Cripps and Loadpull Zoptimum....power?

[RFegg]

in classical way of designing PA, Cripps suggested computing the Roptimum with the formula Ropt=Vdd^2/2Pout.

Now, I have a transistor with maximum deliverable peak output power of say 10W.

Now, I like to design a 1W Power amplifier using this transistor. To compute the optimum load impedance for my desired 1W and
say an acceptable range of IMD, I can do loadpull or use Cripps method above and then use these Z values to broadband match to my eventual load of 50-ohms.

My question is where will I reference my design in terms of Pout for both the Cripps and loadpulls? Will i use the Peak maximum power of 10W or should I use my desired output average of 1W?

 

[vfone]

Load-pull will beat any formulas or estimated equations.

 

[RFegg]

Load-pull will beat any formulas or estimated equations.

I do agree

But my question is in determining the optimum load resistance in Loadpull, will you use your average power of 1W or the maximum rated power of the transistor?

 

[vfone]

Have to use the peak power of the transistor.

 

[RFegg]

Have to use the peak power of the transistor.

this seems to confuse me...
please correct me if i'm wrong but you seem to mean that the compromised design figures for IMD, Gain and efficiency should be referenced or based on the desired Peak Envelope power (PEP) and NOT on its average power.

quite confused because when I'm testing my hardware, i'm setting it such that my output is the average power of 1W and read its IMD, efficiency and other parameters from these settings.
kindly explain it further.

 

[vfone]

I don't know which are your settings but as you just mentioned, for IMD, Gain, Efficiency, and for load-pull characterization, have to use peak power.

 

[RFegg]

thanks for the response.
in summary, my design is a broadband Class AB PA at 1W Average. The input digital modulated signal Peak to average ratio can reach up to 10dB or 10W hence i selected a transistor with PEak output of 10W, but these peaks occur in microseconds only.
Then a digital precorrection requires a minimum IMD to precorrect the output signal to say <-30dBc.

For the Efficiency (as good as can be) and IMD requirement, definitely the optimum loadpull contours will differ between average Po and PEak power.
For the gain requirement, it further depends on the input power. my input power will be higher when i want to output the peak power or PEP (rms).

so if i determine my loadpull using the peak power to an acceptable IMD, Efficiency, shouldn't i worry if these requirements are also met when i reduce the input power to get the average power?

 

[RFegg]

I like to quote a reply from "flyhigh" which is directly related to my topic on;

"
definition of back off power

Hi

An amplifier would be typically used in back-off regime when signal is modulated using some spectrum efficient techniques that is in turn sensitive to nonlinearities.
An example of such a modulation could be OFDM modulation. In its raw version, it is not well characterized by average power as for some symbols peak power could be up to 6dB higher than the average. For these peaks, the nonlinearities have more severe effects than if only average power is considered, so one should "back off" from using maximum linear power of the amplifier (defined as 1dB compression point) and use 3-6 dB less, depending on the system under consideration.

flyhigh"

then seems like yes vfone is right i need to design the IMD requirement based on its P1dB or Peak power...