Originally Posted by amirm
Well, real-world does not include speakers that act like resistors.
No, they are generally reactive loads over at least part of the normal audio range. One of the worst cases generally occurs in the octave of two above the bass roll-off.
You can see a good real world example of this in the 100-400 Hz range. It dips below 4 ohms. This speaker starts its bass roll off around 100 Hz:
I happen to have a pair of the current version of this speaker (Primus P363) which has similar impedance and response characteristics.
I have clipped out a Yamaha RXV 371 (mainstream bottom end AVR) and a Denon AVR 1913 (middle of the road) into these speakers. No lasting effects.
In that sense, bench tests that use such dummy loads (which is the "standard") can overestimate the power that the unit produces.
The point being made here is that back in the bad old days SS amplifiers were built using output transistors that were capable and durable with resistive loads, but could be overstressed with reactive loads. When this problem ("Secondary Breakdown") caught early designers unawares, the amplifier output devices would fail and the amplifier would be permanently damaged and required a whole new set of output and driver transistors, as well as perhaps a few resistors. Been there done that.
This problem was quickly identified, and protective circuits were added that would cause the amplifier to effectively clip before it destroyed itself. This clipping would take place at far lower output voltages than would be expected from bench testing with power resistors. The Crown DC 300 is an example of an early amplifier that was subject to premature clipping. One common speaker that the original DC 300had problems with was the AR-3 which was rated at 4 ohms. Another common amplifier that had fewer problems with premature clipping (but could still have problems with some speakers that were worse than the AR-3) was the Dyna 400.
Problems with premature clipping were circumvented by adding more output transistors. One amplifier that implemented this approach was the Dyna 416 which was basically a Dyna 400 with another output transistor in parallel with each output transistor in the Dyna 400.
There were speakers, particularly the legendary Infinity Kappa 9, that had absolutely ridiculous impedance curves. Notice the dip below 1 ohm from 20 to 40 Hz:
that led to the production of amplifiers with ludicrous numbers of output devices in parallel such as the Threshold 4E. I owned one of these babies and it was absolutely bodacious with low impedance and highly reactive loads. Although rated at only about 125 wpc it was capable of more than 400 wpc at 2 ohms and could easily blow power line fuses on the test bench. I put in a 30 amp circuit!
Kappa 9s and speakers like them have pretty well disappeared from the market because they only work well with a limited number of amplifiers. Speaker manufacturers only make money when audiophiles actually keep the speakers and don't return them right away.
NHT was one of the new age speaker manufacturers that started paying attention to impedance curves. Many of their flagship loudspeakers actually have impedance curves that stay at or above their rated impedance:
Output devices with improved resistance to secondary breakdown were then developed which eliminated the need for doubled-up or tripled or quadrupled output devices. Some come in typical plastic packages which completely obfuscates their outstanding current handling and resistance to secondary breakdown. They are not prohibitively expensive.
There was a very good AES paper that was posted in the other thread that documents this very well.
No doubt not from the current edition of the Journal because this is a solved problem.
Complex waveforms also require higher peak power than simple tones.
That's my point! ;-)
Amplifiers are still tested by audio ragazines with sine waves and resistive loads, even though they are actually used with music and loudspeaker loads which is a completely different thing.