Werner said:
Judging by the responce I think I was taken wrongly.
I was talking from a theoretical point of view based purely on the physics I know. I was stating based on a "perfect" model where there is no switching there are simply two states (I mentioned this in the original description) and therefore the transition between the two would be zero duration (hence no switching as the model is "perfect") and therefore there are two states - current flowing and current not flowing, hence not alternating.
I appologise if I wasn't that clear, thanks for the follow up though.
Isn't the issue here time coherency when reproducing an acoustic pulse ?.
Real sound sources have a fundamental attack wave front, and the relationships of the fourier components will be delayed in time according to the frequency/distance/level/humidity/temperature characteristics of air, and according to the distance to the recording microphone or to the listener.
There is no such thing as a square wave in sound, and by that I mean I don't know of any acoustic square wave generators except for the pressure wave and then the return wave caused by an explosion. (and a big explosion for a low frequency fundamental).
Real triangles and snare drums probably give the fastest slew rate that is likely to be encountered, and they sound quite strongly different according to the live listening distance.
On a reproducing system, the relative arrival times of the fourier components very strongly influence the reproduced sound.
In my experience this is a factor in the correct, distant or up front character of many loudspeakers, and electronics for that matter.
A widerange midrange helps to get acceptable vocals coherency, and with distance staggered drivers can give good overall coherency.
Duntech quote at a specified listening distance a "Pulse Coherency Factor - Propagation time error at 3.5M (11.5 ft) on tweeter axis - Less than 20µS"
This PCF is the key to getting realistic sounds.
Eric.
Real sound sources have a fundamental attack wave front, and the relationships of the fourier components will be delayed in time according to the frequency/distance/level/humidity/temperature characteristics of air, and according to the distance to the recording microphone or to the listener.
There is no such thing as a square wave in sound, and by that I mean I don't know of any acoustic square wave generators except for the pressure wave and then the return wave caused by an explosion. (and a big explosion for a low frequency fundamental).
Real triangles and snare drums probably give the fastest slew rate that is likely to be encountered, and they sound quite strongly different according to the live listening distance.
On a reproducing system, the relative arrival times of the fourier components very strongly influence the reproduced sound.
In my experience this is a factor in the correct, distant or up front character of many loudspeakers, and electronics for that matter.
A widerange midrange helps to get acceptable vocals coherency, and with distance staggered drivers can give good overall coherency.
Duntech quote at a specified listening distance a "Pulse Coherency Factor - Propagation time error at 3.5M (11.5 ft) on tweeter axis - Less than 20µS"
This PCF is the key to getting realistic sounds.
Eric.
Hi Eric,
The issue, IMHO, is very much about phase coherency. On this, I agree, as I agree that finding completely phase coherent recorded material would be difficult if not impossible in analog recordings. I think digital recording techniques offer some hope in that regard.
The sound that comes from most instruments are very rich in harmonics. Harry F. Olson wrote the book "Music, Physics, and Engineering". In this book he shows that the sound created by many instrument have harmonic content very similar to square waves, or triangle waves, or saw-tooth waves. These wave forms exist in music. The trick is capture them in a recording and reproduce them in our listening rooms.
One step at a time.
Rodd Yamas***a
The issue, IMHO, is very much about phase coherency. On this, I agree, as I agree that finding completely phase coherent recorded material would be difficult if not impossible in analog recordings. I think digital recording techniques offer some hope in that regard.
On this, I have to partially disagree. It is true that distance, humidity, and temperature will attenuate the higher frequencies more, and even change the direction of the sound, but the phase relationship doesn't change.
The sound that comes from most instruments are very rich in harmonics. Harry F. Olson wrote the book "Music, Physics, and Engineering". In this book he shows that the sound created by many instrument have harmonic content very similar to square waves, or triangle waves, or saw-tooth waves. These wave forms exist in music. The trick is capture them in a recording and reproduce them in our listening rooms.
One step at a time.
Rodd Yamas***a
Eric,
We went through this debate about a month ago. Look at this thread.
http://diyaudio.com/forums/showthread.php?s=&threadid=5063&perpage=15&pagenumber=3
I also checked through Olson's book and "Sound System Engineering" by Don and Carolyn Davis. The speed of sound through a medium is frequency independent.
Rodd Yamas***a
We went through this debate about a month ago. Look at this thread.
http://diyaudio.com/forums/showthread.php?s=&threadid=5063&perpage=15&pagenumber=3
I also checked through Olson's book and "Sound System Engineering" by Don and Carolyn Davis. The speed of sound through a medium is frequency independent.
Rodd Yamas***a
I have to disagree partially. If You have access to a mic I would suggest hooking it up to Your scope and clap Your hands in front of the mic and You`ll be surprised how much the waveform is similar to a square-wave (of course it`s no perfect square but You`ll see a fast rising leading edge followed by an relatively long durated constant "roof" of the waveform).
The issue with the square wave thing IMHO is not wether or not square waves exist in real life (music in this case) acoustic sensations but wether a square wave is a suitable test signal for evaluating time coherency of the first acoustical wavefront radiated from a speaker. And IMHO it is. How else one could check it?
I cannot understand why the square wave response in developing speakers plays such an underepresented role as it obviously does.
In all other fields of audio equipment tests the square wave response is a common and well regarded signal to evaluate errors in the time domain. Even at CD-players time optimated filters are introduced and the audibiltiy of a few degrees phase shift in the upper frequency range are heftily discussed.
In contrary to speakers. Those still produce very significant time response distortion all over the complete audio range and square waves are distorted unrecognizable when viewed at a scope screen.
Funnily enough this seem to disturb nobody (or only few as Rodd, Charles, Dave and some others here
) with speakers (can You imagine an amp with such strange behavier would be accepted?
) and this drawbacks suddenly are justified while arguing this can`t be heard anyway and square waves don`t exist in music. Kinda wierd IMHO.Hence there seem to be no real interest to proof by sientific research to which extend time distortion can be tolerated respectively until it`s inaudible. The last extensive research works about this topic are decades ago as far I know (I´d be happy if somebody proof me wrong).
Particular today with far better possibities in technology and science and the progress in speaker technology as well, it`s time for some new and serious research work.
I´m convinced (maybe since I heard the Manger driver) that correct time response is as important as frequency response. Although a perfect square response may remain a dream (and due to for instance mechanical restrictions it`s impossible) at least it`s possible to improve current things to some audible extends.
I believe correct time response stands in direct relation to location, distance and size perception of acoustical sensations. We all know our ears analyze and judge wavefronts upon their sequence of arrival (time) of the sound pressure. This allows us to determine position, distance of the source and also its size to some extend.
What seems to be scientifical proofed meanwhile is that this happens mainly inbetween a time window from 20µs to 600µs of the first (direct) arriving wave. For lower frequencies our ears seem to be much less sensitive to this effect (BTW that`s the reason why the single subwoofer concept works at all). So our ears for higher frequencies seems to work time related pressure (amplitude) sensitive while for lower frequencies rather pressure (amplitude) is of importance only (IMHO this is an indication why the woofer supported full range driver approuch with low xover point is so promising and in fact works so good in many cases).
This means there is no need to achieve perfect time response from DC to infinity and makes it a whole lot easier to construct a "perfect" time response speaker.
Besides physical distinct driver locations in multiway speakers (restricted listener area due to interferance effects) another problem are the unfortunaltely necessary crossovers. Because there are only a few kind of crossovers which by their nature allow to achieve perfect transient response and because the time behaviour (bandwith limitations) of the individual drivers have to be taken into account also (integrated or equalized) this makes the task quite difficult but not impossible.
IMHO multiway speakers with more than 3 ways are hardly to make perfect in time response.
But a three way system with a full range (or wide range) driver as middle driver and not too high xover (4-5kHz max.) for the tweeter in junction with the active subtractive XOs (be it the active filler driver concept or the equalized versions) which presents Charles (phase-accurate) in this thread:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=5435
if not perfect at least it looks promising to have better time response than "ordinary" speakers.
Only when we have built a such a "perfect" time response speaker and extensively listened and compared it to common designs we`ll be able to judge if it was worth the hole hassle.
Anyway IMHO it`s worth a try.
Yet, the end of my (couple of)
You said it cocolino. Kudos for your entire post!
I've always believed that excellent square wave response should be a primary goal of speaker designers. It didn't bother me that people told me I was crazy and recently on this forum that it couldn't be done with analog filters. I think it can and this will be my next project.
I'm going to toss together an active second order bandpass filter with F3s of 30Hz and 20KHz and see what the output looks like with a 30Hz square input. The goal is to have the output of the speaker come as close to this as possible.
Does this seem reasonable and who's with me?
I've always believed that excellent square wave response should be a primary goal of speaker designers. It didn't bother me that people told me I was crazy and recently on this forum that it couldn't be done with analog filters. I think it can and this will be my next project.
I'm going to toss together an active second order bandpass filter with F3s of 30Hz and 20KHz and see what the output looks like with a 30Hz square input. The goal is to have the output of the speaker come as close to this as possible.
Does this seem reasonable and who's with me?
Pulse Testing
Rodd, thanks I stand corrected.
Dave, yes I have to agree, but this does give me the question of - Is it better to drive all speakers simultaneously and position them appropriately, or is it better to mount them in the same plane and delay the electronically as is appropriate to achieve correct pulse wavefront ?. (I'm thinking multiway DSP here)
Notably the Duntech passive/staggered baffle soloution works very well indeed.
Christof, I should have stated that low frequency sqare waves cannot exist, because of the impossibilty of pressurising and depressurising a significant listening environment (live or replay) for substantial periods (25mS cycle time for example for 40 Hz).
Did you measure the duration of the flat top of the handclap waveform ?.
Perhaps a measurement of live sound sources can establish a frequency above which live sound sources do indeed emit a square or rectangular wave front, and this fundamental frequency be the yardstick for acceptable reproduced wavefronts.
Non-resonant and correct transient behaviour I agree are more important parameters than absoloute flatness of sinewave tested frequency response, although they are of course somewhat interactive.
Two subwoofers are much better than just one subwoofer IME.
Eric.
Rodd, thanks I stand corrected.
Dave, yes I have to agree, but this does give me the question of - Is it better to drive all speakers simultaneously and position them appropriately, or is it better to mount them in the same plane and delay the electronically as is appropriate to achieve correct pulse wavefront ?. (I'm thinking multiway DSP here)
Notably the Duntech passive/staggered baffle soloution works very well indeed.
Christof, I should have stated that low frequency sqare waves cannot exist, because of the impossibilty of pressurising and depressurising a significant listening environment (live or replay) for substantial periods (25mS cycle time for example for 40 Hz).
Did you measure the duration of the flat top of the handclap waveform ?.
Perhaps a measurement of live sound sources can establish a frequency above which live sound sources do indeed emit a square or rectangular wave front, and this fundamental frequency be the yardstick for acceptable reproduced wavefronts.
Non-resonant and correct transient behaviour I agree are more important parameters than absoloute flatness of sinewave tested frequency response, although they are of course somewhat interactive.
Two subwoofers are much better than just one subwoofer IME.
Eric.
Christoph,
I second Bill's response, excellent post. I believe the idea of the a wide band mid-range driver to provide the bulk of the phase coherency is a good idea. It should work well at modest levels, but you may have issues with IM distortion at higher output levels. Don't take me wrong; I'm not criticizing your approach. I'm just saying we have to pick our battles. I started with horns with their shortcomings and attributes. Our goals, I think, are the same.
Bill,
Please keep us informed on your project. There is clearly interest in this issue.
Eric,
opworm: But, I prefer the idea of aligning the "acoustic centers" of the drivers in a vertical line (actually an arc equal-distant from the listener). Then use electronics to time synchronize the signals to each driver.
Rodd Yamas***a
I second Bill's response, excellent post. I believe the idea of the a wide band mid-range driver to provide the bulk of the phase coherency is a good idea. It should work well at modest levels, but you may have issues with IM distortion at higher output levels. Don't take me wrong; I'm not criticizing your approach. I'm just saying we have to pick our battles. I started with horns with their shortcomings and attributes. Our goals, I think, are the same.
Bill,
Please keep us informed on your project. There is clearly interest in this issue.
Yes, it's a reasonable endeavor. I will contribute where I can.
Eric,
Certainly, both have been done to various degrees of success. I always liked the sound of the Dahquist DQ10's, which used to have 5 different baffles in various different planes for 5 different drivers. It's dangerous to say that one way is correct the other is wrong.
opworm: But, I prefer the idea of aligning the "acoustic centers" of the drivers in a vertical line (actually an arc equal-distant from the listener). Then use electronics to time synchronize the signals to each driver.Rodd Yamas***a
OK. I'm going to start a new thread called the "Square Wave Project".
This will be a joint effort and lots of input will be needed.
What I'm going to hope for is that everybody that gets in on it will stick to the subject and not get sidetracked with comments about enclosure bracing, the cute little town in Germany, how much beer you can drink and all the other things that make threads fall apart.
This will be a joint effort and lots of input will be needed.
