Member
Joined 2009
Paid Member
It maybe less of an issue with first-order electrical cross-overs where there is significant blending of sound between the two amps. This could give beneficial control over the sound. For example, lower harmonics in the bass, especially 2H, tends to reduce warmth which some prefer but they may still enjoy the harmonics higher up. Really Treble harmonics are usually outside of our hearing range of course. And I remember that the late Japanese tube guru, Sakuma-san, whilst controversial did have his followers and he experimented with combining different power tubes into a final output transformer for blending of sound signatures (yes, no cross over in that case). He was also a proponent, when using a 2-way, of employing push-pull for bass and single ended for treble. As far as I know, he used only passive cross-overs, even when bi-amping like that.
For me, I see one advantage with passive being the ability to experiment with different single amplifiers whilst exploring their different sounds because you want to hear them full range.
Last edited:
@bigun, exactly right. Also, once you find a speaker you really like, it is easier to experiment with other components when the rest of the chain is simpler.
@mark. Same here. I didn't mean to derail the conversation with my comment. I respect your opinion and I had the exact same opinion a few years ago. How can a DAC with exemplary measurements have a sound? What are those tube guys smoking over there? Right?
I'll say one more thing and then take this over PM, Mark. Look at the harmonic content of the signal going to the speakers. Does it have the harmonics, however low they may be, in a monotonically reducing fashion? That is, 2nd higher than third, and third higher than fourth, and so on. I'm finding that once you achieve this, suddenly all these marginal effects and things you thought would never make a difference become audible. I would run this test nominally at 1 kHz and assume that the rest of the spectrum has the same order.
I'll give an example. I bought the RMI ADI-2 Pro a few years ago because it has exemplary measurements and it bested all my other DACs at the time. After changing some electronics and getting the right "harmonic content," I traced some glare around instruments to the DAC. I had an old (10+ years old) CS4398 ebay DAC board that I had modded with output transformers from Onetics, Jensen, and another no name set of transformers bought from another guy on diyaudio. Amazingly this DAC with the no name trafos sounds better despite much worse measured jitter and noise. The point of the anecdote is to say that I was and still am very much a "measurement guy." But there are things I am hearing that I cannot explain or at least the harmonic content explanation is the best I've found.
BTW, anyone here is welcome to come and have a listen 🙂
@mark. Same here. I didn't mean to derail the conversation with my comment. I respect your opinion and I had the exact same opinion a few years ago. How can a DAC with exemplary measurements have a sound? What are those tube guys smoking over there? Right?
I'll say one more thing and then take this over PM, Mark. Look at the harmonic content of the signal going to the speakers. Does it have the harmonics, however low they may be, in a monotonically reducing fashion? That is, 2nd higher than third, and third higher than fourth, and so on. I'm finding that once you achieve this, suddenly all these marginal effects and things you thought would never make a difference become audible. I would run this test nominally at 1 kHz and assume that the rest of the spectrum has the same order.
I'll give an example. I bought the RMI ADI-2 Pro a few years ago because it has exemplary measurements and it bested all my other DACs at the time. After changing some electronics and getting the right "harmonic content," I traced some glare around instruments to the DAC. I had an old (10+ years old) CS4398 ebay DAC board that I had modded with output transformers from Onetics, Jensen, and another no name set of transformers bought from another guy on diyaudio. Amazingly this DAC with the no name trafos sounds better despite much worse measured jitter and noise. The point of the anecdote is to say that I was and still am very much a "measurement guy." But there are things I am hearing that I cannot explain or at least the harmonic content explanation is the best I've found.
BTW, anyone here is welcome to come and have a listen 🙂
RA7 my experiences generally seem to parallel yours, hence SE amps driven by low distortion op-amp based electronic crossovers in turn driven by a MiniDSP SHD which certainly measures very well.
I know what I like, and it sounds suspiciously similar to what you describe, not in other words rigorous electronic perfection.
And I haven't found EQing to as close as flat as possible to be sonic nirvana either, rather bass shy, and sort of bright and sterile sounding. The standard HK preferred room response curve seems a good starting point, to which I generally add some extreme low end LF boost and a bit of a scallop in the lower midrange. I have horns so that may play a role here, certainly my goals were different when I was younger and used much less efficient direct radiators. (My goal still is something that sounds neutral at my listening position, and not just to me.)
I know what I like, and it sounds suspiciously similar to what you describe, not in other words rigorous electronic perfection.
And I haven't found EQing to as close as flat as possible to be sonic nirvana either, rather bass shy, and sort of bright and sterile sounding. The standard HK preferred room response curve seems a good starting point, to which I generally add some extreme low end LF boost and a bit of a scallop in the lower midrange. I have horns so that may play a role here, certainly my goals were different when I was younger and used much less efficient direct radiators. (My goal still is something that sounds neutral at my listening position, and not just to me.)
^Yup. With most SE amps at low power levels, the harmonics are ordered in monotonic fashion and there are probably no to very low higher order harmonics.
