I have had many Croft Series 7 hybrids through my hands over the years and they do not run hot. They run warm as owners will attest. Yet, when being technically measured by Stereophile, they managed to get the Croft Integrated amplifier to run very hot and then considered it a fair report.
My suspicion is that a music signal is less stressing to some amp topologies than the ‘steady-state one third power signal into a simulated load’ used by the technical reviewers.
So a topology of amp which tolerates this will have a technical advantage over one that does not. Of course, if you like to listen to a single note all the time….
This is one way in which you can get the contradictory situation where the reviewers rave about the amp, as occurred with the Croft Integrated, while the technician finds fault.
You can bet the Integrated did not run hot at the reviewer’s houses (before and after the measurements were taken) when sounding so musical.
As is often the case their testing is not what they think it is. If it ran hot on test bench they were overloading it in some way and inducing distortion.
The trouble with this kind of ‘science’ is that readers of such reviews tend to feel obliged to bow to the measurement ‘facts’ unquestioningly.
Many years ago Glenn Croft said, ‘When you know what you’re measuring, measure it!’
It seems the measurements were taken from the Tape Out rather than the speaker outputs which would have loaded-down the amplifier’s output and induced the 6db roll-off.
Let’s not forget that Total Harmonic Distortion is also overstated as many Jap amps had vanishingly low figures such as 0.01% distortion and sounded rather thin and unengaging in reviews. In the meantime most speakers distort by about 33%…
If you prefer a more objective perspective regarding magazine technical measurements, Steve Elford of Vertex AQ cites that:
“Fourier transform theorem or FFT software is the wrong way to measure because it’s in the frequency domain and is an iterative process over time so you only see the presence of constant sine waves. By analysing the signal in the time domain, only then can you see the tracking error.”