Summary:
The processed Powersoft amps are among the finest products of their type in the world. Lab Gruppen's processed amps share that distinction. Crown's upgrade of the ITech's with the HD series put them in the running as well. Nobody else is home. I'm confident that these companies have accounting departments that tell them to stop the insanity and aim lower. I for one wouldn't be in this business if the passion for excellence that created these products wasn't allowed to the market.
What began primarily as a review of the Powersoft K3 and K20 with DSP and the M50Q amplifiers turned into something much more involved that required a great deal of learning. In regard to these Powersoft amps, my hat is off to you guys. These are amazing products. May 2012 prove to be a successful year for both Powersoft's and Italy's finances. Your country is responsible for many of the very best things our industry enjoys (not to mention AC Milan!).
Subjective comparisons of the sonic character of the amps under test including Powersoft's damping control and unique take on limiting are still ongoing.
Background:
When it comes to the very largest and very smallest amplifiers, it appears that switch mode is here to stay. These tests include some of the best examples in the world and while they sound very good, it is my opinion that some of the less efficient linear types, such as the Camco Vortex series, are still unrivaled sonically. The Camco's in particular can throw a three dimensional sound field between a pair of high end loudspeakers like nothing else I've ever heard from a touring amp. Even if my opinion on this is true, does it even matter for sound reinforcement? For me it does, but I'm a foolish person. My goal is to make every piece of the chain between the source and listener as good as I can possibly make it. And not go broke.
I do not like switch mode amps. The way they work offends the way I think. But their rate of progress sonically has been amazing and gives me hope. I think we may be at a mid-60's like point where folks like me were complaining about those new fangled solid state amps that sounded like a cat fight at HF's compared to a routine tube amp. Or again in 1982 when Sony released the $900 CDP-101 CD player (yeah, I bought the first one in Boone, NC) and folks like me held our ears and cried out for vinyl's revival - maybe via lasers reading the grooves to improve tracking and avoid wear. It never happened and now transistor playback of digital audio can outperform anything the older mediums could have ever achieved. Much of it is still truly terrible while enjoying great success in the marketplace - what's new?
What Powersoft and Lab Gruppen have done by tightly(1) integrating state of the art processing into the best amp designs in this category has resulted in sonics that are approaching my beloved Lake processors into linear amps.(2)
As a warning - I went through a phase in the late 80's with Mark Levinson solid state and Manley tube amplifiers, Martin Logan electrostatic and B&W 801 dynamic loudspeakers, illegal connections of large transformers to the service entrance of my house to provide split 60V "balanced power", mostly stupid cables, etc. I removed passive crossovers, terminated the sections with resistors, measured crosstalk issues with Doug Rife's MLSSA, physically separated the crossover sections, redesigned them with Peter Schuck's XOPT, turned adjacent inductors 90˚, drove smaller drivers with tube amps, bigger ones with solid state, etc. I have stories, many hugely embarrassing. It is now my goal to offer a semi-educated version of this madness to larger audiences.
Overview:
Anyone familiar with the Lake and Crown DSP control software will immediately recognize some of the beauty of the former and the operational simplicity of the latter in the Powersoft software. To be fair to the Lake, it's ability to control multiple processors simultaneously with a single software interface that allows unified personalities of subsets of those processors, including subsets within individual processors, is unparalleled to my knowledge. With the Lake, processing is carried out in "modules" and these modules can be used anywhere in a networked system to do anything Lake processing is capable of. You'll notice this modular schema in the leading installation processors now, which highlights how far ahead of the curve Lake has been. That kind of flexibility necessary requires software that can make use of it, thus for some it can appear a bit complex at first. Anyone dealing with this level of product should have no complaints IMO.
Fundamentally it is difficult to make powerful DSP systems user friendly and Powersoft has done one of the best jobs I've seen. The only thing that was difficult for me to figure out was how to do the simple stuff - import or export one or more presets to/from the amplifier. The manual didn't provide much help in this area. This is fascinating because the software control of the sophisticated portions of the amp were extremely straight forward. Another complaint is that there is no way I'm aware of to view which preset is currently running on the face of the amp. The information that is readily available there is the best I've ever seen: voltage and current output, temperature, supply voltage, etc., all kinds of things. It is quite slow uploading presets from a computer to the amp, but all the control functions within the software proper whether you're connected to the amp or not respond very quickly. Other things I'd like to see improved is the addition of a 2 second ramp up to full volume when the amp is turned on, the elimination of the HF click when turning the amp off, and a 2 second ramp down then up when switching presets live. At present, if you switch a preset with audio running, it'll mute, let a brief moment of audio through, mute again, then switch audio on under the new preset. Not pretty. No doubt Powersoft doesn't expect such silliness from its end users, but in the heat of combat audio anything can happen.
