Designer’s Guide Community

junkwar2003

New Member

Offline

Posts: 6
WA

IIP3 and IIP2 Simulation in SPECTRE
Jan 16^th, 2004, 6:07pm

I have tried using PSS/PAC mentioned in the design guide to compute the IIP3 and IIP2 for a mixer. It works however I have a problem. My RF freq is at 5.6 GHz and the LO is at 2.8GHz with IF at DC. To get simulation to give meaningful results you would want the second tone that PAC places at least a 1-2 MHz away from the RF i think. The problem is the beat frequency that can be a integer multiple of all freq in the system is around 1 MHz or so when running PSS simulation. This takes hugh amount of time since # of harmonic is increased. I tried it and basically simulation crashed claming insufficient memory. Is there a better way to get IIP3 and IIP3 values using SPECTRE. I have new to SPECTRE so any help would help. Thanks

Back to top

IP Logged

Jitter Man

Community Member

Offline

Am I? Or am I so
sane that u just
blew your mind?

Posts: 50

Re: IIP3 and IIP2 Simulation in SPECTRE
Reply #1 - Jan 16^th, 2004, 9:59pm

Junkwar,
Rather than using PSS as your base analysis, use QPSS. It will allow you to place your output frequency as close to zero as you like without penalty.

Having said this, I would like to suggest a possible solution that is even more efficient. Say you were able to perform the analysis using PSS with a 1MHz output frequency. Then consider what would happen to the result if you changed the output frequency to 100KHz. If the circuit bandwidth is much greater than 1MHz, the results should not change much. Now consider shifting the frequency to 10KHz, 1KHz, 100Hz, 10Hz, 1Hz. The results for all of them should be the same, right? Here is where it gets interesting. Why not take it down to 0Hz? In concept this works, but if you tried this on the lab bench you would not be able to resolve the distortion product at DC from the other signals present at DC (offsets, biases, etc.). But when simulating with Spectre using the procedure outlined in the paper, you get the distortion product out separately from the other DC components. So, while I have not tried it, I believe you could use a 2.8GHz PSS analysis with the large RF at 5.6GHz, the small RF at 5.6GHz, and the LO at 2.8GHz. The output intermodulation distortion would be in the -2 sideband. This approach would be very fast as it would be the same cost as a single-tone PSS+PAC analysis.

Jitter Man

Back to top

IP Logged

junkwar2003 New Member Offline Posts: 6 WA	Re: IIP3 and IIP2 Simulation in SPECTRE Reply #2 - Jan 16^th, 2004, 10:22pm Thanks Jitterman, Unfortunately we do not have SPECTRE RF just spectre so there is no QPSS function. I have tried the other method but that does not work well for IIp3 sims. I am almost thinking about moving to ADS and using harmonic balance simulation for faster results. Parag
Back to top	IP Logged

Jitter Man Community Member Offline Am I? Or am I so sane that u just blew your mind? Posts: 50	Re: IIP3 and IIP2 Simulation in SPECTRE Reply #3 - Jan 16^th, 2004, 10:31pm If you have PSS or PAC, you have SpectreRF. If you do not have QPSS, then you have a really old version of SpectreRF and you should upgrade. Jitter Man
Back to top	IP Logged

junkwar2003

New Member

Offline

Posts: 6
WA

Re: IIP3 and IIP2 Simulation in SPECTRE
Reply #4 - Jan 31^st, 2004, 12:32pm

Another questions. I tired to find some guide on how to use QPSS to measure IIPn measurements but I have not found it. Do you use the same technique as pss/pac? I have upgraded our spectre so now we have qpss and qpac. Another problem I saw as the for QPSS all signals have to be classified as Large and in the QPAC you can only specify one signal as large and rest as moderate. I have two large LO signals and one RF. Thanks

Back to top

IP Logged

Jitter Man

Community Member

Offline

Am I? Or am I so
sane that u just
blew your mind?

Posts: 50

Re: IIP3 and IIP2 Simulation in SPECTRE
Reply #5 - Jan 31^st, 2004, 5:02pm

Parag,
You have two LOs? There are two frequency conversion stages? I recommend that you just try it with one first. The procedure is basically the same as with PSS/PAC. Simply use QPSS and QPAC instead. In this case, the LO would be the large tone, the large RF tone would be the moderate tone, and the small RF tone is applied in QPAC.

- Jitter Man -

Back to top

IP Logged