Yes. I have many amplifiers. None has had difficulty driving the speaker. I have had this darlington amp and an opamp-based CFP which is probably not so stable (nature of darlington and opamp-driven CFP) but there was no problem.
Oscillation can change tonality yes, but what I like most with the speaker is the 'right' tonality so no problem either.
There was this famous distortion test on the net (I forgot the name of the expert who initiate it). The hardest test on it (where as you can guess passed only by a few select people) was still too easy for me (I only tried once).
I don't know what's wrong with my ears (left ears actually), even on a test with ridiculously low THD (by an expert who knows what he is doing) I can pass easily. Honestly I don't think it is possible, but I think the explanation is that there was something that was audible beside the THD alone (It could be something other than THD).
I prefer linearity in everything. In this case, THD at 20Hz, 1kHz and 20kHz are more or less the same. I also like THD at outside the audible range to be very low. When I designed an amp, of course I don't expect the THD to be very low when there is only 2 stages (class-A). But when I design complex amps, I don't want to go very low on THD because I want to prioritize on other parameters. 0.001% at 1kHz is sufficient for me (I don't need to go lower unless it is a 'free lunch').
You are right, they are very basic. You asked about the lowest THD I can discern by ears. That question is probably more important than you may have imagined.
We can measure a lot of things like step responses. But what is the point with all the numbers and visuals if we don't know how they affect our perception??
If we know how they affect our ears, we know what numbers or visuals we need to achieve. A classical example is about phase. Everyone can argue whether it is important or not, audible or not, but I know how to implement this in amplifier design and speaker design because my ears are more sensitive than average people.
Measuring the impedance curve of your speaker is easy. The software is free and the hardware is only a metal resistor. There is right now an ongoing thread here about impedance measurement. What is holding you back from measuring it? (Unless you can't handle the truth that your notch filter is either shorting your amp or doing nothing, that's the only reason I can see.)
Your answer demonstrates that you did not understand THD at all. So time for the next link:
Blog - How Much Distortion Can We Hear With Music? | Axiom Audio
You will never be able to tell if your amp has 0.001% THD or 0.01% by ear. You will have a really high end pair of speakers if they never exceed 1% THD in the audible spectrum, so any harmonic distortion from your amp is simply buried under the sea of speaker distortion. (Does 1.001% sound any better than 1.01%? Are you sure you can tell and even if you are does it matter any more?)
It is for the same reason that the inductor does not matter - not that it does not impact the signal, but because the impact is just a tiny fraction of the ailment introduced by the driver itself. The impact of an inductor is worse than an amp, much worse, but the driver itself is much much much worse, such and such that the inductor becomes the least of worries.
Your answer demonstrates that you did not understand THD at all. So time for the next link:
Blog - How Much Distortion Can We Hear With Music? | Axiom Audio
You will never be able to tell if your amp has 0.001% THD or 0.01% by ear. You will have a really high end pair of speakers if they never exceed 1% THD in the audible spectrum, so any harmonic distortion from your amp is simply buried under the sea of speaker distortion. (Does 1.001% sound any better than 1.01%? Are you sure you can tell and even if you are does it matter any more?)
It is for the same reason that the inductor does not matter - not that it does not impact the signal, but because the impact is just a tiny fraction of the ailment introduced by the driver itself. The impact of an inductor is worse than an amp, much worse, but the driver itself is much much much worse, such and such that the inductor becomes the least of worries.
Cyberstudio, I have never mentioned that I didn't measure my speaker impedance 🙂 It's a long time ago. My intuition says that I have started building speakers before you were born.
That's an unfair assumption, although I don't see how that is relevant.
Technology does not stand still. If you are still clinging onto techniques that are several decades old, you will need to do yourself a favor and update your knowledge. Nowadays people are designing with the proper equipment and tools. Free (and paid of course) software is available to measure frequency response, phase, distortion, impedance, reverberation. Experience is only an excuse for the unwillingness to keep up with the times.
