to godey
Interesting work! Would you consider making the processed files available, either for playback on a sound card or (beware!) from a CD?
Wouldn't FIR filters be a better solution? They have constant delay (and delay may be added on purpose to compensate for driver/baffle geometry) and they can be made as steep as one may like. What is the issue there? Pre-ringing?
If you want to stick with 4th order IIR crossovers, that makes choices tough. There are some excellent 12 and 13 cm Scan Speak and Excel midranges, but they would be pretty directional at 8 kHz, yet 100 Hz being about 25 Hz above their free air resonance, they'd probably add enough of their own phase signature from the 12 dB/oct sealed box rolloff. Going to 8-10 cm midranges will reduce beaming but is going to acerbate the 100 Hz problem.
Even a Manger MSW has its resonance at 70 Hz, so not much of a chance there. The Neo 8 planar speakers could probably streched to go to 200 Hz with little phase shift as their resonance is very weak, but even in a line array, I wouldn't want to use them below resonance because of the excursion.
Did you have any specific drivers in mind that would match your needs?
Eric
Interesting work! Would you consider making the processed files available, either for playback on a sound card or (beware!) from a CD?
Wouldn't FIR filters be a better solution? They have constant delay (and delay may be added on purpose to compensate for driver/baffle geometry) and they can be made as steep as one may like. What is the issue there? Pre-ringing?
If you want to stick with 4th order IIR crossovers, that makes choices tough. There are some excellent 12 and 13 cm Scan Speak and Excel midranges, but they would be pretty directional at 8 kHz, yet 100 Hz being about 25 Hz above their free air resonance, they'd probably add enough of their own phase signature from the 12 dB/oct sealed box rolloff. Going to 8-10 cm midranges will reduce beaming but is going to acerbate the 100 Hz problem.
Even a Manger MSW has its resonance at 70 Hz, so not much of a chance there. The Neo 8 planar speakers could probably streched to go to 200 Hz with little phase shift as their resonance is very weak, but even in a line array, I wouldn't want to use them below resonance because of the excursion.
Did you have any specific drivers in mind that would match your needs?
Eric
Linn Olson mentioned about 300Hz – 3kHz range as no crossover area. Agreed. Shall we assign the crossover point to lets say 3,5kHz then (many do so) because Mr. Olson didn’t refer any further? Not in my opinion. What about the rest of the ear sensitive frequency area? I have great respect towards Olson’s DIY efforts and musings but everything what comes from some well-respected pioneer, shouldn’t be adopted blindly without giving a thought about the background of their sayings. Wasn’t Mr Olson the same guy who planned to build simple, inexpensive and easy to build monitors at first and ended up with huge, complicated, multidriver design he could be satisfied untill powering them with Ongakus or Reichert Silver 300B. Sorry Ariel builders can’t carry the flame any further – never heard a pair of them and that’s not the issue here.
Hey, who started the two-way alignment approach what has became so popular in today's loudspeaker business? Somebody who as been around in music reproduction hobby/business much longer than I can correct me but I think they were PA people. They, I think, had the same reasons back then, as home speaker manufacturers have today – cut down the cost, weight and size of the product by reducing the number of drivers used. But what kind of drivers PA business was/is using and what kind of drivers consumer oriented manufacturers have in their design board. Looking at a typical PA two-way monitor we see 12 to 15 inch woofer what can reach 2Khz at his best and a horn loaded compression driver what can be crossed at 600Hz – 1,5kHz range depending it’s resonant frequency and horn design. Where shall we put the crossover using these drivers? Then the PA loudspeaker manufacturers have those bullet supertweeters with Fs at 5000Hz and 6 to 10 inch midrange drivers. What to build with these –a three way perhaps?
Home speaker designers have woofers covering much wider higher frequency area and the tweeters despite their low resonant freguency (600 – 700Hz, Scan-Speak, Vifa, Seas, Morel) doesn’t like low crossover point (Or do they? That’s another different topic. I have crossed Morels as low as 1,5kHz and I couldn’t detect any hardship on them.) so it’s tempting to make the drivers to cross somewhere between 2 and 3 kHz.
Do PA people concern much about high fidelity? Cabinet compactness and weight are more important if the Spl level is not compromised because of that. Do consumer electronic manufacturers care for high fidelity? Yes and no –depends who is the buyer. For budget and/or SAF minded they sell small two- way mini-monitors (and a subwoofer as requisite supplement), for cost and size no object persons they produce three to five way flagship coffins.