The Powersoft DSP reasonably takes a bit more of an amp-centric approach to processing that fits my needs perfectly. Each channel of the amp has a huge number of input parametric filters that can have independently adjustable skirts similar in concept to the Lake's Mesa filters, and 16 output parametrics that can take on all kinds of personalities, such as 2nd order all-pass filters. The output adds high/low pass filters that can be standard IIR or linear phase FIR types. A key reason I wanted to demo the Powersoft DSP amps in the first place was the ability to import FIR coefficients that allow you to roll your own filter universe. This is the magic that EAW and Fulcrum Acoustic does with different specifics toward a similar objective: to reduce that stuff coming out of the loudspeaker in the acoustic domain that wasn't part of the source in the electrical domain. Hopefully direct import of FIR coefficients will eventually become possible with the Lake processors.
Testing Notes:
All DSP is entirely bypassed (in the case of Powersoft), or with all functions off in the case of the IT8000. Although I'm not sure why anyone would want to bypass the amazing DSP in the Powersoft amps, the ability to route the input directly to the amp and completely eliminate the DSP latency is one of those simple-brilliant ideas I've never seen before. All maximum output tests were conducted with 4Ω resistive loads on both channels (only channels 1 and 2 on the 4 channel M50Q).
Due to a fixed input voltage design, the M50Q and PL380 amps were powered with 120v. The K3, K20, IT8000 and FP14000 were powered with 240v. Everything about running large amps at 208v and especially 240v makes more sense than 120v. My power source was a 75' run of 10 AWG 4-wire cable connected directly to the single phase service entrance of my shop. This is a reasonable simulation of a best case power distribution rig and was quite stiff. Very little voltage drop even with the IT8000, which seemed to dim the lights the most under stress. Powersoft's execution of power factor correction seemed to dim the lights the least and the non-PFC FP14000 was in the middle. Real measurements of amplifier efficiency are not hard to make with the right equipment, but I'm not motivated to do that yet. Powersoft released an interesting PFC white paper that includes comparative measurements between the K10 and FP13000(3).
All clip limiters are engaged to simulate probable use and because it isn't possible to disable the clip limiters on all amps. Even when it is possible, it's still nearly impossible to get any of these amps to flat top a sine wave due the other protection mechanisms the user has no control over. The days where you can turn a sine wave into a square wave with a large amp seem to be over, which is a good thing but makes testing a bit tricky. Looking for the 1% THD threshold to determine max output won't always work either because some amps won't even let you go there. Therefore you have to keep an eye on both THD and the compression of the output signal. Compression of 1dB ends the show per the suggestion of Bruce Hofer of Audio Precision.(4) I stopped increasing level at the 1dB compression threshold or when the next 0.1dB stimulus increase saw a spike in distortion in either the THD or magnitude traces.