What I'm going to hope for is that everybody that gets in on it will stick to the subject and not get sidetracked with comments about enclosure bracing, the cute little town in Germany, how much beer you can drink and all the other things that make threads fall apart.
cocolino said:
I have for the last 20 plus years held with the "mostly full-range" philosophy -- a single driver covering the majority of the spectrum. With much of the spectrum coming from a single source time coherency -- at least in this range -- is maximized.
Until recently this was executed with ESLs actively crossed to woofers (at 100-125 Hz). More recently i have been seeing what i can do with FR cones (and as much by accident as by design Frugal-phile(tm) class drivers). [These easier to drive speakers have also allowed me to play around with small-output power amps]. A modestly priced 3-way with XOs at about 100 (active) and 10k provides a level of performance & musicality that has quite stunned me, especially when one considers the small amount of money that has gone into these. I am hoping that the next iteration (alnico versions of the same driver that cost even less) will allow me to eliminate or push up the HF XO. I also intend on playing with a subtractive XO in the bottom to see if i can lift the slight veil that is inserted with the 4th order LR XOs now in use.
dave
Re: Pulse Testing
Each approach has its own sets of compromises, so i'd say that there probably is no best way, just those that stand-out because they are well implemented.
For instance, digital/DSP has promise but isn't quite there yet in terms of keeping itself out of the way. Stepped baffles have the problems of things bouncing off the steps. Full-ranges have their problems, as do coaxes. The best ones are the ones that take most advantage of the pros and ameriolate the cons.
dave
mrfeedback said:
Each approach has its own sets of compromises, so i'd say that there probably is no best way, just those that stand-out because they are well implemented.
For instance, digital/DSP has promise but isn't quite there yet in terms of keeping itself out of the way. Stepped baffles have the problems of things bouncing off the steps. Full-ranges have their problems, as do coaxes. The best ones are the ones that take most advantage of the pros and ameriolate the cons.
dave
I'm no expert, but I disagree that you can produce a faithful square wave in the real world, based purely on the fact that infinity is not a number but a concept - one for that matter that cannot ever be achieved in a Euclidian universe.
For a square wave to be created it must have a zero rise time on the leading edge which inherently means an infinitely fast change of state (be it in an electrical signal or the compression of air molecules), then a stable state (which too is in theory only attainable through an infinitely long state) and then another infinitely fast change of state at the following edge of the pulse.
I do however agree that it is perfectly acceptable to create approximations - by this I mean trapezia with very steep leading and following edges (although it is questionable how flat the peak of the wave is).
I'm basing this on my knowledge of theoretical physics if you want a second opinion find another physicist?
<center>
___________________________________
</center>
Also with regards to seeing a square wave on a scope - this would be a trapezium possibly even if the input signal was a square wave as the scope is not perfect (i.e. doesn't have a sampling rate of infinity!)
<center>
___________________________________
</center>
For a square wave to be created it must have a zero rise time on the leading edge which inherently means an infinitely fast change of state (be it in an electrical signal or the compression of air molecules), then a stable state (which too is in theory only attainable through an infinitely long state) and then another infinitely fast change of state at the following edge of the pulse.
I do however agree that it is perfectly acceptable to create approximations - by this I mean trapezia with very steep leading and following edges (although it is questionable how flat the peak of the wave is).
I'm basing this on my knowledge of theoretical physics if you want a second opinion find another physicist?
<center>
___________________________________
</center>
Also with regards to seeing a square wave on a scope - this would be a trapezium possibly even if the input signal was a square wave as the scope is not perfect (i.e. doesn't have a sampling rate of infinity!)
<center>
___________________________________
</center>
annex666 -
"Also with regards to seeing a square wave on a scope - this would be a trapezium possibly even if the input signal was a square wave as the scope is not perfect (i.e. doesn't have a sampling rate of infinity!)"
Cros are rated by the frequency response of the vertical amplifiers that drive the vertical deflection plates of the display crt.
A modest 20 Mhz Cro has displayable risetime 1000 times that required at the highest audio frequency (20 kHz) and will display such a waveform perfectly.
Bill, yes good idea.
Eric.
"Also with regards to seeing a square wave on a scope - this would be a trapezium possibly even if the input signal was a square wave as the scope is not perfect (i.e. doesn't have a sampling rate of infinity!)"
Cros are rated by the frequency response of the vertical amplifiers that drive the vertical deflection plates of the display crt.
A modest 20 Mhz Cro has displayable risetime 1000 times that required at the highest audio frequency (20 kHz) and will display such a waveform perfectly.
Bill, yes good idea.
Eric.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.