My neighbor has one on his ST-3. His is using the harmonic for excitation, regulated by the AVR, as the stock unit voltage so high that there is no choice but to use it, which is common with some of the CGG units. He gets about 2-3 years per unit, with heavy use.
FYI- these do NOT regulate to RMS voltage so won't work well with a nasty PF load like a cheap high current , switch mode batttery charger. Output voltage will be too low for a very bad PF load.
One of the issues regarding longevity is that even in the ST-3 with a normal harmonic, the voltage peak is over 400V, while RMS voltage is much lower.
The standard harmonic system of the ST3 sorta regulates output voltage by sensing load current via the harmonic winding, so for a big load spike, like starting a motor, it really works well. I have a dual mode setup, my own design AVR for normal running, but I can switch to harmonic only for running a small switch mode type MIG welder. I have an adjustable resistor to limit harmonic current to get the desired voltage.
For cleaning up the output waveform several methods have been reported depending on what the problem is. The cheapest digital o-scope or usb scope will let you see what needs to be done. The common methods are- adding motor run capacitors on the line output (for some ratcheting and spike distortions), running an AVR on the line voltage instead of the harmonic (if you have "harmonic hump" distortion. Some found adding capacitance after the bridge diodes helped with harmonic hump distortion. If the ratcheting and spikes are really bad, I'd add two 200-400 uH chokes plus capacttance on the line output. A surplus line reactor plus motor run capacitor will also help but is pretty spendy. Putting massive amounts of capacitance (> 60uF) of motor run capacitors starts to affect your peak motor starting capability and starts to show a little in fuel consumption.
For the high frequency EMI on the AC line, I found that adding 0.1 uF snubber capacitors parallel to each diode in the bridge (4 total) made a world of difference. With these small caps (240VAC rated) the EMI went from detectable across much of the AM band from 8 feet away from AC wiring to 8 inches in just a few spots. A standard dual common mode choke filter will also be helpful but I found the snubbers got it good enough for me and only cost me a few bucks.
That's my pitch on ST heads- I envy the guys with Stamford clones who just belt them and forget about it! The variation in different ST heads is huge, in terms of waveform quality. At one point I had 3 different ST-3 rotors and stators on my bench- and none were physicially identical to the others. I"ve seen lots of waveform screen shots from ST-5's also, and they also vary dramatically. The best thing I can say about ST heads is they are cheap, and simple. If you get copper windings and put in good bearings, and a modern bridge rectifier, they are reliable in creating a high THD waveform for thousands of hours. My custom low EMI inverter with a 5 step sine has lower measured THD (measured THD of 12%) than my ST-3 (measured THD of 15%), even when running on mains excitation instead of harmonic. Either runs everything I need just fine, but the timer motor on my washer makes some noise on the generator.