This brings up a point that the great John Roberts has mentioned frequently in these forums - the subjective differences you hear between amps at or near full tilt is more due to the limiting methods employed rather than the heavily advertised aspects of class typology, etc. Lab Gruppen wrote a wonderful little amp testing article that included the following warning concerning listening tests: "The purpose of clip limiters is to prevent serious clipping in the power amplifier. Clip limiters always reduce the peak power of the amplifier. The problem with clip limiters is that even at low listening levels there can be transients which trigger the clip limiter. Yet the user is unaware because the indicator is to slow to indicate it. For this reason and for the reasoning that all clip indicators per manufacture will light at different rates it is important to set all amplifiers with or without clip limiters engaged. To run one amplifier with and one without, will only compare the sound of the clip limiter itself and not the amp."(5)
Switch mode amps to one extent or another output a small component of their 200kHz+ switching frequencies regardless of input. You obviously can't hear this and your drivers and passive crossover components can't reproduce it. The problem is that in order to make high quality low noise measurement gear at the audio frequencies, the input electronics are necessarily limited such that these very high switching frequencies cause all kinds of trouble, mainly slew based distortion. It's a physics thing, so you can't thrown money at it with high slew rate input electronics that are magically low enough in noise. The only way to deal with it is to filter out these above-audio frequencies before they reach the test equipment. The AES covered this in regard to measuring D/A converters in its AES17 standard. Same issue, different application. I read everything I could find on the subject and never saw any "before and after" measurements illustrating this effect, so I did my own. I found articles describing that it was a distortion issue, and why, and an instructive and embarrassing manufacturer slapdown of a clueless reviewer.(6)
I couldn't have understood some of the most important aspects of Audio Precision's excellent discussion of the issues surrounding switch mode amp measurement without the help of the Syn-Aud-Con list participants. It was their input that opened the door wide enough for me to see what was going on. I was absolutely baffled at some of the early results I saw and Ray Rayburn, Pat Brown and Bill Whitlock among others turned the light on. Thank you.
The bottom line is that even the best passive high voltage low pass filters made for this kind of testing necessarily (physics again) impart phase error to the test that should not be reported as an amplifier artifact. The best low pass filters (there are only two(7)) have no material affect on the audio frequency region and thus allow accurate magnitude and distortion measurements. Happily, accurate reporting of a switching amplifier's small signal magnitude, phase and noise can be made without the low pass filter, so we are able to measure everything.
It took me 30+ hours of error masquerading as measurement to learn that.
BTW:
I thoroughly expect to have made some errors myself in method and/or interpretation, though probably not in data. I'd greatly appreciate slapdown, correction or instruction in any regard.
Footnotes:
(End Part 2a)
The processed Powersoft amps are among the finest products of their type in the world. Lab Gruppen's processed amps share that distinction. Crown's upgrade of the ITech's with the HD series put them in the running as well. Nobody else is home. I'm confident that these companies have accounting departments that tell them to stop the insanity and aim lower. I for one wouldn't be in this business if the passion for excellence that created these products wasn't allowed to the market.
What began primarily as a review of the Powersoft K3 and K20 with DSP and the M50Q amplifiers turned into something much more involved that required a great deal of learning. In regard to these Powersoft amps, my hat is off to you guys. These are amazing products. May 2012 prove to be a successful year for both Powersoft's and Italy's finances. Your country is responsible for many of the very best things our industry enjoys (not to mention AC Milan!).
Subjective comparisons of the sonic character of the amps under test including Powersoft's damping control and unique take on limiting are still ongoing.
Background:
When it comes to the very largest and very smallest amplifiers, it appears that switch mode is here to stay. These tests include some of the best examples in the world and while they sound very good, it is my opinion that some of the less efficient linear types, such as the Camco Vortex series, are still unrivaled sonically. The Camco's in particular can throw a three dimensional sound field between a pair of high end loudspeakers like nothing else I've ever heard from a touring amp. Even if my opinion on this is true, does it even matter for sound reinforcement? For me it does, but I'm a foolish person. My goal is to make every piece of the chain between the source and listener as good as I can possibly make it. And not go broke.
I do not like switch mode amps. The way they work offends the way I think. But their rate of progress sonically has been amazing and gives me hope. I think we may be at a mid-60's like point where folks like me were complaining about those new fangled solid state amps that sounded like a cat fight at HF's compared to a routine tube amp. Or again in 1982 when Sony released the $900 CDP-101 CD player (yeah, I bought the first one in Boone, NC) and folks like me held our ears and cried out for vinyl's revival - maybe via lasers reading the grooves to improve tracking and avoid wear. It never happened and now transistor playback of digital audio can outperform anything the older mediums could have ever achieved. Much of it is still truly terrible while enjoying great success in the marketplace - what's new?
