WJW -- George at Audio Regenesis employed the industrial version of the regulator (LM235) on his board, as it is rated at an even higher maximum operating temperature than the LM335 device is that I specified in my article introducing the EFB modification. The LM235 is identical to the LM335 in every other aspect.
As for the regulator's operating temperature, when I was first proving the concept on my unit, I first ran the amplifier for one hour under quiescent conditions to obtain a normal operating temperature, and then drove the unit to full power output with both channels driven for a period of 20 minutes. Since my "heat sink" was the back panel, the temperature of the regulator rose very little from its normally cool operating temperature even with everything all buttoned up during that time. However, the full power output test employed a 1 kHz sine wave test signal, rather than a music signal that would simply reach full power on musical peaks. Using a sine wave is a much more severe test than a test using music would be, as music is so transient in nature, whereas a sine wave is very consistent in achieving full power conditions. When George was doing his own tests of the EFB application for his board, he was using a free air heat sink, a sine wave test signal, full power output tests with both channels driven, but for an even more rigorous 1 hour period of operation with everything all buttoned up. True to the very conservative approach he uses, he achieved an incredible 80+C safety factor for the internal operating temp of the industrial device. Also, before I published my article, and subsequently before George released his board, we both tried to put the circuit through our own form of torture test to induce failure. Since George was most concerned about temperature, he even ran a 337 version of the regulator in the circuit -- without a heat sink(!) -- at full power for one hour, all buttoned up, and even this commercial version of the regulator did not fail or shut down (and it was well over max temp spec too!). I was not as concerned about temperature with my installation, but more concerned about general and transient abuse, so I attempted several rapid AC power on-off sequences that did not induce failure either. Neither did removing the load under controlled full power conditions. So, between all the safe guards and testing we both did on our units, neither of us could induce failure, and temp safety margins of both my original installation, and the finished AR product are quite large indeed. The heat sink on AR's board will certainly become quite warm when all buttoned up under normal conditions (assuming 20C room temp, proper ventilation of the unit, and 50 C internal case temp in the general heat sink area), but due to the heat sink George employs, the internal operating temperature of the regulator itself is really quite low compared to its rating -- even under continuous full power test conditions.
To measure the power output and distortion of your unit, you will need a good THD test set (I use an HP 339A analyzer), two accurate 8 or 16 ohm load resistors capable of sinking at least 20 watts of power each, and a dual channel oscilloscope connected across the load resistors. The THD test set will provide the necessary low distortion test signal to drive the amplifier with, and an accurate AC voltmeter for measurement of the voltage developed across the load resistor, and indication of the THD produced as well. With such a configuration, maximum power output can be tested across the audio bandwidth (at the onset of clipping), as can of course Total Harmonic Distortion. Note that in the article, the THD figures provided are an average of that achieved between the two channels, and made at a power level 1 db down from rated power. This was done to keep the test procedure in accordance with that originally used by Dynaco for the specifications it originally published for the SCA35.
To measure Intermodulation Distortion, a separate IMD analyzer will be required to obtain that data. IMD is measured at the equivalent full rated power of the amplifier, or 17.5 watts developed in each channel, with both channels driven. The figure provided for IMD in the article is also an average obtained between both channels.
I hope this answers your questions!
Dave