I am getting old and I can no longer hear 20kHz. My son is too young to help me with listening tests but my wife is still young and she will be able to help me. Now, many members here did start building speakers before I was born, for sure, and they will suffer even more than myself on the hearing department. We all get old. But it is not a requirement to have a golden ear to be a good speaker designer. What is required instead is a solid understanding of physics, math, acoustics, psychology of perception and expertise in the use of the proper measurement equipment and measurement technique.
Beethoven hardly heard a thing by the time he wrote his 7th Symphony and completely deaf when he wrote the 9th.
We have no shortage of good speaker designers here and one of the crafts I have been lurking around here and learning from them is their ability to read tea leaves from measurement graphs. Literally reading a step response graph and gain insight into how it is going to sound in the real world.
I observed you have a tendency that because you cannot do something you infer other people also cannot do it. For example, by your own admission you cannot relate a measurement graph to a real world sound, and also by your own admission you cannot pull off a passive crossover design using steep slopes. Since you cannot do them, you conclude other people can't, too, but that again is an unfair assumption.
Hehehe so we thought the same thing of each other. English is not my language so may be it made you confused and drew wrong conclusions.What I have said or inferred are quite the opposites.
I deduced from your eagerness with all these measurement stuffs that you are new to this hobby. You showed me websites for beginners, was that because you have assumed that I'm a beginner? Or is it because you are a beginner that you think highly of such things?
But let me tell you that I don't mind you are thinking that I don't understand anything. I thought that you were interested to find out about what I know that other people don't.
Sometimes it's better to postpone a reply and let things settle for awhile. In general a "crossoverless design" approach is worth investigating as a curiosity, as is building whatever makes one happy. The general agreement and recognition is not the most important thing in this hobby of ours, imo.
Thank you for your reply Lojzek. I did learn something new from johnego - I did not think of "crossoverless" seriously as a high-end approach until he pointed me to the Reference 3A MM de Capo BE.
Even beginners understand the importance of measurements in loudspeaker design. To say "I deduced from your eagerness with all these measurement stuffs that you are new to this hobby" is an attack on all experienced participants in this community.
For reasons obvious to us, the "crossoverless" methodology does not mix well with parallel notch filters.
Those are my two real grievances. I take no issue with the crossoverless approach itself.
I have temporarily added johnego to my ignore list - no offense intended to him but you are right that we shall let it sit.
Even beginners understand the importance of measurements in loudspeaker design. To say "I deduced from your eagerness with all these measurement stuffs that you are new to this hobby" is an attack on all experienced participants in this community.
For reasons obvious to us, the "crossoverless" methodology does not mix well with parallel notch filters.
Those are my two real grievances. I take no issue with the crossoverless approach itself.
I have temporarily added johnego to my ignore list - no offense intended to him but you are right that we shall let it sit.
Last edited:
I would be very interested in more conversation directly related to the OP’s original question - both for active and passive crossovers and whether this too makes a difference in the determination.
If measurement results are the primary input to the decision, what characteristics in the results would lead one to start working with one crossover alignment over another?
Hi rscamp,
Well as stated before there is first the behavior of the drivers themself.
Naughty peaks in the frequency response (cone break up), ability to withstand power applied to achieve the target spl ( mainly a problem for fullrangers in the low end, tweeters or CD/horn combo).
Second there is the lobing issue which can be a source of troublesome behavior ( imbalance of the reverberant soundfield in room, non even coverage in the vertical and horizontal plane creating hotspot and coldspot - starved frequency wise).
Third there is subjective choice of designer: what do you want to emphasis or not. Want a (near) perfect transient response? A wide coverage angle or a narrow one? Absolute spl capability, etc, etc,...
There is multiple school of thoughts about that and it is not nescessarely easy to explain (at least for me) because each choice you make induce compromise and more or less impose some choice related to the way you decided to choose.
For example if you want shallow slope (6db/octave) you ll have to find drivers which behave well outside their usable frequency range as there will be a lot of overlap between them. You ll probably have to struggle with their diameter if you want something approaching point source behavior ( 1/4 wavelength center to center distance at fc if you want acoustical coupling).