Why I want to avoid 1,5 to 5kHz range for crossover. Well, I’m not a professional PA or consumer audio speaker builder, so I don’t need to care about their cost, size, weight or whatever cutting reasons (to certain degree of course). I build speakers for myself and considering the effort and money I will put in them, I want a maximum performance and rewarding (within my DIY limits of course) out of them. So why not avoid that suspicious frequency area where my ear should be most sensitive and can detect crossover anomalies more easily. Secondly it is much easier to cover that range with one driver then to blend two dissimilar drivers together and deal with the potential crossover ringing.
To keep my rattling honest I haven’t evaluated so much different crossover points/orders to claim that one is inferior to other. I lack in experience because neither time nor other assets have allowed me to try tens and tens of different driver and crossover configurations in adequate time to make any such comparisons. I wish I could do that. Luckily with today’s technology it has became more possible. IIR digital filter separates or Crossover (Aural) Emulator with IIR now being a feature of those crossover simulation/measurement system packages as LspCad and Sound Easy 6.version allow to experiment with those effects with ease.
So let's experiment.
Cheers,
Argo
Hey, who started the two-way alignment approach what has became so popular in today's loudspeaker business? Somebody who as been around in music reproduction hobby/business much longer than I can correct me but I think they were PA people. They, I think, had the same reasons back then, as home speaker manufacturers have today – cut down the cost, weight and size of the product by reducing the number of drivers used. But what kind of drivers PA business was/is using and what kind of drivers consumer oriented manufacturers have in their design board. Looking at a typical PA two-way monitor we see 12 to 15 inch woofer what can reach 2Khz at his best and a horn loaded compression driver what can be crossed at 600Hz – 1,5kHz range depending it’s resonant frequency and horn design. Where shall we put the crossover using these drivers? Then the PA loudspeaker manufacturers have those bullet supertweeters with Fs at 5000Hz and 6 to 10 inch midrange drivers. What to build with these –a three way perhaps?
Home speaker designers have woofers covering much wider higher frequency area and the tweeters despite their low resonant freguency (600 – 700Hz, Scan-Speak, Vifa, Seas, Morel) doesn’t like low crossover point (Or do they? That’s another different topic. I have crossed Morels as low as 1,5kHz and I couldn’t detect any hardship on them.) so it’s tempting to make the drivers to cross somewhere between 2 and 3 kHz.
Do PA people concern much about high fidelity? Cabinet compactness and weight are more important if the Spl level is not compromised because of that. Do consumer electronic manufacturers care for high fidelity? Yes and no –depends who is the buyer. For budget and/or SAF minded they sell small two- way mini-monitors (and a subwoofer as requisite supplement), for cost and size no object persons they produce three to five way flagship coffins.
Why I want to avoid 1,5 to 5kHz range for crossover. Well, I’m not a professional PA or consumer audio speaker builder, so I don’t need to care about their cost, size, weight or whatever cutting reasons (to certain degree of course). I build speakers for myself and considering the effort and money I will put in them, I want a maximum performance and rewarding (within my DIY limits of course) out of them. So why not avoid that suspicious frequency area where my ear should be most sensitive and can detect crossover anomalies more easily. Secondly it is much easier to cover that range with one driver then to blend two dissimilar drivers together and deal with the potential crossover ringing.
To keep my rattling honest I haven’t evaluated so much different crossover points/orders to claim that one is inferior to other. I lack in experience because neither time nor other assets have allowed me to try tens and tens of different driver and crossover configurations in adequate time to make any such comparisons. I wish I could do that. Luckily with today’s technology it has became more possible. IIR digital filter separates or Crossover (Aural) Emulator with IIR now being a feature of those crossover simulation/measurement system packages as LspCad and Sound Easy 6.version allow to experiment with those effects with ease.
So let's experiment.
