That is absolutely correct, but in real life not fully attainable (a power amplifier, the one that came extremely close to this goal was Halcro's DM38 - go to stereophile test review and measurements).
Thus one must looking for a compromise, which consists in setting priorities regarding the reduction of those THD components that are responsible for particularly unpleasant impressions (crispy and harsh sound e.g. at Cassandra Wilson's "SSSS") while listening test.
H2 and H3 THD components don't make this unwanted effects.
Check out also Nelson Pass' papers "ZEN"
Last edited:
Well certain tube guitar pre-amps and certain microphone tube preamps certainly sound good with their characteristic distortion. That said once a dense mix is made and mastered adding a layer of IMD (products of distorting the whole mix) is (in my opinion) not going to sound good.
But if you are listening to a vocal or perhaps piano (or acoustic guitar) then you might enjoy adding a little distortion of a certain type on playback.
But for me that is not the subject of op-amp swaps. I use a certain tube amp or preamp when I am trying to create a certain type of distortion. But when I am recording or playing back I want the recording equipment (ADC/DAC) and associated op-amps not to change the signal (to the extent possible).
But if you are listening to a vocal or perhaps piano (or acoustic guitar) then you might enjoy adding a little distortion of a certain type on playback.
But for me that is not the subject of op-amp swaps. I use a certain tube amp or preamp when I am trying to create a certain type of distortion. But when I am recording or playing back I want the recording equipment (ADC/DAC) and associated op-amps not to change the signal (to the extent possible).
The answer is the memory (type) distortion. Here is my post, the second half has a comparison measurement:
https://www.diyaudio.com/forums/solid-state/25428-cancelling-memory-distortion-2.html#post6608493
Slow means you cannot hear complex passages clearly, because the amplifier has high level of memory type distortion. The amplifier smears details.
Nothing to do with slew rate. A low slew rate high IMD amplifier (like tube amplifiers) can sound better than a high slew rate low IMD another.
There are other clear mechanisms that are, in fact, known to be the cause of what you are calling "memory distortion." Varying Vce causes varying capacitance at the output of a transistor. The quiescent current shifts, also.
Unless you are talking about the hysteresis present in audio transformers, there is not memory distortion.
Unless you are talking about the hysteresis present in audio transformers, there is not memory distortion.
So, is the issue that the term 'memory distortion' is defined to mean hysteretic distortion only?
In other words, if Vce varies due to a prior signal, say, a high amplitude low frequency signal causes some discharge of power rail filter caps as power is delivered to the load, then if the power rail is still recovering as a low level signal passes through, and if distortion results, what type of distortion should we call it?
In other words, if Vce varies due to a prior signal, say, a high amplitude low frequency signal causes some discharge of power rail filter caps as power is delivered to the load, then if the power rail is still recovering as a low level signal passes through, and if distortion results, what type of distortion should we call it?
Interesting read: IMD - Something New
TLDR:
IMHO: the goal should still be to reduce IMD as much as possible as only that can lead to a transparent reproduction for complex types of music.
TLDR:
- Only asymmetrical distortion (which also causes even harmonics) will produce sum and difference frequencies in the IMD
- Difference products will appear as subharmonics which might perceptually lead to "warmer" sound because of added content in the lower frequencies
IMHO: the goal should still be to reduce IMD as much as possible as only that can lead to a transparent reproduction for complex types of music.
Its interesting that if reproduction system distortion is reduced enough, then you can clearly hear what is wrong with every recording you listen to. Its not fun, not emotionally engaging. You don't listen for long before you want to go do something else.
The most well designed high end stuff, devices that are special, those things are a pleasure to listen to even though they don't measure perfectly. Unfortunately, that stuff is luxury goods, not cheap. One example, Pass HPA-1, is described in Stereophile as follows:
After auditioning the HPA-1 with a variety of 'phones from Audeze, AudioQuest, and Master & Dynamic, JA declared that "in bass quality and authority, and in midrange transparency, the Pass Labs HPA-1 is without peer." JA also brought the HPA-1 to his test bench, where it stood up to everything he could throw at it (well, not literally), prompting the appraisal: "superb audio engineering." One of HR's long-term reference headphone amplifiers. Recommended Components: Fall 2020 Edition Headphones | Stereophile.com
After having an HPA-1 here on loan for awhile, and comparing it against two better measuring Neurochrome HPAs, I often preferred the HPA-1. Didn't prefer it for analytical listening though, the less good measuring of the two Neurochrome HPAs was preferred when I was trying to optimize the sound of my AK4499 DAC. I wanted to clearly hear what was still not optimal with the DAC so I could try to further improve it. Some time later the HPA-1 here was modified by its original designer. After that I preferred it for everything. Unfortunately, I eventually had to return it. No way to buy a modified one either.
The most well designed high end stuff, devices that are special, those things are a pleasure to listen to even though they don't measure perfectly. Unfortunately, that stuff is luxury goods, not cheap. One example, Pass HPA-1, is described in Stereophile as follows:
After auditioning the HPA-1 with a variety of 'phones from Audeze, AudioQuest, and Master & Dynamic, JA declared that "in bass quality and authority, and in midrange transparency, the Pass Labs HPA-1 is without peer." JA also brought the HPA-1 to his test bench, where it stood up to everything he could throw at it (well, not literally), prompting the appraisal: "superb audio engineering." One of HR's long-term reference headphone amplifiers. Recommended Components: Fall 2020 Edition Headphones | Stereophile.com
After having an HPA-1 here on loan for awhile, and comparing it against two better measuring Neurochrome HPAs, I often preferred the HPA-1. Didn't prefer it for analytical listening though, the less good measuring of the two Neurochrome HPAs was preferred when I was trying to optimize the sound of my AK4499 DAC. I wanted to clearly hear what was still not optimal with the DAC so I could try to further improve it. Some time later the HPA-1 here was modified by its original designer. After that I preferred it for everything. Unfortunately, I eventually had to return it. No way to buy a modified one either.
Last edited:
Speak for yourself. I prefer listening to Jelly Roll Morton on a revealing system than on one that distorts.
FWIW, from a musician's point of view with respect to subjectively evaluating ADC/DAC SQ, a single hit on a small symphonic triangle is a great subject. Clarity vs distortion is quite apparent. It might prove useful for audiophiles as well.
@scottjoplin: All reproduction systems distort. How we measure distortion with fixed sine waves does not give us complete insight into how revealing a reproduction system can be. Otherwise measurements would completely predict how a device sounds, which no one has seriously claimed. Measurements are better thought of as 'figures of merit,' nothing more.
@skitron: Try woodblocks instead.
@skitron: Try woodblocks instead.
They can help with mids but that's about it imo. Triangle is a complex yet clean signal that sits squarely in the frequency range that electronics struggle with i.e. where all opamps start climbing distortion wise. JMO of course.
Last edited:
Midrange is hard to get right. Its often too bright, or too muddied. Its where human voices sit, and we are very sensitive to noticing when they don't sound quite right. Imperfection in the exact sound of triangles and cymbals is less objectionable so long as there isn't too much in the way of high order IMD at higher frequencies.
Last edited:
As a musician who also mixes, mids are actually the easiest for me, but probably because they have received the most attention over the years for the very reasons you cite. But for evaluating ADC/DAC I'm not going to even entertain using something the mids aren't right, so it becomes a game of clarity and detail in the highs, hence my comment. Of course there is no one size fits all concerning tools for subjective evaluations. That said, woodblock is my go-to for evaluating which reverb to use in a mix.
I already knew the effects mentioned. Newbies may understand the thermal error more easily. That's why I wrote about it. Of course, you can write the self heating of the resistors, the dielectric absorption of the capacitors.
You must be kidding about memory of transformer. Think about MC pickup, loudspeaker, microphone. All of them are (half)transformers. So I do not think that audio transformers have significant memory type distortion. Hysteresis is not time dependent. Hysteresis is a linear distortion like THD.
False assumptions. The ear itself generates several percent of IMD.
Psychoacoustic/Ear-Related IMD | Audio Science Review (ASR) Forum
Measurement of the non-spectral memory type distortion is the key:
http://peufeu.free.fr/audio/memory/patents/aes-measurement.zip