This may compromise the absolute spl depending on x over frequency. The driver behavior related to directivity will come into play (this is diameter related) and you ll have to balance this too into the equation...
If you target to cover a large area ( live sound or theater for example) you ll have to pay attention to lobing issue so you can t use whatever kind of slope and should at the same be sure the drivers won t be damaged by power applied to them ( because you ll have potential use of high power to achieve high spl in this case) so you must forget shallow slope...
Then there is the audibility of the kind of filter (solution) you will use as all have strength and weakness and some may like or dislike some kind ( LR2 are favored by some, 6db/octave by others, some claims preringing in dsp FIR is objectionable some don t...).
Everybody have his/her own preference about that once you know what to listen...
One good way to discover that is to read the WhitePaper at Rane site. They have LR and Bessel kind covered and they are understandable for non mathematicians (but there is error in the LR one for example but nothing to serious).
Active / passive well this is a can of worm... let s say that an active solution can do same things that a passive solution. Active DSP can do things passive can t (FIR).
Each solution have its pro and cons but in some case one will be superior to the other ( for low freq xover you may find that active is at the end easier and less costly for example, passive may be more coherent in some case...).
Imo active dsp is the way to go if your source is mainly digital. This can impose some issue though...
Well as stated before there is first the behavior of the drivers themself.
Naughty peaks in the frequency response (cone break up), ability to withstand power applied to achieve the target spl ( mainly a problem for fullrangers in the low end, tweeters or CD/horn combo).
Second there is the lobing issue which can be a source of troublesome behavior ( imbalance of the reverberant soundfield in room, non even coverage in the vertical and horizontal plane creating hotspot and coldspot - starved frequency wise).
Third there is subjective choice of designer: what do you want to emphasis or not. Want a (near) perfect transient response? A wide coverage angle or a narrow one? Absolute spl capability, etc, etc,...
There is multiple school of thoughts about that and it is not nescessarely easy to explain (at least for me) because each choice you make induce compromise and more or less impose some choice related to the way you decided to choose.
For example if you want shallow slope (6db/octave) you ll have to find drivers which behave well outside their usable frequency range as there will be a lot of overlap between them. You ll probably have to struggle with their diameter if you want something approaching point source behavior ( 1/4 wavelength center to center distance at fc if you want acoustical coupling).
This may compromise the absolute spl depending on x over frequency. The driver behavior related to directivity will come into play (this is diameter related) and you ll have to balance this too into the equation...
If you target to cover a large area ( live sound or theater for example) you ll have to pay attention to lobing issue so you can t use whatever kind of slope and should at the same be sure the drivers won t be damaged by power applied to them ( because you ll have potential use of high power to achieve high spl in this case) so you must forget shallow slope...
Then there is the audibility of the kind of filter (solution) you will use as all have strength and weakness and some may like or dislike some kind ( LR2 are favored by some, 6db/octave by others, some claims preringing in dsp FIR is objectionable some don t...).
Everybody have his/her own preference about that once you know what to listen...
One good way to discover that is to read the WhitePaper at Rane site. They have LR and Bessel kind covered and they are understandable for non mathematicians (but there is error in the LR one for example but nothing to serious).
Active / passive well this is a can of worm... let s say that an active solution can do same things that a passive solution. Active DSP can do things passive can t (FIR).
Each solution have its pro and cons but in some case one will be superior to the other ( for low freq xover you may find that active is at the end easier and less costly for example, passive may be more coherent in some case...).
Imo active dsp is the way to go if your source is mainly digital. This can impose some issue though...
Last edited:
For a starter, I would refer rscamp and the OP to a previous thread:
Differences in tweeter's sound by different kind of crossover
What I was trying to say in the beginning was that by the time the system parameters are defined we will have been so severely constrained that the choice of crossover type will become extremely limited, and I cited the classic 2-way as the example.
