A better way to show IIP3 curves rather than PSS PAC?

mixer question

hi ;
to design a mixer on spectre RF,i made 2 tone test by PSS with single tone & combining the other one by PAC ---> it results in a very broken curve & 3rd order curve became upper then 1st order curve

is there a better way to show the famous IIP3 curves rather PSS PAC ..???

thanks in advance

Re: mixer question

for the broken curves, what step size did u use?

for the second point, did u terminate the output properly w.r.t. to the input? (both referencing to 50 ohm load)?

i used qpss and pac together to get teh iip3 curve.

Re: mixer question

Hello

First try a 1 dB compression simulation, this requires only one tone..
it's well known that 1dB≈IP3-10dB, this is rough, and the diff may be more ~ 12dB or so in DSM technology.

Now, for IP3 use actual 2 tones "in port, use second frequency", this simulation may be heavy so recommended to set accuricy to moderate

For IP3 curves, I always had funny looking curves, dont wory, make sure:

1. To start from really small input "-30dBm or less"
2. When plotting the IPN curves, you will be asked for ref power "Default is min", choose a power at which the IP3 curve really has slope of 3 "such that the asymptote drawn by the plot window is tangential"
3.Too wide or too narrow a step in power are both bad, a resonable value I always use is ~ 3 dBm step

Hope this helps