Langston

Warning! Bogus subjectivity ahead! : ) I finally gave the amps a listen and I don't hate Class D anymore. I ran through my usual test songs and the first thing that hit me was that I could hear everything. "Wow, that's good, I didn't think Class D could do that" was my first response. My Klipschorns are sounding very good with the Crites cast frame woofers and the Dave Harris Tractrix horn and B&C DE750 in a two-way system. I'm not finished tinkering, but these things are so good that they've allowed me to do other things for a bit, such as mess around with amps. The loudspeakers are sealed against their corners and the ceilings are 10'. This isn't the end-game room for them, but it's the most convenient for experimenting. Klipschorns NC400 Setup AHB2 Setup After a while of listening, "dryness" became my primary impression, everything was there but it wasn't pulling me into the music like usual. Maybe I'm just tired or something I thought, so it was time to switch the AHB2 back to the horns. The moment I started playing music again it was over. The AHB2 brought back the liquid, not more detail, but the smooth spread of the recording environment was filling the space between the loudspeakers again with increased depth and pulling me into the music. The difference was pretty significant, though I could easily see myself enjoying the NC400's if I hadn't heard the AHB2. The NC400 reminds me in a way of my MC275 tube amps, just the opposite end of the spectrum. The MC275 is too liquid and too soft, the detail is there but too laid back for my taste. Whatever that means. : ) I have a huge, long-term, personal bias against Class D amps and this is a big change for me toward respecting the technology. For now I'm going to use them for LF. Then I had an idea, why not arrange a way for you guys to hear the difference for yourselves? In my next post you'll get access to (3) music files: 1. The original 96k/24bit test track. 2. The same recorded from the output of the AHB2. 3. The same recorded from the output of the NC400. God bless you and your precious family - Langston

Good question - I measure everything before using it and rarely see DC offsets on the output of line-level audio components exceeding about 3mV. The rule of thumb is that anything under 50mV is acceptable, under 20mV is good and under 10mV is great - but based on my experience I'd fix it if it were above 5mV. Just my opinion - I'm no expert. You'd make the DC measurement between center and ring on an RCA and between pins 2 and 3 on an XLR output. Give the component 5 minutes to warm up, disconnect anything from its inputs and if it has a gain knob mfg's say to leave it at minimum, but I sweep it through its range. When you sweep the gain you need to let it stay put for a few seconds to let the circuit settle. The NC400 compares the DC on its inputs to its outputs. If it sees DC on the output when it's not being fed to the input, it'll fault and go into protect mode. If it sees DC on its input, it'll obediently amplify it along with AC, if any. One story from my concert production days - I had a friend who ran another company that was using the same EAW line array at the time that I was. EAW released updated firmware and presets for the UX8800 processor (included FIR processing for the mids and highs), and I of course measured the thing to study the update differences prior to feeding my amp racks with it. My friend just did the upgrade and took out (48) high frequency drivers at the beginning of a show. EAW replaced them all for free, but man what a hassle.

Last set of measurements.. I don't like the shape and pattern of the THD I'm seeing here from the NC400 at approx. 1 watt, but it's so far under the probable threshold of audibility that I'm stretching things to complain. Classic 1kHz S/N at full throttle (1% THD) for both amps into 4Ω. AHB2 NC400 I selected two swept twin-tone IMD measurements for the amps and the NC400 is neck and neck with the amazing AHB2, both at 1 watt output. First up is DFD, or difference frequency distortion measurement that begins at mean (average) frequency of 20kHz with twin-tones 40Hz below and above that mean. Thus measurement #1 is actually equal levels (1 watt) of 19,960Hz and 20,040Hz. Distortion components to the 5th harmonic are recorded and plotted. This is repeated for 31 logarithmic steps down to a mean frequency of 250Hz. The second IMD measurement is classic SMPTE, but with the addition of a swept modulation frequency. Normally twin-tone SMPTE distortion measurements use 40Hz at a 4x (12dB) higher level than the 7kHz frequency of interest and a single value is calculated. In the following the same method applies, except that the offending (modulation) frequency begins at 1kHz and logarithmically steps down 31 times to 40Hz. Again at 1 watt. Much more revealing, just as the swept DFD measurement is much more revealing that the typical 19kHz/20kHz twin-tone. Finally, the multitone FFT plot has the NC400 as the winner. Fascinating amp! God bless you and your precious family - Langston

