Consistency between pss, hb, qpss et.al.
时间：04-04
整理：3721RD
点击：
Hi,
In Cadence (spectre) I have created a simple passive, voltage mode mixer with a 50 Ohm port as input and output floating.
So simplify, I use a vcvs to decouple port from the mixer input and shunt the port with 50 Ohm for matching.
I then drive the LO with an ideal flo=1GHz square wave and set the port to sinusoidal, f1=1.1 GHz. The input amplitude is 50mV.
1.) qpss: I run qpss with the two tones 1G (Large) and 1.1G (moderate) and then plot the output of qpss_fd. The difference between fundamental (100M) and 3rd order harmonic (300M) is 104.6 dB ("HD3").
2.) qpss+qpac. For QPAC I evaluate at a second tone, f2=1.09G and then plot the IIP3 curve. It matches the result from 1.), IIP3=31dBm.
3.) pss: Similarly, I use the two tones 1G and 1.1G and the beat frequency auto-calculates to 100MHz. The shape of pss_fd looks qualitatively the same as qpss_fd. However the difference between 100M and 300M is now 91.91 dB; 13dB worse than with qpss!
4.) I also tried hb. The results are the same if I use 2 or 3 tones. Again, the spectrum hb_mt_fd looks qualitatively the same as the previous ones but the difference is now 94.43 dB. Still much worse than qpss but a bit better than 3dB better than pss.
I run all simulations with "conservative".
The differences are a bit too much for me to trust any of them.
Any ideas?
Thank you!
In Cadence (spectre) I have created a simple passive, voltage mode mixer with a 50 Ohm port as input and output floating.
So simplify, I use a vcvs to decouple port from the mixer input and shunt the port with 50 Ohm for matching.
I then drive the LO with an ideal flo=1GHz square wave and set the port to sinusoidal, f1=1.1 GHz. The input amplitude is 50mV.
1.) qpss: I run qpss with the two tones 1G (Large) and 1.1G (moderate) and then plot the output of qpss_fd. The difference between fundamental (100M) and 3rd order harmonic (300M) is 104.6 dB ("HD3").
2.) qpss+qpac. For QPAC I evaluate at a second tone, f2=1.09G and then plot the IIP3 curve. It matches the result from 1.), IIP3=31dBm.
3.) pss: Similarly, I use the two tones 1G and 1.1G and the beat frequency auto-calculates to 100MHz. The shape of pss_fd looks qualitatively the same as qpss_fd. However the difference between 100M and 300M is now 91.91 dB; 13dB worse than with qpss!
4.) I also tried hb. The results are the same if I use 2 or 3 tones. Again, the spectrum hb_mt_fd looks qualitatively the same as the previous ones but the difference is now 94.43 dB. Still much worse than qpss but a bit better than 3dB better than pss.
I run all simulations with "conservative".
The differences are a bit too much for me to trust any of them.
Any ideas?
Thank you!