Peak power

Sorry, I can't honestly answer your question. The peak power of a noise like signal is theoretically infinite - but I've never measured this... I guess it is possible to define a peak power if you have an associated probability of that level being exceeded. Investigate the maths of the Raleigh distribution of band-limited Gaussian noise to find out more. See: https://en.wikipedia.org/wiki/Rayleigh_distribution

The following may be of interest:

To measure the average power of a modulated carrier you need to turn it into heat using resistor and thermocouple, as is the case with some of the best wide band power head measuring sets. See pictures and read: https://www.satsig.net/cgi-bin/yabb/YaBB.pl?num=1265438382

Typical measuring equipment however, such as spectrum analysers, use diode type detectors. These are calibrated assuming a CW sine wave input. If the input is modulated (i.e. noise-like) you get a different result, but it is still accurate; you just need to apply adjustment factors.

Using a diode type detector:
To correct for a noise like input add exactly 1.05 dB. This corrects for the difference between a sine wave and a Raleigh band limited noise distribution.
If you using a spectrum analyser you may have log mode or linear mode. In log mode you need to add 1.45 dB. This is because the log amplifier does not amplify the higher level signals as much as the lower level signals.
On a spectrum analyser the shape of the resolution bandwidth filter matters. The noise bandwidth of the filter is typically 1.05 to 1.13 times the displayed resolution bandwidth (e.g.RBW = 100 kHz, 300kHz, 1 MHz etc). Take off about 0.5 dB to get an answer corresponding to the displayed RBW.

Using a spectrum analyser note that the result relates to what is getting through the filter. If the filter is less than the carrier bandwidth then you need to multiply by the filter bandwidth versus the carrier bandwidth (e.g. 1 MHz RBW versus 28 MHz carrier bandwidth). If the filter bandwidth is much more than the carrier bandwidth you may get a false higher reading due to the adjacent noise floor. In all cases you are measuring the noise floor under the carrier as well as the carrier.

Measurement of carrier power is best done by turning the modulation off and measuring in CW mode. A spectrum analyser reading is accurate, provided the noise floor is well down.

If you are measuring a modulated carrier with a spectrum analyser then look for the MARKER NOISE ON function key. The displayed result is dBm/Hz. This is accurate and takes account of all the correction factors. Just multiply by the carrier bandwith e.g. 10 log (1.1 times the -3dB bandwidth) to get the total carrier power.

Obtain HP Application Note 150 -4
https://uspas.fnal.gov/materials/11ODU/AN150-4.pdf

Best regards, Eric.

As I said earlier, the peak power is theoretically infinite for a very small percentage of time. The notion of 'peak power' only means anything if associated with a particular percentage of time. Don't worry about it.

Design and set your clear sky carrier level (dBW EIRP) exactly as specified by the satellite operator.

You will need a large enough HPA such that the output back off is at least about 1 dB. More output back off is preferred. 6 to 7 dB output back off is typical for Ku band VSAT systems to allow for 6 dB of uplink power control during heavy rain.

If you operate multicarrier then you need more output back off (e.g. at least 3 dB) to minimise intermodulation to an acceptable level. Consult the HPA specification. Some HPAs are more linear than others and may operate nearer saturation.

Note that higher order modulation methods like 8-PSK and 16-QAM need better linearity and thus more back off.

The need for adequate back off applies in IF (70 MHz and L band) combiner/splitter networks (transmit as well as receive) as well as transmit HPA amplifiers.

When connecting devices (modems, IF amps, upconverters, HPAs) together avoid extreme gain settings such as near maximum or minimum.

Best regards, Eric.