That is SUCH A GOOD question! : ) With loudspeakers ALL THAT MATTERS is acoustic output. Then there is the electrical output in the lovely and pristine world of active crossovers, but reality is the COMBINED acoustic response of the raw driver in its cabinet or horn WITH the electrical filter that produces the output from the loudspeaker that we hear, for better or for worse. Sorry for the "shouting", but I think the capitalized words might be helpful to some. All that cool research that went on in the 70's onward and documented in AES papers, largely initiated by the Aussie genius Neville Thiele, completely changed the loudspeaker world from the Stone Age into what we have now. I ended up using 2nd order electrical crossovers of different slope functions (Linkwitz Riley and Butterworth) combined with EQ along with the natural rolloffs of the drivers in their horns to achieve an actual Linkwitz Riley 6th order acoustic crossover very near 400Hz. I use software that allows me to combine the electrical and acoustic domains into a single result to figure out what is likely to sound the best prior to actually implementing it. Not rocket-science any more. : ) One more thing - 6th order is too steep for good time domain performance, but in my case I'm able to get away with it because I use an overall FIR filter to effectively turn the multiple drivers into a single impulse. It's taken years, but I've found that you need to do things as correctly as possible without FIR filters, and then use them for polish after you've done everything you can through classical (minimum phase) methods. The Klipschorn bass with the DE750 required very little "polish". I'm still looking forward to giving the DCX464/Tractrix my best effort. Ugh! Thank you for pointing that out. That was an error that I'm going to fix. The ACOUSTIC "hay stack" as they call the low end rise in the concert production world, is only 2dB in my case with the Klipschorns in my home. 7dB was the electrical reduction of the highs that enabled the acoustic 2dB hay stack. I'm a newbie to the genius of the Klipschorn bass and am now only about 1dB higher in the bass than the highs. 2dB muddied the female vocal just a tad. In concert production I generally had a 6dB acoustic hay stack for the subs and ran the passbands above that as flat as possible. Not that it matters, but I figure that anecdotes from my time in pro audio might be fun. : ) God bless you and your precious family - Langston

I love surprises, but I'm getting antsy too, maybe tonight, but first more useless trivia! : ) Another easy one - gain comparison with the AHB2 switched to low-gain: Now for something interesting, an output magnitude comparison using the 4Ω resistor and the 4Ω nominal moving coil loudspeaker load simulator. The resistor is of course a flat line at 4Ω, but the simulator has the following impedance: The AHB2 has a very low output impedance which minimizes changes in the magnitude response of the loudspeakers it drives down to inaudible levels. The NC400 takes this much further and behaves almost like a short circuit. This is achieved by its ground referenced (non-bridging) design, simple circuit path (more on this in a bit) and lots of feedback. Something else the NC400 stands alone with in the realm of Class D amplifiers is an extremely low and well behaved noise floor with great suppression of the SMPS power line fundamental and harmonics. The AHB2 completely suppresses SMPS power line noise, which I have yet to see elsewhere. Now we're getting close to an the most surprising thing I learned about the NC400, and a hint is provided in the following phase vs. group delay plot. The blue traces of the NC400 show evidence of the benign 65kHz noise peak seen above, but notice the absolutely flat LF group delay. That's impossible UNLESS.. ..UNLESS this thing is DC coupled from input to output - which it is! This amp just became my first choice for powering low frequencies. The following shows DC amplification at precisely the same gain as AC. You are going to have to be careful not to feed this thing DC from a pre-amp, etc., but it would probably have to be broken to do that. Folks with loudspeakers without passive crossovers are going to have to take the most care because there's no capacitor to interrupt DC going to their delicate HF driver(s). I restored an old Crown DC300 to use on my bench to test woofers for rub & buzz and offset distortion experiments. One of the hassles with rub & buzz testing is the noise generated by the portion of the test that takes the driver near Xmax. What you can do with a DC amplifier is offset the driver near Xmax and then drive it with a much smaller signal AC test stimulus, then repeat using the opposite offset. Which is what's gonna happen to your drivers without capacitors in series with them if you feed the NC400 with a mix of AC (music) and DC from a downstream component. Best case will be distortion, worst case.. Well you know. A DC amplifier like this is simpler to design, requires fewer components (such as capacitors), and has the best LF transfer function in town. Both my NC400 amps are completely DC-free on their outputs when feeding them AC, so there's some fancy servo (feedback) going on I'm sure. The NC400 runs stone cold into low loads at high levels. It's quite the reverse of Class A. Fascinating amp. Wonder what it sounds like? : ) One more set of measurements to come. God bless you and your precious family - Langston

