Roger Sanders of Sanders Sound Systems (a leading manufacturer of electrostatic speakers today) has a white paper on amplifiers. He says transistor amplifiers are better for electrostatic speakers because the low impedance outputs can drive the capacitive load, which he characterizes as 1 ohm at 20kHz, with more extended frequency response. And, because it's easier to get high power in transistorized amplifiers. He claims most audiophiles have undersized amplifiers, and when they hear amplifiers as sounding different that is because they clip differently. But the solution is not to find the amplifier that clips better so much as finding the amplifier that has enough power to never clip.
Interesting, and pretty much in line with my thinking.
I don't think, however, I voltage clip either my "200W" (actually something like 500W into 4 ohms at clipping, as published in Audio review) Aragon, or my "300W" (actually something more like 900W into 4 ohms at clipping, as measured in Stereophile review) Krell very frequently. In fact I'm pretty aware of clipping points because I know the max output of my DAC (which drives the power amp), the digital attenuation I'm using at a particular time, and the digital output level. (However, there's some wiggle room between what the nominal output is and peak output...so I'm not totally sure how this works out...actually scoping the output for clipping like Sanders suggests is not a bad idea and I've often thought about doing it, just never gotten around to it.) I believe I generally do not clip either of these amplifiers based on the levels of signals they are receiving, however I do not know. However, I come far closer to clipping than most might imagine. At the max levels I listen to, I'm right under those levels by just a db or two. And that's with 900W into 4 ohms on tap.
He does mention also current limiting. Current limiting and clipping are somewhat different. I might be getting more current limiting than I know about, though it's doubtful there is very much since both amps are capable of more than 60A peak currents (IIRC the Krell is around 100A).
Sanders claims that average listeners with 90dB efficiency dynamic speakers in typical listening rooms need as much as 500W on tap to avoid clipping.
That's way more than most people think. I'd guess more like 200W but I haven't examined this by doing the measurements Sanders has.
Meanwhile I find (in sighted testing only) the Krell to sound better. Less current limiting perhaps??? I believe that the Krell has flatter output at 20kHz into the Acoustats than the Aragon. I've seen this in acoustic measurements but never measured it exactly. It may be because the Aragon is a low feedback amplifer with a damping factor of only 450 below 1kHz, whereas the Krell is a high feedback amplifier with damping factor of 900 or higher and extends to higher frequencies (though, this is not a Krell specification and I can't remember where I saw this measured). The Krell has zero loop feedback, but the output stage has it's own local feedback (drivers and outputs), as well as more bigger transistors, and power bandwidth to 300kHz. The Aragon is a more traditional design with loop feedback, but less than 2dB of loop feedback, and power bandwidth of 60kHz, but still lots of transistors for low impedance despite low feedback.
A friend of mine who originally didn't think much of the Aragon has been studying the circuit in great detail, and decided it's actually very good. The low feedback fits perfectly with this design, and is probably close to the Baxandall recommendations (either very low feedback, or high feedback, but not middle feedback).
He also points out that the Quad ESL-63's have a very amplifier friendly resistive load, not like the capacitive load Sanders describes. This was part of the ESL-63 delay line design. However, that is unique to Quad ESL's as far as I know.
Interesting, and pretty much in line with my thinking.
I don't think, however, I voltage clip either my "200W" (actually something like 500W into 4 ohms at clipping, as published in Audio review) Aragon, or my "300W" (actually something more like 900W into 4 ohms at clipping, as measured in Stereophile review) Krell very frequently. In fact I'm pretty aware of clipping points because I know the max output of my DAC (which drives the power amp), the digital attenuation I'm using at a particular time, and the digital output level. (However, there's some wiggle room between what the nominal output is and peak output...so I'm not totally sure how this works out...actually scoping the output for clipping like Sanders suggests is not a bad idea and I've often thought about doing it, just never gotten around to it.) I believe I generally do not clip either of these amplifiers based on the levels of signals they are receiving, however I do not know. However, I come far closer to clipping than most might imagine. At the max levels I listen to, I'm right under those levels by just a db or two. And that's with 900W into 4 ohms on tap.
He does mention also current limiting. Current limiting and clipping are somewhat different. I might be getting more current limiting than I know about, though it's doubtful there is very much since both amps are capable of more than 60A peak currents (IIRC the Krell is around 100A).
Sanders claims that average listeners with 90dB efficiency dynamic speakers in typical listening rooms need as much as 500W on tap to avoid clipping.
That's way more than most people think. I'd guess more like 200W but I haven't examined this by doing the measurements Sanders has.
Meanwhile I find (in sighted testing only) the Krell to sound better. Less current limiting perhaps??? I believe that the Krell has flatter output at 20kHz into the Acoustats than the Aragon. I've seen this in acoustic measurements but never measured it exactly. It may be because the Aragon is a low feedback amplifer with a damping factor of only 450 below 1kHz, whereas the Krell is a high feedback amplifier with damping factor of 900 or higher and extends to higher frequencies (though, this is not a Krell specification and I can't remember where I saw this measured). The Krell has zero loop feedback, but the output stage has it's own local feedback (drivers and outputs), as well as more bigger transistors, and power bandwidth to 300kHz. The Aragon is a more traditional design with loop feedback, but less than 2dB of loop feedback, and power bandwidth of 60kHz, but still lots of transistors for low impedance despite low feedback.
A friend of mine who originally didn't think much of the Aragon has been studying the circuit in great detail, and decided it's actually very good. The low feedback fits perfectly with this design, and is probably close to the Baxandall recommendations (either very low feedback, or high feedback, but not middle feedback).
He also points out that the Quad ESL-63's have a very amplifier friendly resistive load, not like the capacitive load Sanders describes. This was part of the ESL-63 delay line design. However, that is unique to Quad ESL's as far as I know.
No comments:
Post a Comment