Differnece in avg. signal power between MATLAB/ADS?

Question asked by drgz on Jul 14, 2010
Latest reply on Jul 18, 2010 by drgz
Hello,

just got a quick question; I've set up a co-simulation test-bench in ADS so I can test my RF power amplifier design with different signals at stimulus generated in MATLAB, however I'm experiencing some trouble with normalizing so that the average signal power is as I want it to be. If I run a single-tone simulation with the tone generated in ADS in a HB simulator, the PA goes into compression with an input power close to 30dBm. Hence, I would expect that if I change the stimulus to i.e. a 16QAM signal with a 6dB PAR, it would be sufficient to set the average power of the 16QAM signal just below 30dBm in order to drive the PA into compression.

The thing is that I have to scale up the signal from MATLAB significantly in order to experience this; hence, the avg. power of the signal as I set it in MATLAB is not the same as the avg. power as ADS want it - so the question is; what is the relation?

In order to scale the QAM signal in MATLAB I use this code:

%%% Updated code %%%
px_av = (1/2)*mean(abs(x).^2)/refImp; % average power of qam signal, refImp the reference impedance
sf = sqrt(10^((pavg-30)/10)/px_av); % scaling factor for scaling amplitudes to desired average power (pavg in dBm)
x = sf*x; % scale the signal to the desired level

Based on the single-tone simulations an avg. power of ~28dBm should be enough - however, in order to drive the PA into compression I need to set the avg. power to ~35dBm; which confuses me. Anyone who see what I do wrong or got any tips for a solution?

Best regards,

drgz