I've finished the measurements unless (we) think of something else. This is the most interesting amp measurement session I've ever had, and I've measured dozens of these things over the last few decades, though mostly professional stuff. I haven't listened to the NC400 yet and to tell you the truth, I'm not looking forward to it. Why? Because it measures, overall, about the same or better (!) than the AHB2. In my experience, these are the two best measuring audio amplifiers in the world. I'm just worried the NC400 isn't gonna sound good, as 100% of my experience to date with Class D amps has been. If it doesn't sound good it's going to show me that I'm no closer to correlating amp measurement to the listening experience than I was 30 years ago! Let's start with the easy stuff. The Bench The Power Dedicated 120/240/30A. Nope - I haven't messed up YET. The NC400 negotiates world voltages automatically, the AHB2 has to be ordered to spec. Mine are 120V. Single frequency power into a resistor at 120V and 240V. 20Hz and 50Hz were at 3% THD, 1kHz and 5kHz were at 1% THD. The AHB2 performs to spec, thus isn't included. NC400 120V 4Ω 20Hz NC400 240V 4Ω 20Hz NC400 120V 4Ω 50Hz NC400 240V 4Ω 50Hz NC400 120V 4Ω 1kHz NC400 240V 4Ω 1kHz NC400 120V 4Ω 5kHz NC400 240V 4Ω 5kHz Finally, let's channel a little Nelson Pass and look at the 1kHz 1% THD sine with residual amplified 10x for both amplifiers. And then let's look at the same at 1dB lower than these clipping levels. AHB2 1% THD 1kHz AHB2 1dB under clipping NC400 1% THD 1kHz NC400 1dB under clipping Much more interesting stuff to come.. God bless you and your precious family - Langston

Benchmark will let you return the AHB2 after 30-days if you don't like it. "Risk Free" is BS 'cause you might keep it! : )

Nah, I was just throwing out a teaser based on the few measurements I made yesterday that showed me the NC400 was serious and required a planned measurement session. I drove the NC400 1dB shy of clipping (defined as 1% THD, 1kHz, 8Ω) and saw S/N numbers in the mid 120's. I was also using convenience outlets on my bench, when I do the real measurements I'll be using a dedicated 120V/30A and 240V/30A supplies. I got the AHB2 S/N out of the manual and have never given it a full measurement run-through because I trust the designer so much. He's actually been one of my mentors in this stuff and (almost) everything I've ever fussed at him about concerning one of his products (the DAC3 phase via USB was a recent one), I got a polite slap-down and he'd walk me through the solution. On the (almost), I do want to submit for the record that I was right once a long time ago concerning the latency of the original DAC1. He used a bunch of math to calculate the serial group delays of every component in the system using mfg. specs for each part and I had a shiny new calibrated AP analyzer and, genius that I am, hit a button and measured it - Ha! : )