What Powersoft and Lab Gruppen have done by tightly(1) integrating state of the art processing into the best amp designs in this category has resulted in sonics that are approaching my beloved Lake processors into linear amps.(2)
As a warning - I went through a phase in the late 80's with Mark Levinson solid state and Manley tube amplifiers, Martin Logan electrostatic and B&W 801 dynamic loudspeakers, illegal connections of large transformers to the service entrance of my house to provide split 60V "balanced power", mostly stupid cables, etc. I removed passive crossovers, terminated the sections with resistors, measured crosstalk issues with Doug Rife's MLSSA, physically separated the crossover sections, redesigned them with Peter Schuck's XOPT, turned adjacent inductors 90˚, drove smaller drivers with tube amps, bigger ones with solid state, etc. I have stories, many hugely embarrassing. It is now my goal to offer a semi-educated version of this madness to larger audiences.
Overview:
Anyone familiar with the Lake and Crown DSP control software will immediately recognize some of the beauty of the former and the operational simplicity of the latter in the Powersoft software. To be fair to the Lake, it's ability to control multiple processors simultaneously with a single software interface that allows unified personalities of subsets of those processors, including subsets within individual processors, is unparalleled to my knowledge. With the Lake, processing is carried out in "modules" and these modules can be used anywhere in a networked system to do anything Lake processing is capable of. You'll notice this modular schema in the leading installation processors now, which highlights how far ahead of the curve Lake has been. That kind of flexibility necessary requires software that can make use of it, thus for some it can appear a bit complex at first. Anyone dealing with this level of product should have no complaints IMO.
Fundamentally it is difficult to make powerful DSP systems user friendly and Powersoft has done one of the best jobs I've seen. The only thing that was difficult for me to figure out was how to do the simple stuff - import or export one or more presets to/from the amplifier. The manual didn't provide much help in this area. This is fascinating because the software control of the sophisticated portions of the amp were extremely straight forward. Another complaint is that there is no way I'm aware of to view which preset is currently running on the face of the amp. The information that is readily available there is the best I've ever seen: voltage and current output, temperature, supply voltage, etc., all kinds of things. It is quite slow uploading presets from a computer to the amp, but all the control functions within the software proper whether you're connected to the amp or not respond very quickly. Other things I'd like to see improved is the addition of a 2 second ramp up to full volume when the amp is turned on, the elimination of the HF click when turning the amp off, and a 2 second ramp down then up when switching presets live. At present, if you switch a preset with audio running, it'll mute, let a brief moment of audio through, mute again, then switch audio on under the new preset. Not pretty. No doubt Powersoft doesn't expect such silliness from its end users, but in the heat of combat audio anything can happen.
The Powersoft DSP reasonably takes a bit more of an amp-centric approach to processing that fits my needs perfectly. Each channel of the amp has a huge number of input parametric filters that can have independently adjustable skirts similar in concept to the Lake's Mesa filters, and 16 output parametrics that can take on all kinds of personalities, such as 2nd order all-pass filters. The output adds high/low pass filters that can be standard IIR or linear phase FIR types. A key reason I wanted to demo the Powersoft DSP amps in the first place was the ability to import FIR coefficients that allow you to roll your own filter universe. This is the magic that EAW and Fulcrum Acoustic does with different specifics toward a similar objective: to reduce that stuff coming out of the loudspeaker in the acoustic domain that wasn't part of the source in the electrical domain. Hopefully direct import of FIR coefficients will eventually become possible with the Lake processors.
Testing Notes:
All DSP is entirely bypassed (in the case of Powersoft), or with all functions off in the case of the IT8000. Although I'm not sure why anyone would want to bypass the amazing DSP in the Powersoft amps, the ability to route the input directly to the amp and completely eliminate the DSP latency is one of those simple-brilliant ideas I've never seen before. All maximum output tests were conducted with 4Ω resistive loads on both channels (only channels 1 and 2 on the 4 channel M50Q).
Due to a fixed input voltage design, the M50Q and PL380 amps were powered with 120v. The K3, K20, IT8000 and FP14000 were powered with 240v. Everything about running large amps at 208v and especially 240v makes more sense than 120v. My power source was a 75' run of 10 AWG 4-wire cable connected directly to the single phase service entrance of my shop. This is a reasonable simulation of a best case power distribution rig and was quite stiff. Very little voltage drop even with the IT8000, which seemed to dim the lights the most under stress. Powersoft's execution of power factor correction seemed to dim the lights the least and the non-PFC FP14000 was in the middle. Real measurements of amplifier efficiency are not hard to make with the right equipment, but I'm not motivated to do that yet. Powersoft released an interesting PFC white paper that includes comparative measurements between the K10 and FP13000(3).
