Testing this amplifier has been a highly informative and sobering experience for me. John Curl, on whose equipment the measurements were made, has tested many switching amps (he had several on hand, including a Tripath output stage) and was not as surprised by the results as I was.
If I had seen the ZR1600 on my lab test bench in 1965 (when I got started) I would have been alarmed. I would have checked for proper biasing of the output stage and the feedback loop for enough compensation. The distortion measurements on the ZR1600 were reminiscent of the early 60's. Here they are:
IM distortion 60/7000Hz, 1W level: 0.5%.
THD 1W/5 kHz: 0.8%
THD 10W 5 kHz: .28%
Power output at clipping, 8 Ohm load, one channel driven: 200W
Power output at clipping, 4 Ohm load, one channel driven, 220W
Dynamic headroom, 8 Ohms: minus 1.76 dB
Dynamic headroom, 4 Ohms: minus 4.5 dB
Power bandwith at any level below clipping: 10Hz to above 40kHz
1 kHz rise time: 1 ns (!!)
1W 1 kHz distortion spectrum: 14 harmonics visible, 2 and 3rd, 4th and 5th about equally high
Noise: considerable, rising steeply above 20 kHz
Waveform fidelity: sine waves at all frequencies thickened with switching noise riding along
1 kHz square wave: 1 cycle of overshoot and large amount of switching noise on top of wave which was greatest at low (1W or less) listening levels
All of these measurements with the exception of the square wave risetime are exceptional poor. The unit failed to reach its rated power before clipping, and oscillated when doing so.
It is the nature of switching amps that one output device is one while the other is off, leading to a measurement phenomenon that looks rather like crossover distortion. The amp designer has decided not to filter the switching artifacts from the output and they are a major component thereof. It is likely he counts on the self-inductance of moving coil tweeters (like the treble horns in pro speakers) to roll off the switching frequency, suggesting the amp might sound better on tweeters with higher self-inductance (most domes, for example) than with ribbon tweeters which lack such a component.
Summary: as John Curl says, it is difficult to know how high the distortion of the amplifier is, since noise is its biggest component. On some measurements distortion was within the manufacturer's 0.5% THD spec, on others the measurement was in excess of 3% (with 80kHz filter engaged on the distortion meter). By any standard there is a LOT of distortion.
I will contact the manufacturer for a suggested output filter so that the power measurements may be made more accurately. It is quite possible the waveform clipping we observed was a noise component and not the signal at all. Special equipment and test procedures may be needed in order to make accurate power and distortion measurements.
On to critical comparative listening.