Very true, but this is such an important question I'm going to have a go at it, though it's obvious you already understand the essentials. Holding all else constant, increasing the electrical signal to noise (S/N) ratio when driving a highly efficient horn loaded loudspeaker system within a few meters or less of a critical listener is a very big deal. The wider the dynamic range of the music, the more important it becomes. The quietest passages of Copland's Appalachian Spring or Vaughan William's The Lark Ascending lifting from a silhouette of playback system silence is breathtaking and more than worth the effort to achieve. Woe to you vinyl worshippers! : ) So.. What to do? You have to have a music source, amplifier and loudspeakers. You can choose how many components (at your pleasure or peril) to add to this, but they are potentially additive noise sources. Since we're talking electrical S/N optimization, we start with the amplifier because we're stuck with it's noise level as the absolute best (quietest) our system can be. The stuff that comes before it can only make it worse AND THAT is what must be avoided and the goal of "gain optimization". Note: when adding noise, two components with equal noise levels will increase total noise by 3dB. Of course volume knobs can increase it more, but let's not make this messy. If you can keep the noise level of one components at least 10dB quieter than the other, there will be no increase in total noise. Want to calculate this for yourself? Analog audio components such as amplifiers have S/N ratios all over the map, but I aim for at least 96dB (16bit CD resolution). I have a Mark Levinson 532H that specs ">85dB" and a Benchmark AHB2 that specs 130dB. The latter sounds better and this may be part of the reason. (The Hypex NC400 measured in the 120's on my bench today) By S/N, I refer to the largest undistorted signal a component can output vs. a noise floor that pretty much stays the same no matter the volume control setting, if any. With a big amplifier feeding high efficiency horn loudspeakers, you will be throwing away S/N and possibly hear noise through the loudspeakers even without music playing. An amp that outputs 10x more power than you will ever use will have you 10dB closer to its noise floor during playback. You have to buy a less expensive amp to achieve better sound! Paul Klipsch was a genius. On the preamp, etc., that feeds your amplifier, you are able to drive it to its full output and you should. This will take full advantage of its S/N ratio, but you'll probably have to turn down the gain on your amp if it has that feature or pad the input like Claude does. This will allow you to keep your amplifier's S/N ratio as the limiting factor in your system. Like everything else in audio, there are diminishing returns where additional optimization (and money) won't audibly help, but we certainly should do the best we can with what we have. Pro amps have always had input gains. Just do it. : ) God bless you and your precious family - Langston

Both Hypex NC400 amps measure the same and are so much better than the miniDSP amp that I'm going to drop the comparison after this post and compare the NC400 to the AHB2, which I consider to be the measurement and sonic champ in my limited experience. Another really weird and wonderful thing about the NC400 is that it's self-oscillation noise is benign enough that I can measure the amp directly with my APx515 without the need of an external switching amp filter. Until this point this filter was required for every Class D amp I've measured. The Audio Precision AUX-0040 filter is so well designed that it is nearly transparent to the majority of audio measurements, but it's nice not to have to use it. This in itself sets a new bar for a Class D amp. The output the NC400 is load-insensitive under 20kHz, another first for a Class D amp in my experience. I tinkered around with several APx515 measurements and the results were so good that I decided to turn everything off and plan what I'm gonna do. Plus the day is getting busy 'cause I just turned 62. : ) God bless you and your precious family - Langston

Finished the build and Claude was absolutely right - it was so easy that I put together both amps in 3 hours in slow motion (my top speed). Parts, fit, design, everything is class A (pun!). Pictures now, bench tomorrow, weeping or joy Saturday. : ) God bless you and your precious family - Langston

Never lose sight of audio's Prime Directive: If it sounds good, it is good. : ) Short Answer: I don't know. Try it both ways and if you can't tell a difference that is in favor of the inconvenient method, carry on with the remote control! : ) The technical aspects in your case lend toward setting the OPPO at its maximum and using your integrated amp for volume control. The OPPO uses digital gain control and the integrated amp probably uses analog gain control. Reducing digital gain causes a reduction in bit depth (resolution) of 1 bit per 6dB of volume reduction, which can be a greater compromise than reducing the analog signal to noise ratio by turning down your integrated amp's volume. Nevertheless, in most cases you will probably find that the controlling volume in the digital domain sounds better than doing so in the analog domain, i.e., keep doing it the way you're doing it, but try it both ways to convince yourself. A final thought is that digital audio playback software like Roon and JRiver Media Center process volume control at 64bit resolution, thus losing 1bit per 6dB gain reduction is not going to compromise 16bit or 24bit music playback. Here's some testing a guy did last year with Roon's volume control that demonstrated this.