All clip limiters are engaged to simulate probable use and because it isn't possible to disable the clip limiters on all amps. Even when it is possible, it's still nearly impossible to get any of these amps to flat top a sine wave due the other protection mechanisms the user has no control over. The days where you can turn a sine wave into a square wave with a large amp seem to be over, which is a good thing but makes testing a bit tricky. Looking for the 1% THD threshold to determine max output won't always work either because some amps won't even let you go there. Therefore you have to keep an eye on both THD and the compression of the output signal. Compression of 1dB ends the show per the suggestion of Bruce Hofer of Audio Precision.(4) I stopped increasing level at the 1dB compression threshold or when the next 0.1dB stimulus increase saw a spike in distortion in either the THD or magnitude traces.
This brings up a point that the great John Roberts has mentioned frequently in these forums - the subjective differences you hear between amps at or near full tilt is more due to the limiting methods employed rather than the heavily advertised aspects of class typology, etc. Lab Gruppen wrote a wonderful little amp testing article that included the following warning concerning listening tests: "The purpose of clip limiters is to prevent serious clipping in the power amplifier. Clip limiters always reduce the peak power of the amplifier. The problem with clip limiters is that even at low listening levels there can be transients which trigger the clip limiter. Yet the user is unaware because the indicator is to slow to indicate it. For this reason and for the reasoning that all clip indicators per manufacture will light at different rates it is important to set all amplifiers with or without clip limiters engaged. To run one amplifier with and one without, will only compare the sound of the clip limiter itself and not the amp."(5)
Switch mode amps to one extent or another output a small component of their 200kHz+ switching frequencies regardless of input. You obviously can't hear this and your drivers and passive crossover components can't reproduce it. The problem is that in order to make high quality low noise measurement gear at the audio frequencies, the input electronics are necessarily limited such that these very high switching frequencies cause all kinds of trouble, mainly slew based distortion. It's a physics thing, so you can't thrown money at it with high slew rate input electronics that are magically low enough in noise. The only way to deal with it is to filter out these above-audio frequencies before they reach the test equipment. The AES covered this in regard to measuring D/A converters in its AES17 standard. Same issue, different application. I read everything I could find on the subject and never saw any "before and after" measurements illustrating this effect, so I did my own. I found articles describing that it was a distortion issue, and why, and an instructive and embarrassing manufacturer slapdown of a clueless reviewer.(6)
I couldn't have understood some of the most important aspects of Audio Precision's excellent discussion of the issues surrounding switch mode amp measurement without the help of the Syn-Aud-Con list participants. It was their input that opened the door wide enough for me to see what was going on. I was absolutely baffled at some of the early results I saw and Ray Rayburn, Pat Brown and Bill Whitlock among others turned the light on. Thank you.
The bottom line is that even the best passive high voltage low pass filters made for this kind of testing necessarily (physics again) impart phase error to the test that should not be reported as an amplifier artifact. The best low pass filters (there are only two(7)) have no material affect on the audio frequency region and thus allow accurate magnitude and distortion measurements. Happily, accurate reporting of a switching amplifier's small signal magnitude, phase and noise can be made without the low pass filter, so we are able to measure everything.
It took me 30+ hours of error masquerading as measurement to learn that.
BTW:
I thoroughly expect to have made some errors myself in method and/or interpretation, though probably not in data. I'd greatly appreciate slapdown, correction or instruction in any regard.
Footnotes:
- By "tightly", I mean that they are able to monitor current, voltage and heat of the amplifier and use that information to do things that are wonderful and effectively impossible when using external processors.
- I love the high power and efficiency afforded by switch mode amps with everything under the midrange.
- PFC - Power Factor Correction, Powersoft S.r.l., circa 2005.
- Measuring Switch-mode Power Amplifiers, White Paper, Audio Precision, Bruce Hofer, 2003.
- Lab Gruppen, Testing Procedure, circa 2005.
- Limitations in Making Audio Bandwidth Measurements in the Presence of Significant Out-Of-Band Noise, Audio Precision, Bruce Hofer, 2005.
- Audio Precision's AUX-0025 and Chris Strahm's LinearX LF280 that I use.
(End Part 2a)
Last edited: