Do all amplifiers sound the same thread

[Steve81]

A speaker is a comparatively slow mechanical system. Talking about microseconds makes sense with amps, but not with speakers. So this begs the question of what duration of a signal is audible? Does a high amplitude microsecond pulse have any residual impact on the signal a speaker plays?

I'd hazard a guess and say you're approaching this the wrong way in talking about a microsecond pulse (or maybe I'm just full of it, which is always a strong possibility). Consider a simple case of a 1kHz sine wave. Obviously at 1kHz we're talking 1 millisecond to cover the full cycle, but the actual peak of the wave is only a fraction of that.

 

[Irvrobinson]

A speaker is a comparatively slow mechanical system. Talking about microseconds makes sense with amps, but not with speakers. So this begs the question of what duration of a signal is audible? Does a high amplitude microsecond pulse have any residual impact on the signal a speaker plays?

It looks like the measured pulse here lasted about 0.1 milliseconds or 100 microseconds. So I'm thinking a speaker would not have time to start movement at the cone before a microsecond signal was gone (frequency and size of the driver would matter).

Anyone know or care to speculate what the audible effect of such a short pulse would be?

Would the effects of clipping hang around longer to effect the speaker's motion?

Who mentioned microseconds? Most techies abbreviate milliseconds as "ms" and microseconds as "us", and of course nanoseconds, picoseconds, and femtoseconds as ns, ps, and fs. Peaks on the order of 10s of milliseconds seem likely to be audible, and speakers will reproduce them.

I remember some years ago Carver Corporation made a big deal out of some researcher needing thousands of watts to reproduce the transient peak of a scissor snip. I don't know how true it is, but the folklore is still on their web site.

 

[KEW]

Who mentioned microseconds?

Uh... how about ADTG and you?
I understand you are arguing that microseconds are too short to matter, but I don't understand your question above!

That's a good point. If the peaks are only a microsecond duration, that seems like very insignificant exposure to me. Would a microsecond duration peak even be AUDIBLE to us? Or would our neural perception just completely ignore that signal altogether ?

The problem is that music is a bunch of peaks happening in quick succession, and the peaks aren't a microsecond, they're more likely several milliseconds. Distort them by clipping the peaks or whatever and it seems likely to cause audible anomalies. I wouldn't discount these peaks as an audible factor so quickly.

Looking at the impulse plot below, it seems that 0.1 ms or 100 microseconds would be audible. I don't know for certain that an impulse test is audible, but it seems like a non-relevant test of speaker performance if it was inaudible.

A speaker is a comparatively slow mechanical system. Talking about microseconds makes sense with amps, but not with speakers. So this begs the question of what duration of a signal is audible? Does a high amplitude microsecond pulse have any residual impact on the signal a speaker plays?

It looks like the measured pulse here lasted about 0.1 milliseconds or 100 microseconds. So I'm thinking a speaker would not have time to start movement at the cone before a microsecond signal was gone (frequency and size of the driver would matter).

Anyone know or care to speculate what the audible effect of such a short pulse would be?

Would the effects of clipping hang around longer to effect the speaker's motion?

Just to be clear, I understand that the chart is in ms, but it is also apparent that the duration of the impulse is on the order of 100 microseconds.

 

[Irvrobinson]

Looking at the impulse plot below, it seems that 0.1 ms or 100 microseconds would be audible. I don't know for certain that an impulse test is audible, but it seems like a non-relevant test of speaker performance if it was inaudible.

Just to be clear, I understand that the chart is in ms, but it is also apparent that the duration of the impulse is on the order of 100 microseconds.

The pulse used to create that graph is audible, if that's what you're trying to get at.

 

[cpp]

The pulse used to create that graph is audible, if that's what you're trying to get at.

All I see on the chart is a relationship to time what about the Hz ? A person actually hearing something with a 100 microsecond impulse would depend on the Hz being presented.

 

[PENG]

Lots of good points made in the last 12 hours. I think we should step back a little and look at the practical side. Let's say we are talking about average SPL of 85 dB C scale on the RS meter. I used Diana Krall just for the sake of measuring average SPL because her CD albums tend not to have DR more than 12 to 15 dB. For my quick experiment I don't see the need to use a high end SPL meter (I could probably have borrowed one if I felt the need) but I have no reason to believe those CDs contain peaks anywhere near 120 dB.

I also would not trust the peak SPL registered by the RS meter even in the fast selection. So in addition to viewing the analog watt meters on my SA-7, I also used a Hioki meter to measure current and a Fluke 80V to measure voltage. The Fluke's spec says: "Peak capture to record transients as fast as 250 µs", definitely a far cry from ps but better than ms. If there was any peaks lasting around 0.25 to 1 ms I would have captured it, so I am quite confident with the data I obtained in my quick measurements of SPL vs Amp output. I also took some measurements using some of my best discs that contain what I would consider huge dynamic peaks. The results still indicate to me that in my world, 20W would be plenty, to cover the rarest and highest peaks that I would encounter. At the end of the day I am quite happy with the amps that I have on hand, knowing that they offer headroom that I would never need.
Back to those 120 dB picoseconds thing, I really have to do some research before saying much other than at the moment I cannot imagine what kind of media source I have that could have music signals with such short peaks. 1 picoseconds is 0.000000000001 seconds, that's looking like a delta function. A 10,000 kHz fundamental wave would have a period of 1/10,000s or 0.1 ms, so for that signal to last say 0.05 ms, it would be a just a half wave. I just can't see what musical instrument would create a very high peak for just one cycle or less. My head is exploding just thinking about so thank a lot guys, obviously special thanks to MidnightSensi2 who started the whole thing and Irv who is perpetuating this madness...

For now I will continue to think that while music signals typical do have successions of peaks, from what I have seen and heard from my media source collection, those peaks are not high enough for me to worry about clipping, not even close.

 

[ahblaza]

Thanks for hopefully ending this madness

 

[Irvrobinson]

Back to those 120 dB picoseconds thing

You should really monitor your drinking, or whatever it is you're using.

 

[cpp]

Thanks for hopefully ending this madness

Oh boy the madness is over, now off to the mountains with the family for Christmas.

Merry Christmas all

 

[AcuDefTechGuy]

Oh boy the madness is over, now off to the mountains with the family for Christmas.

Merry Christmas all

When you say "mountains", do you mean no WiFi, do Data, no iPads so you can check in on AH or emails?

 

[PENG]

When you say "mountains", do you mean no WiFi, do Data, no iPads so you can check in on AH or emails?

Well in that case please allow some of our fellow AH stay home for their Merry Christmas with family members, friends and pets, hifi gear and speakers..

 

[PENG]

The pulse used to create that graph is audible, if that's what you're trying to get at.

Agree, but can you ask your musician wife which instruments (just anything above say 4000 hz otherwise my gear can catch any peaks) would produce that pulse so I know what to expect?

 

[Irvrobinson]

Agree, but can you ask your musician wife which instruments (just anything above say 4000 hz otherwise my gear can catch any peaks) would produce that pulse so I know what to expect?

Out of curiosity, what would you estimate the time duration is of a rim shot on a snare drum?

 

[PENG]

Out of curiosity, what would you estimate the time duration is of a rim shot on a snare drum?

The fundamental freq period (1/f) would be more than a ms but the T for the harmonics could be much shorter, but we are talking peaks high enough to cause our 300/500W amps to clip so let's focus on the fundamental frequency and the lower freq that demands much more power. As you said, that's just my instant guessimate. I am interested to do more research on this thing so please share if you have some interesting links, as long they come for reliable and/or authoritative sources. You know I don't go by hearsays, neither do you I am sure.

 

[Irvrobinson]

The fundamental freq period (1/f) would be more than a ms but the T for the harmonics could be much shorter, but we are talking peaks high enough to cause our 300/500W amps to clip so let's focus on the fundamental frequency and the lower freq that demands much more power. As you said, that's just my instant guessimate. I am interested to do more research on this thing so please share if you have some interesting links, as long they come for reliable and/or authoritative sources. You know I don't go by hearsays, neither do you I am sure.

I haven't seen any research into rim shots, but they are the loudest sound I've recorded over forty years or so, and on the OmniMic I've seen peaks over 110db at six feet. They also hurt my ears. The RTA display shows that the energy from a rim show is quite high in frequency too. Obviously, a rim shot is a highly variable sound, that depends on the drum and the type of stick, and how the rim is hit, but on the last one I recorded there was a lot of energy centered on approximately 1500Hz. Assuming the Salon 2s are about 83db/watt, 110db would take over 500 watts, though I'm not sure I've ever reproduced one at really realistic levels. And, of course, in real music there are many frequency ranges being reproduced simultaneously.

 

[PENG]

Assuming the Salon 2s are about 83db/watt, 110db would take over 500 watts, though I'm not sure I've ever reproduced one at really realistic levels. And, of course, in real music there are many frequency ranges being reproduced simultaneously.

To reproduce that in realistic levels I would make sure there is at least 2kW amp power on hand and 4 pairs of Salon 2s. I never have doubt about the need of tons of power for that kind of experience. Mind you, all would be moot points if the recording equipment are not up to par and/or the recording engineer doesn't know how to do it right. In reality though, not too many people would or could tolerate average SPL of 110 dB. For me, maximum I can tolerate is around 85 dB, with peaks most likely below 110 dB. Given the speakers that I have, that's about their limits anyway.

I should add that very few mid range drivers can take 500W peaks, let alone continuous. Tweeters would take even less. I wouldn't want to listen to violin concertos at anywhere near 110 dB.

 

[Irvrobinson]

In reality though, not too many people would or could tolerate average SPL of 110 dB. For me, maximum I can tolerate is around 85 dB, with peaks most likely below 110 dB.

You're so focused on your own conclusion that you're missing my point altogether. I'm not talking about 110db average levels, I'm talking about 110db peaks for a particular instrument being played a certain way. Frankly, I don't listen with 110db peaks on my system, ever. Well, just that once when I was trying to truly simulate my wife's drum kit, but that was it. If I get really rambunctious I listen to 102-104db peaks on well-recorded contemporary jazz, but the average levels are worst-case in the mid-high 80s, though my more normal listening levels are high 70s-low 80s, and with an 80db average I see peaks on the OmniMic in the 95-100db range, and mostly in the 90s.

My point is, PENG, that if you want to reproduce live music levels for some instruments, and saxophones and trumpets fall into this category incidentally, you're going to need more than 20W of peak power into 83db/w sensitive speakers. You can make all the fun you want of this and my statements, based on some measurements that you feel satisfy your own particular needs, and I'm only pointing out that there are other environments, systems, and considerations where more power is called for, and necessary to avoid clipping.

 

[PENG]

You're so focused on your own conclusion that you're missing my point altogether. I'm not talking about 110db average levels, I'm talking about 110db peaks for a particular instrument being played a certain way. Frankly, I don't listen with 110db peaks on my system, ever. Well, just that once when I was trying to truly simulate my wife's drum kit, but that was it. If I get really rambunctious I listen to 102-104db peaks on well-recorded contemporary jazz, but the average levels are worst-case in the mid-high 80s, though my more normal listening levels are high 70s-low 80s, and with an 80db average I see peaks on the OmniMic in the 95-100db range, and mostly in the 90s.

My point is, PENG, that if you want to reproduce live music levels for some instruments, and saxophones and trumpets fall into this category incidentally, you're going to need more than 20W of peak power into 83db/w sensitive speakers. You can make all the fun you want of this and my statements, based on some measurements that you feel satisfy your own particular needs, and I'm only pointing out that there are other environments, systems, and considerations where more power is called for, and necessary to avoid clipping.

Let me be absolutely clear, I have never ever even thought of making fun of what you are saying. If it came across that way then I must have used the wrong words in a post or two. Everytime I said I only needed 20W without clipping I referred to my own 90 dB/W/M speakers. Like you said, for my own needs. I wasn't making fun of the picosecond peak things either. I meant it when I said I needed to research the topic some more. That's why I am still online. By the way, so far from what I have read, most music transient attacks seem to last longer than a ms, not microseconds or less, but I am still searching, too early to make up my mind yet.

Of course I understand with your 83 dB/W/M and other factors you need much more power. In regard to your example of needing 500W for the 110 dB peaks for the energy around your reported frequency for the rim shots, I still think few mid range drivers could handle that kind of successive peaks, the Salon2 mostly likely can, but few people own such large high end speakers. My skepticism about people claiming huge improvements is never about people who has the obvious need like yourself.

Right from start I said I was sure there were many reasons why people hear huge improvements after adding external amps, some are believable base on science, such as truly demanding speakers, large room, listening loud, defective components being compared to but there are also reasons such as Placebo, improver comparison methodology or just gross exaggeration of any such gains. Sound like I am repeating the same things I have been saying may be I should in fact monitor Alcohol content instead..

I do apologize for misquoting the 110 dB being continuous, my mistake on that for sure.

 

[KEW]

The pulse used to create that graph is audible, if that's what you're trying to get at.

Thus, we can conclude that 0.1 ms or 100 microseconds is audible. We don't know that is the shortest period, but we can conclude that a signal does not have to be several ms long to be audible. This is in the context of your statement that sounds may need to be several or 10's of ms long to be detected.

That is the point I was trying to get at from my first post of the impulse response. Apparently I did not do a very good job of it.

 

[KEW]

By the way, so far from what I have read, most music transient attacks seem to last longer than a ms, not microseconds or less, but I am still searching, too early to make up my mind yet.

By transient attack, do you mean the rise time of the amplitude? This is really interesting! I guess that makes sense for a wind or bowed instrument, but I would think a percussive instrument would be almost as fast as the pitch allows.

I'll put my money on the triangle as fastest sound of a common instrument!

I wonder how plucked compares to hit? The top end of a harp seems pretty fast!

Neat stuff!