Answer As low as possible, such that your system is just able to achieve the maximum volume you want to hear. Background Your playback system will sound better if you set the "gains" correctly. This is more accurately called optimizing "gain structure" of an audio system. Well designed electrical audio components output two things, (a) noise, and (b) signal. Noise is bad and at a fixed and low level, the signal is the good thing, the music. The signal is not fixed in level and can be very low (near the noise) or very high - it can even be so high that it starts distorting. The goal is to run the signal as high as possible so that the distance between the noise and signal is greatest. Example 1: Some people say you should set your amplifier gain at maximum for it to sound best. This is false because it forces you to turn down the downstream components, such as your preamp, to achieve the listening level you want. By turning the preamp signal down, you just pushed the music closer to the noise floor. Example 2: Turn the amp gain way down so that you have to turn your downstream components at or near their maximums to hear a low sound level out of the loudspeakers. If that sounds distorted, turn down each downstream component until the distortion goes away, then turn them up again until you find the "just beginning" to distort volume settings. During normal listening, you should be close to those maximum volume settings. THE LAST STEP is to turn up your amp gains so that you get the sound volume you want out of your loudspeakers. This is the ideal. Example 3: If you're into measuring things, this is how you optimize gain structure with little monetary investment. God bless you and your precious family - Langston

Postscript After the miniDSP SHD Power fail, I decided to build a couple of diyAudio ACA amp kits for the Heresy II's, which is largely @Chris A 's fault for thinking so highly of the designer. Using one per channel in parallel mode sounds wonderful - I'm thrilled. You do have to feed them with enough voltage to drive the little amps' to their max (10dB gain), but that proved easy enough by wiring pin-2 and pin-1 of the balanced outs of a DAC2 HGC to the unbalanced amp inputs. The unbalanced RCA outputs of the DAC2 didn't produce enough voltage swing to get the desired volume with some uncompressed recordings. And then I decided to take a final step into the Dark Side (a.k.a. Class D amplifiers) before calling it quits, and this is largely @ClaudeJ1 's fault for thinking so highly of the Hypex NC400 mono amp kits. A pair of them are due to me Thursday and they'll hit the bench Friday and I'll listen Saturday. Probably. If they don't work out, I'm going to make him swap his K-402's for them! : ) God bless you and your precious family - Langston

Just a little closer and I'll slap some sense in you!

@RandyH is spot-on. Your Klipschorns are stunning loudspeakers when operating according to design, and new capacitors in your AK-3 networks is a good idea and requires little commitment of time and money. The second step is to setup your listening room properly, after that you may be so thrilled you won't want to modify the loudspeakers. If not, you'll be in a better position to determine exactly what kind of changes you'd like to make. Have fun! : ) God bless you and your precious family - Langston