To elaborate, for true full range performance typically a 7" woofer is used to get sufficient bass. However that woofer may not go up to the midrange very comfortably. So a "low going" tweeter is used, with a low crossover point and a steep slope crossover to protect the tweeter and to cut down tweeter distortion. First and second order will not be possible and we will be looking at 3rd order Butterworth or 4th order Linkwitz Riley. Also the 7" is physically large and if the tweeter is large as well will lead to a complex lobing pattern, which can be reduced, but not completely mitigated, by a steep crossover slope.
If you pair the same tweeter with a 4" "woofer" then because there is more "overlap" between the two drivers the crossover point can be set higher and we may now have the opportunity for a 2nd order Linkwitz-Riley crossover. But that will not have sufficient bass and you will have to turn it to a 3-way for a satisfying experience.
By now it becomes obvious that a 3-way will have far greater freedom in driver selection and crossover point/type selection.
Differences in tweeter's sound by different kind of crossover
What I was trying to say in the beginning was that by the time the system parameters are defined we will have been so severely constrained that the choice of crossover type will become extremely limited, and I cited the classic 2-way as the example.
To elaborate, for true full range performance typically a 7" woofer is used to get sufficient bass. However that woofer may not go up to the midrange very comfortably. So a "low going" tweeter is used, with a low crossover point and a steep slope crossover to protect the tweeter and to cut down tweeter distortion. First and second order will not be possible and we will be looking at 3rd order Butterworth or 4th order Linkwitz Riley. Also the 7" is physically large and if the tweeter is large as well will lead to a complex lobing pattern, which can be reduced, but not completely mitigated, by a steep crossover slope.
If you pair the same tweeter with a 4" "woofer" then because there is more "overlap" between the two drivers the crossover point can be set higher and we may now have the opportunity for a 2nd order Linkwitz-Riley crossover. But that will not have sufficient bass and you will have to turn it to a 3-way for a satisfying experience.
By now it becomes obvious that a 3-way will have far greater freedom in driver selection and crossover point/type selection.
Last edited:
Thank you krivium and cyberstudio for your comprehensive and immediate responses to my query!
I am somewhat aware of basics such as avoiding regions of frequency where a driver misbehaves or would have a power handling issue. I am a bit less familiar with the lobing issue and much less familiar with what impact a filter choice would have on reproduction characteristics other than frequency response. It sounds like I'll have to learn these impacts and decide priorities for myself.
Are there any rules of thumb? Just as an example off the top of my head, if a midwoofer driver exhibits a resonance near its upper limit at X Hz then it is best to select a 1st, 2nd, 3rd, etc. LP filter that would attenuate the peak by at least Y dB?
Do you mean this site? Rane Library
Are you referring to overall driver diameter rather than the diameter of the sound radiating surface? With most tweeters for example, there is a baffle with radius much larger than the diaphragm.
I was thinking of starting out with active using a miniDSP due to the apparent speed and flexibility in making changes.
I'll take a look!
I was thinking a bit along these lines with regard to frequency coverage. I was looking to select drivers for a 2-way system with characteristics that would provide quite a bit of overlap in the crossover region enabling more flexibility and experimentation with crossover frequencies, slopes, etc. This would tend to limit midwoofer size and require a tweeter with a low Fr and perhaps a larger diameter. I should probably start a new thread on that rather than going too far off topic here.
I am somewhat aware of basics such as avoiding regions of frequency where a driver misbehaves or would have a power handling issue. I am a bit less familiar with the lobing issue and much less familiar with what impact a filter choice would have on reproduction characteristics other than frequency response. It sounds like I'll have to learn these impacts and decide priorities for myself.
Are there any rules of thumb? Just as an example off the top of my head, if a midwoofer driver exhibits a resonance near its upper limit at X Hz then it is best to select a 1st, 2nd, 3rd, etc. LP filter that would attenuate the peak by at least Y dB?
