Designer’s Guide Community

nano_RF

Community Member

Offline

Posts: 50
madison

AM/PM simulation in cadence
Mar 28^th, 2006, 8:00am

Hello All,

Does anybody know "How to simulate AM/PM distortion of an amplifier in Cadence". There seems to be a behavioral test bench called AM_PM_test_ckt in cadence's rfLib. Is this the on;y way to test AM/PM distortion. Even for that I am still trying to figure out a way to simulate an actual amplifier (instead of a behavioral model).

I was reading through Ken Kundert's paper "introduction to rf simulation and its application". In this he mentions that a PAC analysis would be able to compute this. Ken, could you please elaborate more on this ?

Thanks,
--Vikas

Back to top

IP Logged

Jess Chen

Community Fellow

Offline

Posts: 380
California Bay Area

Re: AM/PM simulation in cadence
Reply #1 - Mar 28^th, 2006, 10:38am

Vikas,

The AM_PM_test_ckt in rfLib will not directly help you because it is strictly a baseband equivalent test circuit designed to define the AM/PM parameters in a baseband equivalent PA model. (I know because I created that circuit.) However, the test circuit does point to a possible answer to your question. You could do the same test with passband counterparts. Specifically, I believe you could:
1. drive your amplifier with a single sinusoidal source with a variable power,
2. do a swept pss analysis, sweeping the power level in your sinusoidal source,
3. plot the phase of the fundamental at the output of your amplifier against the variable input power (or voltage). The resulting plot should characterize the AM/PM distortion.

There may be newer methods in the Cadence documentation. I have not looked for them yet.

-Jess

Back to top

IP Logged

nano_RF Community Member Offline Posts: 50 madison	Re: AM/PM simulation in cadence Reply #2 - Mar 28^th, 2006, 11:40am Thanks Jess, The method you suggested seems to be simple and straightforward. I am going to try that, and will give a feedback on that. Yeah, I was aware that you had written that behavioral circuit. I got your name off the behavioral AHDL files. Thanks again for responding. --Vikas
Back to top	IP Logged

Ken Kundert Global Moderator Offline Posts: 2386 Silicon Valley	Re: AM/PM simulation in cadence Reply #3 - Mar 29^th, 2006, 6:41am What is AM/PM distortion? Is this the same as AM/PM conversion? -Ken
Back to top	IP Logged

Jess Chen

Community Fellow

Offline

Posts: 380
California Bay Area

Re: AM/PM simulation in cadence
Reply #4 - Mar 29^th, 2006, 8:30am

Hi Ken,

I briefly looked at your paper, the one Vikas cited, and it seems you use the terms in the context of periodically driven circuits, circuits with mixers? However, some authors define AM/PM,AM conversion in the context of power amplifiers to describe saturation(AM/AM) as well as a phase shift that that sometimes accompanies it (AM/PM). Both effects occur with only one periodic signal. I have not thought about it long enough but I suspect the two definitions are consistent if we drive the PA with the sum of a large tone at one frequency and a small tone at another. I've seen AM/PM conversion and AM/PM distortion refer to the same thing. The first reference listed below uses "conversion" while the second uses "distortion".

Peter Kennington, "High-Linearity RF Amplifier Design", Artech House.
Steve Cripps, "RF Power Amplifiers for Wireless Communications". Artech House.

-Jess

Back to top

IP Logged

nano_RF

Community Member

Offline

Posts: 50
madison

Re: AM/PM simulation in cadence
Reply #5 - Mar 29^th, 2006, 8:58am

Hello Jess, Ken,

I believe they both refer to the same thing. Usually we would like to have an amplifier that has a constant phase irrespective of Input power level. A recent article associate this to the feed-forward path (Cgs, miller cap) of an amplifier that in turn causes it to have a different phase shift depending on the input power level.

Jess, about the simulation set-up I have a Psin port (with AM Modulation) applied to the input of the Amplifier. And I did a swept pss analysis (beat frequency= AM Mod frequency) by varying the input power level(prf). Then I looked at the Output Phase Vs Input Power level @ fundamental(AM Mod Frequency). From the simulation Output Phase did not change much. However this result seemed misleading when i looked at the Phase of the Input signal (from the Input port) @ fundamental. One should expect that this should have a constant phase irrespective of the Power level i am supplying from it. Instead it has weired type of phase variation with input power level.

Any idea what am i doing wrong in here?

Another thing I want to know is about the Psin Port. When one defines say AM with the prf(dBm) in the same port. Say, I have defined a Psin Port with following thing...
Frequency= 1GHz
Amplitude(dBM) = 5 dBm
AM Mod Index = .85
AM_mod_frequency = 1MHz

In this case I want to know if 5dBm refers to the average power of the AM modulated signal. Cause I am seeing that a constant envelope 5dBm signal has a lesser Peak-Peak voltage than the 5dBm AM modulated signal.

Thanks,
--Vikas

Back to top

IP Logged

Jess Chen

Community Fellow

Offline

Posts: 380
California Bay Area

Re: AM/PM simulation in cadence
Reply #6 - Mar 29^th, 2006, 10:43am

Hi Vikas,

Try setting just the amplitude under the "Source" heading, not the "Modulation" heading. In other words, just use a simple UNMODULATED sinusoidal source, a pure tone. Make the power of the tone the pss sweep variable. �The PSS sweep will do a separate simulation for each power level in your sweep. Plot the phase of the output fundamental frequency against the power in the pure tone. My main point is that you do not need a modulated signal to characterize AM/PM conversion in power amplifiers.

(Just for completeness, there is something I call PM/AM conversion. From a baseband perspective, this is a linear effect caused by gain mismatch and quadrature error in direct conversion IQ mixers. Again, this can be characterized with a single tone.)

-Jess

Back to top

IP Logged

Andrew Beckett Senior Fellow Offline Life, don't talk to me about Life... Posts: 1742 Bracknell, UK	Re: AM/PM simulation in cadence Reply #7 - Mar 29^th, 2006, 11:43am There is also the modulated mode in PAC, which will compute AM/PM conversion (this is a linear analysis though) - the same is available in PXF. Not quite sure if this is what you're after or not, but it's probably worth a look. Andrew
Back to top	IP Logged

T New Member Offline Posts: 1	Re: AM/PM simulation in cadence Reply #8 - Aug 27^th, 2010, 2:46pm Dear Jess, Could you share some details on building block(s) and mechanism(s) that is/are behind what you call PM/AM conversion? Thanks Thomas
Back to top	IP Logged