiSIM2 models for this application, as they do not have the inherent limitation of the BSIM3/BSIM4 models.

 

If you don't have access to support.cadence.com, let me know.

 

best regards,

Tawna

One more thing... Strongly nonlinear -> use pss shooting. However, I suspect what you are seeing is due to the models.

Hi Tawna, Thank you. Actually I was referred to that aleady. I forgot to tell that I am not using BSIM but UTSOI (based on PSP) so I should not suffer from that. (However, the other experiments I did in the 180nm use BSIM). In any case, shooting takes terribly long, hb is fast. Is it still recommended to use shooting here because of my rectangular LO, right?

Upon further investigation it seems to me that shooting is more trustworthy (unfortunately because it's terribly slow) and hb often produces garbage results. I base this conclusion on:

1.) Adding an AC coupling cap (like 1F) between the ideal LNTA and the mixer changes IIP3 significantly (up to 24dB!!) although all common mode voltages are the same and there is no DC current. It should not change anything. Usually without cap, hb brings much better IIP3 than with cap. For shooting the result seems to be the same in both cases.

2.) I check DC IIP3 by sweeping DC input and calculate IIP3 from the Taylor series coefficients of the DC output voltages. Now, using very low frequencies for qpss (like fLO=1kHz, f1=1100 Hz, f2=1090 Hz) with shooting approaches the DC result. However, qpss with hb is significantly lower (something like 30dB).

From 1+2 it can be observed that hb is sometimes significantly above, sometimes significantly below shooting. For (1) I guess the "great looking IIP3 result" I get is just bogus :(

Hence it seems I can't trust hb.

Any further remark is much appreciated!

Could you contact customer support? I think it would be best if we could see this so that we can either check whether your hb settings are appropriate, or see if there is a specific problem that needs fixing.

Thanks,

Andrew

Hi Andrew, the problem is I am student and hence do not have access to the support. Life would be so easy with it I guess ;-)

Hence I need to rely on public information and volunteers from forums. Let me take this opportunity to thank you so much for all the voluntary help you give in this forum and which help me already so much.

As a reference to my previous post, this shows a plot of "DC"-IIP3 vs. common mode voltage of the mixer with small load resistance (of 1mOhm or so). It can be seen that qpss/shooting curves approach the one obtained using dc Analysis (and extracting Taylor series coefficients via derivatives). It can also be seen that the hb versions are far off:

http://snag.gy/xSczk.jpg

This curve shows IIP3 vs. frequency for shooting/hb:

http://snag.gy/20yFX.jpg

The variants for shooting are consistent (with and without ac coupling cap) and make more sense: Higher frequency -> lower IIP3.

hb on the other hand are much smaller, and produce different results.

There should be a way of you getting support. If you're under the Europractice scheme (mostly in Europe) you can go to Europractice for support, and if they need support in turn they come to us in the UK. For other University programmes it is done typically via the designated owner of the programme at that institution - often a Professor.

There is some detail on the process here: 

Without seeing your actual design and simulation setup, I doubt we can do much more...

Regards,

Andrew.

If you don't have access to support.cadence.com, let me know.

best regards,

Tawna

Hi Tawna,

I hope you are still reading this.

Could I come back to this offer to get an account?


Thank you so much!

If you are a student, you should talk to your Department Chair to get access to Cadence Online Support. I"m not sure if you are in the US or Europe. Processes differ depending on your geographical area.

best regards,
Tawna