Do you mean this site? Rane Library
Are you referring to overall driver diameter rather than the diameter of the sound radiating surface? With most tweeters for example, there is a baffle with radius much larger than the diaphragm.
I was thinking of starting out with active using a miniDSP due to the apparent speed and flexibility in making changes.
I'll take a look!
I was thinking a bit along these lines with regard to frequency coverage. I was looking to select drivers for a 2-way system with characteristics that would provide quite a bit of overlap in the crossover region enabling more flexibility and experimentation with crossover frequencies, slopes, etc. This would tend to limit midwoofer size and require a tweeter with a low Fr and perhaps a larger diameter. I should probably start a new thread on that rather than going too far off topic here.
Hi,
Yes Rane library. Worth reading the full notes! Professional no nonsense approach.
The ones i was refering to are under the active x over section.
Mini dsp is a nice cheap way to enter active. Don t forget to invest into a measurement microphone and REW.
About driver diameter i was refering to both the overall physical size ( relevant within the CtC 1/4 distance at fc) and the radiating area. The issue will come with woofer rather than tweeter: from omni radiation the higher you go in freq the narrower the radiation pattern become.
If you followed the "econowave" example you may have noticed that the waveguides (seos) does have a recommended lowest frequency and a recommended diameter for woofer associated. This corespond to the point where there is a directivity match between the beaming of woofer and waveguide. For example a seos 12" will match a 12" woofer at around 1200hz, both have a radiation pattern of 90° (+/- 45°) at this frequency, a seos 15" and a 15" at 900hz/1khz, a seos 18" at around 800hz, etc,etc...
If you try to match to a dome tweeter a 15" at something like 2khz you ll run into trouble (from a radiation point of view) as the tweeter will radiate much wider than the woofer...
You should try to read about Dunlavy interviews. He was on the shallow slope camps and had some interesting point about that.
Yes Rane library. Worth reading the full notes! Professional no nonsense approach.
The ones i was refering to are under the active x over section.
Mini dsp is a nice cheap way to enter active. Don t forget to invest into a measurement microphone and REW.
About driver diameter i was refering to both the overall physical size ( relevant within the CtC 1/4 distance at fc) and the radiating area. The issue will come with woofer rather than tweeter: from omni radiation the higher you go in freq the narrower the radiation pattern become.
If you followed the "econowave" example you may have noticed that the waveguides (seos) does have a recommended lowest frequency and a recommended diameter for woofer associated. This corespond to the point where there is a directivity match between the beaming of woofer and waveguide. For example a seos 12" will match a 12" woofer at around 1200hz, both have a radiation pattern of 90° (+/- 45°) at this frequency, a seos 15" and a 15" at 900hz/1khz, a seos 18" at around 800hz, etc,etc...
If you try to match to a dome tweeter a 15" at something like 2khz you ll run into trouble (from a radiation point of view) as the tweeter will radiate much wider than the woofer...
You should try to read about Dunlavy interviews. He was on the shallow slope camps and had some interesting point about that.
Bookmarked. Thanks!
Budgeted.
Long sonar arrays that we used to build have large aperture and high directivity so I'm a bit familiar with that concept. I need to read up on the interference between the two sources.
This one? Loudspeaker designer John Dunlavy: By the Numbers... | Stereophile.com
Yes John Dunlavy.
This is the interview i was thinking about.
I tend to agree with his design's philosophy but i took an other way than shallow slope passive: FIR.
They don t store energy and give very steep slope. Which help with 'conventional' design and open the choice of driver much more (his approach require some very well behaved drivers).
😉
If you want to have some visual feedback of what happen at FC with passive crossover of different slope and driver layout download XDir which is a freeware and offer a way to determine lobing behavior in the vertical plane ( the most troublesome).
You ll see that playing with it it offer some interesting approach like in d'appolito design where you can lessen some artefact like floor/ceiling bounce if used with correct fc and ctc of drivers...
Talking about d appolito and destructive interference between driver principle here is a good reading:
Biro Technology
Linkwitz lab site have some very good info on the subject too ( even if i don t agree with everything).
