The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> PSS + PAC versus QPSS for out-of-band IIP3 of a filter
https://designers-guide.org/forum/YaBB.pl?num=1284021633

Message started by aaron_do on Sep 9^th, 2010, 1:40am

Title: PSS + PAC versus QPSS for out-of-band IIP3 of a filter
Post by aaron_do on Sep 9^th, 2010, 1:40am

Hi all,

this is related to my question in the RF measurements section. I thought I would post it here since my question seems to be more about RF simulators now.

Basically I simulated a low-pass filter for out-of-band IIP3, where the two tones are outside of the filter band. I found that using a large + moderate signal in a QPSS analysis or two large signals in a PSS analysis gives simulated IIP3 very close to my measurements. However, when using a PSS + PAC analysis with one large tone and one small signal, the simulated IIP3 is significantly better (10 dB) than measurements.

This may have something to do with the interaction between the nonlinearity and the memory in the filter. I was hoping somebody could shed some more light on this, perhaps with some references, and also confirm that this is normal.

thanks,
Aaron

Title: Re: PSS + PAC versus QPSS for out-of-band IIP3 of a filter
Post by pancho_hideboo on Sep 9^th, 2010, 3:39am

aaron_do wrote on Sep 9^th, 2010, 1:40am:

Basically I simulated a low-pass filter for out-of-band IIP3, where the two tones are outside of the filter band.

Maybe input level of two tones for LPF are around or over compression point of LPF.

See http://www.designers-guide.org/Forum/YaBB.pl?num=1277203226/1#1

Also see http://www.designers-guide.org/Forum/YaBB.pl?num=1205223090/3#3

Show me netlists regarding analysis statements and signal sources for the followings.
(1) QPSS
(2) PSS
(3) PSS/PAC

Title: Re: PSS + PAC versus QPSS for out-of-band IIP3 of a filter
Post by aaron_do on Sep 9^th, 2010, 6:11am

Hi pancho_hideboo,

Quote:

Maybe input level of two tones for LPF are around or over compression point of LPF.

I don't think so. I swept the input from 100uV to 10mV. The IM3 curves were all nice 3x gradient curves. I will try and post more info when I get back to my lab.

cheers,
Aaron

Title: Re: PSS + PAC versus QPSS for out-of-band IIP3 of a filter
Post by RFICDUDE on Sep 9^th, 2010, 7:35pm

What was the delta in filter gain (in dB) between the two out-of-band tones?

Did the large signal tone have more or less attenuation from the filter than the PAC tone?

I would think that the PSS tone would need to be at the frequency with the least amount of attenuation or else the answer might not be accurate. The PSS-PAC method creates a periodic linear (time-variant) result based upon the PSS tone result. If the PSS tone was at a frequency of higher attenuation then I don't know if it gives a correct answer when the PAC tone is at a higher amplitude.

It would be good to repeat the simulation for the following cases
1. PSS at least lower attenuation frequency and PAC at the other
2. Vice versa
3. Narrow PAC-PSS tone separation at the lower attenuation frequency
4. Narrow PAC-PSS tone separation at the higher attenuation frequency

Then you should have enough information to confirm if the frequency relationship of the PSS-PAC tones matter when there is frequency variation in the circuit.

I will be curious to hear the result.

I understand why QPSS gets it right because it considers the amplitude of the moderate tone in forming the solution and the circuit response is weakly nonlinear (if you aren't driving it to absurd IMD levels). So, for QPSS it should not matter which tone is large and which is moderate.

Also, you should be a bit careful with interpreting out of band linearity with active filters. It is possible to get the attenuation you expect, but the linearity be very poor (out-of-band) because of the reduced loop gain in the feedback amplifiers. Interestingly, the IMDs which end up closer to the pass band will be better suppressed due to the increased loop gain. It will really make you scratch your head when you see -30 dBc IMD out-of-band, but the inband linearity is still great.

Title: Re: PSS + PAC versus QPSS for out-of-band IIP3 of a filter
Post by aaron_do on Sep 10^th, 2010, 9:12am

hi,

thanks for the replies. I seem to have jumped the gun a bit. After studying my results I found that in fact neither simulation results match the measurements, although they do give different results. I may have to remeasure some data, in which case I'll get back to you afterwards (it may take a while).

btw, I did in fact use the large tone at the lower attenuation level. I'll see if I can follow your suggestions, RFICDUDE, and figure out what's going on.

thanks,
Aaron