With speakers, yes, the classic B&K curve is a good start. I ended up with something similar for tonal balance. Flat from 400 Hz to 10 kHz, slight boost below 400 Hz, more boost below 100 Hz. The below is an in-room LP measurement with complex gating in Holm.
With speakers, yes, the classic B&K curve is a good start. I ended up with something similar for tonal balance. Flat from 400 Hz to 10 kHz, slight boost below 400 Hz, more boost below 100 Hz. The below is an in-room LP measurement with complex gating in Holm.
Attachments
Last edited:
I somewhat agree with you regarding the harmonics profile of the signal chain. I think this applies mostly to the speaker. The rest of the equipment chain isn't as influential in proportion IMO, unless you're used to vacuum tubes and transformers coloring up the audio with harmonics. I'm a class A solid state guy myself and have an updated ADI2 pro fs with the AK4493 chip set. I think its superior to just about any other dac I've heard or owned, including some one off proprietary designs costing 5 figures. Its neutral which conveys the honest truth without adding anything else. I've owned just about every known multi bit DAC chipset, including all incarnations of TDAs, CSs, PCMs, AKMs, ESS, Wolfson etc and the three that stand out are TDA1541A S2, AK4493 and ES9038 Pro.
My ears prefer the lowest amount of both even and odd order harmonics, with the exception of a small amount of H2 harmonics in the top end. Otherwise the sound is too boring and engaging. Maybe thats why I like the sound of lighter paper cone drivers from the midrange on up. Bass is a completely different story... that needs to be as tight as possible while reaching way down low, but with a touch of warmth in the mid bass. Passive xovers help "loosen up" the low end just enough to make it sound interesting that way. Its not just the inductor series resistance that affects this. Larger iron core inductors are perfect for this purpose.
For many years, I had been using FIR minimal phase with very long taps (over 200K), but recently I performed serious ABX test, and found well designed IIR can be more preferable to FIR filters including linear phase, while IIR sounds slightly tighter in general. I'm sure it's depends on the digital processor, and I use DMG Audio EQ in special IIR mode. Don't blindly believe FIR is always better.
While it's questionable whether this is about phase itself, phase does have its fundamental influences.
IIR is the way things are in nature. While FIR can be useful in specific situations, the phase change of IIR is necessary and correct in others.
IIR is the way things are in nature. While FIR can be useful in specific situations, the phase change of IIR is necessary and correct in others.
I'll have to agree with ra7 here... for the longest time I figured the DAC and amplifier should only play a minor role. Once they measure good or even great it should be sufficient, as it would be swamped by the speaker's misbehavior anyway.
Same goes for amplifiers, as a measurement kind of guy I figured the researchers had it all figured out already and we should be done once the output level is sufficient.
That is until some DIYaudio friends came over with a variety of amplifiers, to test in my own setup. This test scattered that believe. Even though an amplifier can have way less harmonic distortion, it's specific profile does seem to shine trough in the end listening.
I was quite impressed with a relative limited output amplifier known here as the Fetzilla. Somehow that combination worked quite well with my speakers, and rather different than most of the others in our test, including my own amp at the time. There was another "monstrous" mosfet amp in the test that acted the same, but way more powerful. A clone of a Goldmund Telos.
It kind of shattered my believe that all competent designed amps will sound the same (or close enough not to matter). Looking further into the subject, the points ra7 made about harmonic structures seem to add up quite well. The design philosophy of that Fetzilla amp is following the proposed idea:
I don't mean to say the Fetzilla is the best example of such an amplifier, just that it's the one that made it clear to me that the profile of the harmonic structure does matter to me, personally. I won't be stating: "all competently designed amps sound the same" any longer. I'd say: it depends...
As far as active vs passive: I'll use whatever I need to get where I want to be. Right now I'm finishing up a hybrid experiment that is part passive/part active. As for FIR filters, they can do a lot, but mostly they do what you tell them to do. There's quite a bit of danger in that 😀.
Same goes for amplifiers, as a measurement kind of guy I figured the researchers had it all figured out already and we should be done once the output level is sufficient.
That is until some DIYaudio friends came over with a variety of amplifiers, to test in my own setup. This test scattered that believe. Even though an amplifier can have way less harmonic distortion, it's specific profile does seem to shine trough in the end listening.
I was quite impressed with a relative limited output amplifier known here as the Fetzilla. Somehow that combination worked quite well with my speakers, and rather different than most of the others in our test, including my own amp at the time. There was another "monstrous" mosfet amp in the test that acted the same, but way more powerful. A clone of a Goldmund Telos.
It kind of shattered my believe that all competent designed amps will sound the same (or close enough not to matter). Looking further into the subject, the points ra7 made about harmonic structures seem to add up quite well. The design philosophy of that Fetzilla amp is following the proposed idea:
I don't mean to say the Fetzilla is the best example of such an amplifier, just that it's the one that made it clear to me that the profile of the harmonic structure does matter to me, personally. I won't be stating: "all competently designed amps sound the same" any longer. I'd say: it depends...