Another approach which is very interesting in its use of FIR is the Horbach-Keele way, geometrical layout and special FIR filter profil to control vertical directivity. Very clever.
About sim, Lspcad or other 'big' simulation/CAD dedicated software offer even more accuracy and simulation potential but learning curve is steep and you ll need to invest money!
This is the interview i was thinking about.
I tend to agree with his design's philosophy but i took an other way than shallow slope passive: FIR.
They don t store energy and give very steep slope. Which help with 'conventional' design and open the choice of driver much more (his approach require some very well behaved drivers).
😉
If you want to have some visual feedback of what happen at FC with passive crossover of different slope and driver layout download XDir which is a freeware and offer a way to determine lobing behavior in the vertical plane ( the most troublesome).
You ll see that playing with it it offer some interesting approach like in d'appolito design where you can lessen some artefact like floor/ceiling bounce if used with correct fc and ctc of drivers...
Talking about d appolito and destructive interference between driver principle here is a good reading:
Biro Technology
Linkwitz lab site have some very good info on the subject too ( even if i don t agree with everything).
Another approach which is very interesting in its use of FIR is the Horbach-Keele way, geometrical layout and special FIR filter profil to control vertical directivity. Very clever.
About sim, Lspcad or other 'big' simulation/CAD dedicated software offer even more accuracy and simulation potential but learning curve is steep and you ll need to invest money!
Last edited:
Well, I have now read through most of the references kindly provided. I was met with a mixture of things I already knew, things I had previously misunderstood, things I hadn't considered, things I just didn't know and things that simply made me slap my forehead and go "D'oh!". 🙂
It looks like I was generally on the right track with driver selection and placement for my next project but now I understand more about why. And overall I would have to say there was more of a leaning towards active speaker systems in the references than I expected given the majority of projects in this forum.
With regard to crossovers, there is no single, best solution. But I guess I have a bit better idea what is likely to work well with the target 2-way configuration. I look forward to experimenting. All of the above is fodder for a new thread if I can pick your brains on a few details before moving forward.
It looks like I was generally on the right track with driver selection and placement for my next project but now I understand more about why. And overall I would have to say there was more of a leaning towards active speaker systems in the references than I expected given the majority of projects in this forum.
With regard to crossovers, there is no single, best solution. But I guess I have a bit better idea what is likely to work well with the target 2-way configuration. I look forward to experimenting. All of the above is fodder for a new thread if I can pick your brains on a few details before moving forward.
Hi Rscamp,
Yes there is no better. It is all about design goal, philosophy behind it and... compromise!
About active or passive here again there is no absolute better way. Each solution have theorical and practical advantage it is up to the designer to make choice.
I m clearly in the active dsp camp as i have a background in pro audio and given the constraint of the field this solution is often considered easier to apply, solve many issue you ll face ( amplifier power needed for PA make passive approach hopeless for good sound ).
That said and despite my own theorical and pratical preference the best loudspeaker i ever heard were passive... so don t be an extremist in the choice you made and stay open minded to other approach. There is a lot to learn. Always.
And this is the 'raison d etre' of this forum. So lucky we are to have access to such knowledge ( some very, very smart people in here) and variety of point of view.
Yes there is no better. It is all about design goal, philosophy behind it and... compromise!
About active or passive here again there is no absolute better way. Each solution have theorical and practical advantage it is up to the designer to make choice.
I m clearly in the active dsp camp as i have a background in pro audio and given the constraint of the field this solution is often considered easier to apply, solve many issue you ll face ( amplifier power needed for PA make passive approach hopeless for good sound ).
That said and despite my own theorical and pratical preference the best loudspeaker i ever heard were passive... so don t be an extremist in the choice you made and stay open minded to other approach. There is a lot to learn. Always.
And this is the 'raison d etre' of this forum. So lucky we are to have access to such knowledge ( some very, very smart people in here) and variety of point of view.
Last edited:
- Status
- Not open for further replies.