Maybe nothing, depending on how REW is presenting the phase curve. I'm not a REW user*, but if it's set to "wrap" phase at ±180˚ points, then the 4-6kHz wrapping may simply be due to the curve kissing the -180˚ point. If REW is presenting the phase curve "unwrapped", then ignore me. : ) This is the kind of thing @Chris A has probably defined already, but it's the minimum amount of frequency delay (the initial zero-cross of the sinusoidal) possible for a perfect device with a finite passband. Minimum phase is not measured, it is a mathematical transform of the magnitude (SPL) response. Loudspeaker drivers of good quality without external filters (passive, active, or acoustic) usually behave like minimum phase devices quite well. When plotted against the measured** phase curve, the difference reveals non-minimum phase behavior. This difference is called excess phase, thus the excess phase curve is the measured phase curve with the minimum phase component subtracted out of it. This same concept extends to the measured, minimum and excess group delay curves. There is much that can be learned by looking at each of these three curves and the REW manual has some excellent instruction on this. God bless you and your precious family - Langston * I've read the manual and in my opinion it would be a good deal if it sold for $1,000. I sent the author $50 a few years ago to thank him for his contribution to our sport. ** Assuming the measured phase curve is measured correctly, i.e., the microphone and electronics have their inaccuracies compensated for, the physical measurement setup is correct, and the user sets measurement system to remove the correct TOF (time of flight between microphone and driver). Correct TOF removal is no piece of cake and it's a fascinating subject. (PM me if you want this paper)

Your assumptions are correct, doubling the inductor and resistor values while cutting the capacitor values in half will result the same electrical response for the 16Ω version of the driver. You will lose 3dB in acoustic sensitivity (loudness at the same volume control setting) with the 16Ω driver, which means you'll have to pad the higher frequencies by 3dB to maintain the same balance. The new woofer will still go just as loud, but your amplifier will have to supply 3dB more voltage to it to do so. Proof that the same voltage will be applied to the 16Ω driver Proof that the new low pass network will see the same impedance from the new Zobel network and driver, just doubled in value God bless you and your precious family - Langston

Apparently not. This is a calibrated measurement of a pair of 1986 vintage Heresy II's. They are unmodified and function correctly. God bless you and your precious family - Langston

The golden ears thing was definitely a joke, particularly for myself. I was just attempting to draw attention to what many here experienced before I did - an electroacoustic transduction system with the capability of bridging those two domains with very little loss. So little loss that sonic differences between amplifiers isn't an item of debate for me anymore, as this thread indicates. My hearing is nothing like it was decades ago, but I'm now listening to my music collection with a sense of "I've never really heard this before". I hear page turning on music stands and musician coughs that I never knew were there. Some albums are not making it through this, others are so much better than I knew I'm kicking myself for not turning my attention to home audio earlier. Now I'm rarely able to listen to portions of songs while doing subjective testing - I get involved and forget what I'm doing. It's wonderful. : ) God bless you and your precious family - Langston

The miniDSP SHD Power amp outputs are bridged and non-inverting. As is typical with bridged amps, it struggles a bit with 4Ω loads. There are two processing modules included, the standard miniDSP stuff which is quite competent and the Dirac Live module that only accepts FIR convolution files generated by the Dirac Live software. It's a shame that you can't upload your own FIR files to this thing because I can do a heck of a lot better than their blind/automated routine. I'm impressed they tried to make inverse filtering available to everyone, but it's a mess unless your loudspeakers and room are so good they could do without it. My overall opinion is that the unit is excellent except for its Class D amp. Thus I recommend the SHD processor without the amp if you want to play with the Dirac Live software. The intrinsic delay on this thing (digital in to amp out) is ridiculous, but fine for most home audio applications. Digital in is asynchronous, internal processing and digital outputs are fixed at 96k. Delay at 48k SR is 25.27ms Delay at 96k SR is 24.71ms IMO, the sound is transparent through the processing modules, but the amp set a new record for awful. I apparently have obtained @Chris A 's "golden ears" (due to my modified Klipschorns). This is yet another Class D failure for me. Flat, dull and even irritating. The SHB Power amp changed the character of the music from involving to boring. The representation of the recording environment was gone. Instrument and vocal placement were correct but flat. I didn't want to listen to it after a couple of minutes. I spent about 10 hours trying to find a measurement method that made sense as to why this thing sounded so bad, but I failed. Everything was consistently an order of magnitude (10x) or more worse than the Benchmark DAC/amp combo, so nothing in particular caught my eye as an "Ah Ha!" measurement. I still think the >20kHz garbage holds the answer, but somebody smarter than me will have to unravel it. God bless you and your precious family - Langston

