Sorry to be a bit evasive on this because you are again asking for a simple answer to something that can be quite complex. I would expect IMD terms to consist of a mix of AM-AM IMD and also AM-PM IMD so it all gets a bit muddy when trying to define the behaviour of IMD terms at -30dBc. For example AM-PM IMD terms tend to be in phase quadrature with the main RF cycle so they could be at -30dBc and in this case these IMD terms would not add or detract from PEP. But I certainly agree that with a simple model of an amplifier it is possible for IMD to add to PEP if the phase is right. In this case I guess you can simply add the voltage as per your sums but only if the amplifier's non linear characteristic places the IMD term at the right phase. I'd usually expect the IMD to be caused by some compression of the peak part of the waveform as the large signal gain of the amplifier is usually 'less' at the peak voltage point of the time dependent waveform compared to parts of the waveform that are at lower voltages. In this case the IMD terms won't be at the right phase to add to PEP. They would reduce it. However, I guess some amplifiers can achieve the phase that adds the IMD to PEP at certain drive levels. I think I could easily design an amplifier that would break the PEP rule in that pdf article. Note that you don't need much in the way of AM-PM to get -30dBc IMD and the amount of AM-AM and AM-PM effect varies according to what part of the RF cycle the amplifier is at. The AM-PM IMD tend to be in quadrature and the AM-AM IMD can be at a wide range of angles I guess and there are also issues with memory effect when trying to characterise IMD behaviour. so it all gets far too complicated for me to want to analyse fully... If you want a clearer and better answer than I can give then you could try emailing an RF amplifier design house? Where I work we design RF power amplifiers (as modules within our product line) but we aren't really that concerned with detailed IMD analysis so I only know the basics. Yes, it's not the correct term, that's why we ended up arguing for the last few pages However none of this stuff about IMD is relevant to this thread so it's probably time to stop talking about IMD... Yes, I appreciate that you knew this and that's the bit that is really relevant to ham power meters. I did look to see what meters are available these days and the nicest one appears to be the LP100A. This looks to be very well engineeered and is very versatile. But it does look to use a Vpeak detector so I would expect the measurement uncertainty to degrade the accuracy spec if there are harmonics appearing at -40dBc.