I pulled this over from the old PSW forum as background to an upcoming SpeakerPower amp module post.
Purpose:
My recent upgrade to the Danley TH118 subs has motivated me to look into larger amplifiers than the PL380 that has been perfect for the TH115's. I've also been underpowering the low passband of my KF730 boxes on the order of 3dB. I've justified this mentally because the low passband only covers 90Hz to 200Hz and the fact that I overlap the early part of this spectrum with subs. Now I'm going to drive those hungry 10" drivers to their limits. The KF730's are 16Ω boxes and the series pair of woofers is rated for 700 watts continuous, thus 1,400 watts "program" per box. A daisy chain of (4) to (6) of those boxes demands an amp that effectively didn't exist when the KF730 first came out. That, boys and girls, is not the case anymore.
The net result of the amp upgrade will be approx. 3dB more output from the same loudspeaker arrays from 200Hz downward. Same truck pack, same weight, same size, but more rig for the gigs. Just need to pony up around $10k (assuming sales of my existing amps) and I’m in. Is 3dB under 200Hz worth 10 grand for my (24) KF730’s and (8) TH118’s? Heck yes.
The higher frequencies are powered to their program ratings already with an amp that I've yet to find an equal to sonically. On that note, I’m going to add a fair amount of subjective commentary in this review that I used to shy from. I’m over it. Class D and the variants I’ve heard are not ready for critical HF reproduction yet. There, I said it - and it’s a fact unless you’re deaf, don’t care, think MP3 is good, or are trying to earn a living in this business.
Measurement Setup:
Room temperature 20C, voltages at the input and output of the amps were monitored with a Fluke 192C Scopemeter that has fully isolated inputs. Sine waves in various flavors were used for all tests. Long-term tests were done with 500ms on / 500ms off steps with distortion curves monitored via the amazing STEPS module in the ARTA measurement suite. Benchmark DAC1 and ADC1 converters with instrument grade +30dBu I/O capacity were optically connected to my MacBook Pro running XP in Bootcamp. A 21dB pad was used to feed the input of the ADC1 for a 51dBu maximum input capability. At the 4Ω resistive loads I placed on both channels of each amp in the test, this allows a maximum level of 18,904 watts to be measured cleanly. My new SuperBucket™ was used to take the heat. This thing presents (2) 4Ω loads with 14,400 watts long-term dissipation and peaks numerically similar to the national debt. You can also use banana plug jumpers to configure single 8Ω and 2Ω loads at double the wattage, assuming the jumpers survive.
SuperBucket
Measurement Theory:
Everybody and their mother have different ideas about how to best measure the performance of an amplifier. Some folks think the old fashioned way of feeding continuous sines is the deal because they apparently use their amps to arc weld during the middle of the week. On the plus side, you’ll get absolutely comparable results from every manufacturer, on the down side those results are not very meaningful for our application. My use of 500ms on / 500ms off sine steps yield effectively identical results as continuous sines in my experience and this is included. Since I use amps for powering loudspeakers with music, I prefer using shorter lengths of sines that approximate musical behavior as much as possible so I can have more meaningful metrics for the intended application. That's where the rub is – just what kind of stimulus comprises the most meaningful metric? For some reason folks keep looking for a _single_ metric to describe amp power, but I think that's a recipe for failure. In our application there is a long-term load as well as much, much higher momentary peaks that need to pass cleanly up to the point where the loudspeaker system begins to sound bad or fails. Those two thresholds are separated by about 6dB of voltage and 20dB of attitude. I've included both long-term and short-term tests in a way that makes the most sense to me. As of today. And it’s my party so I can do what I want. I’d love contrasting views as always.
Take a trip around the block and you'll notice that Lab Gruppen and Yamaha/NEXO, to name two, are measuring their amps in ways that attempt to account for the behavior of music as opposed to the behavior of welding machines. Both use sines to allow distortion measurement, which is critical. Lab uses a 33ms on / 66ms off step which seems quite reasonable as a single metric, especially at lower and mid frequencies. It might be a little severe (long) for the higher frequencies, but it keeps things consistent, repeatable and logical. Yamaha/NEXO uses a 20ms on / 500ms off step and only measures the last sine wave in the 20ms burst, and that at 1kHz apparently. This probably yields higher power ratings than reasonable and certainly shouldn’t be the only metric considered.
There's a guy named Don Keele that I've been following since the 80's that has contributed much to the loudspeaker industry, and by extension, the amplifiers that power them. The logic behind his attempts to measure both gizmos in a way that simulates actual usage makes sense to me and correlates with my testing and listening efforts, particularly as applied to limiter settings. As a side point, I think advanced limiter topologies that simultaneously deal with long-term heating and peak mechanical issues of loudspeakers in a sonically pleasing way are the next area of serious improvement in the quest for high SPL and good sound. Driver design will need to go hand-in-hand with this. NEXO is a forerunner here.
So what about Keele's short-term tests? Read Shaped Tone Bursts beginning on pg. 6. The earlier pages are an important intro.
Want his stimuli and have time to wait for 300MB? Click here if you dare. I neutralized a very expensive amplifier that was supposed to go to a little outfit that rhymes with “hair” after I was done with it. I wasn't careful enough with the 20Hz tone burst. Whoops.
Bottom line from Keele? From a post he made last November on the SAC list: "What we need is that 100 Watt continuous power amp that can pass 10 kW peaks for short periods of time."
I've had to chew on this concept for years to appreciate it, but I'm convinced he's right and my tests show more than a few amp manufacturers are convinced as well and are building amps accordingly. According to the needs of music instead of welding, imagine that.
Continued on Amp Tests (2 of 2) since this forum won't allow over 10k characters per post.