Interesting. Striking similarities between my scope measurements of the miniDSP amp. The ASR guy has a $35k APx-555 with a 1MHz bandwidth. My APx-515 has a 96kHz bandwidth and measurement of devices with significant ultrasonic output can cause slew induced distortion in the analog input stages. This is a fact for the analog input of ANY analyzer that is exposed to high level signals or noise above its passband. Thus I use an external low-pass filter designed for these kinds of measurements when using the APx-515. No need to bother with this external filter with my 200MHz scope, which is what I used to start this thread. ASR didn't need the external filter because his analyzer can handle the 450kHz slew, BUT he did engage a 40kHz AES-17 filter (internal option for all AP analyzers*) to suppress the ultrasonic garbage in his results as shown here. This is an 8th order filter, thus the ASR measurement of the ultrasonic noise is wildly understated. The use of the AES-17 filter for digital audio type measurement is like grading on a curve - it gives the garbage producers better results with the excuse that "nobody can hear that stuff anyway". Of course we can't hear it directly, but we do hear the flat/hard results that are likely the result of it through aliasing and/or other mechanisms. I'm going to experiment with some of @Chris A 's IMD thoughts to see if I can coorelate an objective measurement to what we hear. God bless you and your precious family - Langston * The AES-17 filter cannot protect the analog input stages from slew based distortion, its only function is to low-pass audio measurements with significant out of band noise that is within the analog stage's slew limits so that standard distortion metrics behave as if the noise weren't there. In the old days they called this kind of thing cheating. 02 August Edit: After reading through many of the responses on ASR that follow the review, I figured out that the reviewer refers to his external 40kHz low pass filter between amp and analyzer as an "AES-17" filter. This is technically incorrect. Even Audio Precision refers to this "AUX-0040" as a "Switching Amplifier Measurement Filter" and nothing else. The AES-17 filters are internal to the analyzer and follow very different specifications per AES standards whereas the AUX-0040 is an AP product per their own specification. Thus when ASR mentions the AES-17 filter in connection to Class D amp reviews, it's a reference to the AUX-0040. Sorry for the confusion, but it doesn't affect our discussion.

One of my audio projects has been trying to mate an integrated amp to a pair of bone stock, but perfectly operating Heresy II's for one of my sons. I'm into FIR convolution and thought I'd try out a miniDSP SHD Power unit that includes a bunch of great looking features, including an everyman's version of FIR filter generation. The D/A measures very well as does the standard miniDSP processing. I haven't bothered with the "Dirac Live" FIR filter software yet because I'm having a horrible time with the amp section. All I wanted to do today was listen to it in a shootout with my Benchmark AHB2 to make sure it was acceptable. I don't expect anything near equivalence, but I can't even match gains with my DAC3/AHB2 amp combo because the output of this miniDSP Class D "amp" has so much garbage riding on top of the audio signal that it isn't possible to measure with precision. So tomorrow I'll listen and try to match gains by ear to see if the amp, in spite of it's ultrasonic garbage, sounds good. The following plots include comments. I'm sure this will be hard to tell*, but I'm wildly biased against Class D amps. They happened during my tenure in concert production and the companies that employed them for anything other than subs instantly cut the quality of their sound in half. To be optimistic. Great for subs though and cut those amp racks down by half or more, both in size and weight. At the end I include some old scope measurements of the Class D garbage out of these amps as well as some non-Class D's that I was considering for sub duty. If this miniDSP amp is as bad sounding as it measures, I'll be looking for something else for these little Heresy's. Suggestions solicited, even Class D if you are convinced it sounds good and I can return it without hassle. Lots of web sites are saying Class D was "hard" and "flat" (an understatement), but now they're wonderful. I hope so, but I'm not holding my breath on it. God bless you and your precious family - Langston * sarcasm. Frequency Domain Time Domain 1 Time Domain 2 Blast from the Class D past

The most exciting phrase to hear in science, the one that heralds new discoveries, is not 'Eureka!' but 'That's funny...' - Isaac Asimov