As far as active vs passive: I'll use whatever I need to get where I want to be. Right now I'm finishing up a hybrid experiment that is part passive/part active. As for FIR filters, they can do a lot, but mostly they do what you tell them to do. There's quite a bit of danger in that 😀.
Last edited:
Does phase even apply to nature?
Seems to me, sounds that emanate from nature (production) have to be in phase.
They are what they are....how can they not be in phase?
By phase, don't we really mean relative phase that attempts to mathematically describe how well reproduction is accomplished?
Crossovers don't exist in nature afaict.
I believe the need for xovers is probably audio reproduction's biggest problem.
And I can't see how the idea of IIR applies to natural sounds.
It's clear enough that natural sounds come from bandwidth limited sources, but bandwidth rolloffs occur at the ends of their spectrums, how ever wide or narrow the spectrums are (eg, lions roar vs bird singing).
These natural spectrums with their rolloffs occur all over the 20-20kHz spectrum, at all kinds of different orders.
Which rolloffs at what frequencies do we want our phase warping IIR crossovers to mimic?
Seems to me, we simply want the relative phase of our audio output, to be unchanged from the audio input....whatever the composition, or however we describe, the audio input.
Seems to me, sounds that emanate from nature (production) have to be in phase.
They are what they are....how can they not be in phase?
By phase, don't we really mean relative phase that attempts to mathematically describe how well reproduction is accomplished?
Crossovers don't exist in nature afaict.
I believe the need for xovers is probably audio reproduction's biggest problem.
And I can't see how the idea of IIR applies to natural sounds.
It's clear enough that natural sounds come from bandwidth limited sources, but bandwidth rolloffs occur at the ends of their spectrums, how ever wide or narrow the spectrums are (eg, lions roar vs bird singing).
These natural spectrums with their rolloffs occur all over the 20-20kHz spectrum, at all kinds of different orders.
Which rolloffs at what frequencies do we want our phase warping IIR crossovers to mimic?
Seems to me, we simply want the relative phase of our audio output, to be unchanged from the audio input....whatever the composition, or however we describe, the audio input.
Everywhere. Imagine yourself on a childs swing, if you had no control of the phase of your kicking. (Even worse, if you were only allowed to kick at the zero (middle) position.)
I figured the nicer ones may do that, but oem has been pretty much the bottom of the barrel, traditionally.
I just replaced my oem head unit with an aftermarket unit finally. The processor was very noisy in the one that was removed.
I also got rid of the variable pwm feature that dimmed the headlights as the interference was horrid, even with a dedicated supply.
Allen, I can't catch the relevance of that analogy. Seems more like a resonance thing, with who knows what other physical principles going on too...????
So going back to audio alone....
If i may repeat a line from my prior post...
By phase, don't we really mean relative phase that attempts to mathematically describe how well reproduction is accomplished?
Relative phase...relative is the key work imo.
Our measurements are all relative to something.
Single channel, relative to some stimulus signal, a sine sweep or noise, etc. Dual channel, one channel relative to the next, again with some stimulus signal.
When anything in nature produces sound, is is likely to be bandwidth limited.
I think we can all agree to that.
If i could measure the frequency response of a dog barking, a dog with limited bandwidth, it will mathematically appear that the lower frequency sound of the bark has lagging phase vs the upper.
But the bark is the bark, and no frequency parts of it are delayed in reality.
They are only delayed mathematically in depiction against unlimited bandwidth, which is something immaterial to the barks spectrum.
I've read Richard Heyser say any attempt to measure and mathematically describe the frequency response of a bandwidth limited sound source, will show a downward sloping phase curve, making it appear the lower frequencies are lagging the upper. And say that doesn't mean any frequencies actually arrive later, it just means there is no other way to describe it mathematically.
Anyway, my 2c at take #2.....
on why i don't think we can make valid comparisons of natural sounds vs reproduction of them, using frequency response and xover terminology.
heck, production has no frequency response at all...it's not responding to anything...it's just makin sound 🙂
I've run this simulation to show what a phase shift looks like in the time domain. With input, the output Voltage doesn't move at first.. also it continues after the signal generator stops, ergo it has been delayed.
Such filters as this can be used to represent the effect of limited bandwidth.
Attachments
If the high frequencies of the speaker are delayed a lot relative to the low, the bark may not sound like a bark anymore. Right, Mark? In reality, crossovers don't introduce so much phase shift and so it is never a problem.
Speakers themselves produce the most distortion in the chain, but it is mostly of the benign nature. I am finding that once you get that H2, H3 in the right proportion, suddenly you can hear down the chain and pin-point the sound of different components.
Member
Joined 2009
Paid Member
Yep, i know all too well about phase shifts in the time domain.....from bandwdth filters. Heck, aren't I the phaseholic that's always bashing IIR xovers' phase shifts? 🙂
What i've been trying to convey, rightly or wrongly, is that the understandings we bring from measurements of limited bandwidth speakers, reproduction gear......don't seem to apply to natural sounds, productions.
Iow, natural sounds, productions, are not IIR, and are not matched by IIR speaker systems. Natural sounds don't match anything other than, they are what they are, imho.