Purpose:
My recent upgrade to the Danley TH118 subs has motivated me to look into larger amplifiers than the PL380 that has been perfect for the TH115's. I've also been underpowering the low passband of my KF730 boxes on the order of 3dB. I've justified this mentally because the low passband only covers 90Hz to 200Hz and the fact that I overlap the early part of this spectrum with subs. Now I'm going to drive those hungry 10" drivers to their limits. The KF730's are 16Ω boxes and the series pair of woofers is rated for 700 watts continuous, thus 1,400 watts "program" per box. A daisy chain of (4) to (6) of those boxes demands an amp that effectively didn't exist when the KF730 first came out. That, boys and girls, is not the case anymore.
The net result of the amp upgrade will be approx. 3dB more output from the same loudspeaker arrays from 200Hz downward. Same truck pack, same weight, same size, but more rig for the gigs. Just need to pony up around $10k (assuming sales of my existing amps) and I’m in. Is 3dB under 200Hz worth 10 grand for my (24) KF730’s and (8) TH118’s? Heck yes.
The higher frequencies are powered to their program ratings already with an amp that I've yet to find an equal to sonically. On that note, I’m going to add a fair amount of subjective commentary in this review that I used to shy from. I’m over it. Class D and the variants I’ve heard are not ready for critical HF reproduction yet. There, I said it - and it’s a fact unless you’re deaf, don’t care, think MP3 is good, or are trying to earn a living in this business.
Measurement Setup:
Room temperature 20C, voltages at the input and output of the amps were monitored with a Fluke 192C Scopemeter that has fully isolated inputs. Sine waves in various flavors were used for all tests. Long-term tests were done with 500ms on / 500ms off steps with distortion curves monitored via the amazing STEPS module in the ARTA measurement suite. Benchmark DAC1 and ADC1 converters with instrument grade +30dBu I/O capacity were optically connected to my MacBook Pro running XP in Bootcamp. A 21dB pad was used to feed the input of the ADC1 for a 51dBu maximum input capability. At the 4Ω resistive loads I placed on both channels of each amp in the test, this allows a maximum level of 18,904 watts to be measured cleanly. My new SuperBucket™ was used to take the heat. This thing presents (2) 4Ω loads with 14,400 watts long-term dissipation and peaks numerically similar to the national debt. You can also use banana plug jumpers to configure single 8Ω and 2Ω loads at double the wattage, assuming the jumpers survive.
SuperBucket
Measurement Theory:
Everybody and their mother have different ideas about how to best measure the performance of an amplifier. Some folks think the old fashioned way of feeding continuous sines is the deal because they apparently use their amps to arc weld during the middle of the week. On the plus side, you’ll get absolutely comparable results from every manufacturer, on the down side those results are not very meaningful for our application. My use of 500ms on / 500ms off sine steps yield effectively identical results as continuous sines in my experience and this is included. Since I use amps for powering loudspeakers with music, I prefer using shorter lengths of sines that approximate musical behavior as much as possible so I can have more meaningful metrics for the intended application. That's where the rub is – just what kind of stimulus comprises the most meaningful metric? For some reason folks keep looking for a _single_ metric to describe amp power, but I think that's a recipe for failure. In our application there is a long-term load as well as much, much higher momentary peaks that need to pass cleanly up to the point where the loudspeaker system begins to sound bad or fails. Those two thresholds are separated by about 6dB of voltage and 20dB of attitude. I've included both long-term and short-term tests in a way that makes the most sense to me. As of today. And it’s my party so I can do what I want.
I’d love contrasting views as always.Take a trip around the block and you'll notice that Lab Gruppen and Yamaha/NEXO, to name two, are measuring their amps in ways that attempt to account for the behavior of music as opposed to the behavior of welding machines. Both use sines to allow distortion measurement, which is critical. Lab uses a 33ms on / 66ms off step which seems quite reasonable as a single metric, especially at lower and mid frequencies. It might be a little severe (long) for the higher frequencies, but it keeps things consistent, repeatable and logical. Yamaha/NEXO uses a 20ms on / 500ms off step and only measures the last sine wave in the 20ms burst, and that at 1kHz apparently. This probably yields higher power ratings than reasonable and certainly shouldn’t be the only metric considered.
There's a guy named Don Keele that I've been following since the 80's that has contributed much to the loudspeaker industry, and by extension, the amplifiers that power them. The logic behind his attempts to measure both gizmos in a way that simulates actual usage makes sense to me and correlates with my testing and listening efforts, particularly as applied to limiter settings. As a side point, I think advanced limiter topologies that simultaneously deal with long-term heating and peak mechanical issues of loudspeakers in a sonically pleasing way are the next area of serious improvement in the quest for high SPL and good sound. Driver design will need to go hand-in-hand with this. NEXO is a forerunner here.
So what about Keele's short-term tests? Read Shaped Tone Bursts beginning on pg. 6. The earlier pages are an important intro.
Want his stimuli and have time to wait for 300MB? Click here if you dare. I neutralized a very expensive amplifier that was supposed to go to a little outfit that rhymes with “hair” after I was done with it. I wasn't careful enough with the 20Hz tone burst. Whoops.
Bottom line from Keele? From a post he made last November on the SAC list: "What we need is that 100 Watt continuous power amp that can pass 10 kW peaks for short periods of time."
I've had to chew on this concept for years to appreciate it, but I'm convinced he's right and my tests show more than a few amp manufacturers are convinced as well and are building amps accordingly. According to the needs of music instead of welding, imagine that.
Continued on Amp Tests (2 of 2) since this forum won't allow over 10k characters per post.
Last edited:
