Everything physical system is time variant, even if only a little. However, time variant/invariant are human-defined concepts. Time-invariant is a conceptual/mathematical model that says if the same input produces the same output regardless of when the input is applied, then we classify the behavior as time-invariant.
Its up to the person doing the modeling to determine how much imperfection rises to the level of constituting a 'different' output. If the model works well enough for the problem at hand, then its a good model for that use case. Maybe not for other cases.
Its up to the person doing the modeling to determine how much imperfection rises to the level of constituting a 'different' output. If the model works well enough for the problem at hand, then its a good model for that use case. Maybe not for other cases.
Within the respective camps, some consensus has seemed to develop.
Camp 1:
Nobody can hear distortion below 0.5%. Don't waste your money. All electronics sound the same. Feedback is good because it lets you get good numbers out of crappy circuits. We need a lot.
Camp 2:
All distortion is bad and must be eliminated, regardless of audibility. OK, I'll accept a miniscule amount of distortion so long is it is only 2nd or 3rd order. Feedback is good because it makes good circuits even better. We want a lot.
Camp 3:
Gimme plenty of 2nd harmonic, and maybe a little 3rd on the side. Distortion is good if it's musical. No feedback, please.
I haven't seen such post. Would you mind helping me out by quoting 1 or 2 of those? Thanks in advance.
Is seems to me that Syn and Russel are talking about two different instances of time. One, which is prevailing during the excitation of a stimuli and the output of that stimuli (differing wether the system is time invariant or not) and an other one that is about if said output would be equal to the result of an identical excitation but made some time later. The latter seem to relate to the temperature change in an amp...
//
//
Sorry I wasn't precise, arguing.
If distortion threshold is not one of the starting points for discussion and there is no consensus about that, then I think we're loosing time here, only ego fighting, that's pointless IMHO.
Btw designing low distortion gear is an interesting technical challenge, no question about that...
Did I ever claim all three camps play here on diyaudio? No, that's your assumption. Why would you be so eager to call me a liar?
I have worked with engineers in Camp 1. They were proud of their ability to make the cheapest amp that met the specs. Most of the folks in diyaudio Solid State seem to be in Camp 2. Many of the folks in the tube fora are in Camp 3.
And, no, I'm not going to link examples for you. This is just a casual discussion, not a dissertaion.
This thread is becoming a monster.
Russell, I like your second and third options.
Which to choose?
HD
Russell, I like your second and third options.
Which to choose?
HD
I'm libertarian. Each must choose one's own way, and let others do the same.
I'm personally in Camp 2, but that should be obvious looking at my posted amp design. Moar linearity! Moar gain! I tried Camp 3 for awhile. I wandered lost in the no-man's land between camps 1 and 3, once, also. Nyquist, Bode, and Baxandall showed me the path home to my people in Camp 2.
Non-linearity does not cause time invariance, unless the non-linear event changes the system in a permanent way. I'll explain:
On a guitar string, there's elastic strain and inelastic strain. Elastic strain is normal when sounding a note. The elastic string vibration contains the fundamental and harmonics and is modulated by the other strings, while the string remains unchanged and undamaged. The instrument remains time-invariant. OTOH, inelastic strain causes a permanent deformation to the string. The guitar is now out of tune, or worse, the string is broken. The instrument's response to an input is now different. That is time variance.
Then there was Pete Townshend's guitars. They drastically changed their behavior at the end of concerts. They became very time-variant at the end of the show.
The second statement is true: everything is time-variant, but sometimes we can ignore it. Of course all musical instruments have some time-variance: they are subject to aging, which is a function of time, hence time-variance. Instruments and audio gear also all have noise. Thermal noise is uncorrelated to the input, and it causes the output to change slightly for the same input. Noise is a function of time, and if the system's output is a function of time, independent of the input, then it's time-variant.
I listen all product designed by Camp 1, 2, and 3. I like Camp 2. But I also tried to design solid state amp with Camp 3 philosophy with little feedback. I share a lot of my design for free, I am happy if other people happy using my design. A lot people in my country like Camp 1 because of their economic background and their education background. They are never listen product in Camp 2 and 3.
Do yo expect all people are same? 😛
If you want to sell your product, do some marketing research. What is your target market? If you want to make for yourself, you are free to explore all concept. Usually listening skill is improve overtime. So, if your amplifier now make you satisfied, may be in the future you do not.
Id' argue that everything below -80 dB HD is purely academic, even below -60dB I'd want to see a controlled test demonstrating it's audibility.
Listening Test - Instructions
Distort audibility tester | Distort documentation
Anyone claiming otherwise should try it by themselves. Its a reality check.
Was discussed already, but not much response. Wonder why.
https://www.diyaudio.com/forums/sol...dback-huge-benefit-audio-188.html#post6285302
If I can design an amplifier with THD below average human threshold, I will build it. In engineering, you must design with safety margin. If average human threshold in THD is -60dB, I will design with target at least -70dB.
If you KNEW that average human threshold in THD is -60dB and you design an amplifier with THD -60dB, you are not good engineer.