Cheers,
Argo
Response to capslock questions
capslock,
1. Would you consider making the processed files available, either for playback on a sound card or (beware!) from a CD?
No, it would be too much trouble to work through the legalities of such a distribution. Besides, I would like to see others perform an INDEPENDENT assessment. Music selections should include passages with "delicate harmonic structure", such as well recorded violins, vocal groups that effectively employ harmony (preferably with minimal instrumental accompaniment), and brass instruments recorded so that they sound "like the live experience" (most don't). Very important is that the recording provide LISTENING PLEASURE when played back unaltered. If this is not the case, the subsequent tests are totally meaningless. The recording should be at the high sample rate (96khz is convenient) so that the discrete-time filter (IIR) will more faithfully replicate the analog filter to be used in the final implementation (an assumption). True high resolution recordings are preferred, not Compact Disc (a topic I would rather not address in this thread).
2. Wouldn't FIR filters be a better solution?
I specified IIR digital filters with the assumption that the desired final implementation will be analog. If a digital implementation is considered, then obviously the FIR filter is a consideration. The issues for the FIR filter implementation include: Tradeoffs between pre-ringing and phase distortion (our perception processing at the neural network is in the time domain), and Tradeoffs between filter slope and resulting response in three-dimensional acoustic space (each driver will ring individually, and only when exactly on-axis will the ringing of all drivers cancel). Another consideration is the faithful playback of high resolution formats such as SACD, for which high quality digital signal processing is somewhat problematic.
My investigations included 2nd, 4th, and 6th-order L-R crossovers, on-axis only. No simulation of driver weaknesses, such as cone break-up, was employed (obviously an important consideration).
3. Did you have any specific drivers in mind that would match your needs?
Yes its true, no single driver quite covers the frequency span if high listening levels are desired. Dang! A number of options exist, though all seem to require a first-order crossover somewhere between 200hz and 300hz, and additional attenuation (significant) at lower frequencies to avoid over-excursion of the wideband driver. The options that may suit my requirements include:
a. An electrostatic loudspeaker for the midbass and high frequencies along with a lower-midbass assist from an array of high quality midbass cone drivers for below the first-order crossover (these must have good high frequency extension) and an additional true "woofer" below that using a 4th or 6th order crossover.
b. A wideband driver such as a Jordan JX92s or Fertin for the midbass, along with a tweeter crossed above 6khz (Neo3pdr or ESg2) as a substitute for the electrostatic above. Slightly lower crossover points may be considered.
c. A Manger driver as a substitute for the electrostatic above.
The above designs are complex multi-driver systems that require careful integration, and an electronic crossover would be helpful. The complexity alone should be quite sufficient to convince most DIY builders to simply not worry about phase distortion. Such designs will be costly and take years to design and build. For this reason I encourage others to experiment with phase distortion so that energy is not needlessly wasted for those who are not as affected by phase distortion as I am. Most people do not perceive sound the way I do (a generally true statement for everyone, I believe).
Regarding the wonderful drivers from Scan Speak and Excel, they do have a problem at the upper extreme of their range, and so a compromise in the upper crossover frequency (certainly placing the crossover well below 6khz) would probably result in a better design if these drivers are to be used. These would be good designs, but I am spoiled by my experience with nearly-full-range electrostatics.
capslock,
1. Would you consider making the processed files available, either for playback on a sound card or (beware!) from a CD?
No, it would be too much trouble to work through the legalities of such a distribution. Besides, I would like to see others perform an INDEPENDENT assessment. Music selections should include passages with "delicate harmonic structure", such as well recorded violins, vocal groups that effectively employ harmony (preferably with minimal instrumental accompaniment), and brass instruments recorded so that they sound "like the live experience" (most don't). Very important is that the recording provide LISTENING PLEASURE when played back unaltered. If this is not the case, the subsequent tests are totally meaningless. The recording should be at the high sample rate (96khz is convenient) so that the discrete-time filter (IIR) will more faithfully replicate the analog filter to be used in the final implementation (an assumption). True high resolution recordings are preferred, not Compact Disc (a topic I would rather not address in this thread).
2. Wouldn't FIR filters be a better solution?
I specified IIR digital filters with the assumption that the desired final implementation will be analog. If a digital implementation is considered, then obviously the FIR filter is a consideration. The issues for the FIR filter implementation include: Tradeoffs between pre-ringing and phase distortion (our perception processing at the neural network is in the time domain), and Tradeoffs between filter slope and resulting response in three-dimensional acoustic space (each driver will ring individually, and only when exactly on-axis will the ringing of all drivers cancel). Another consideration is the faithful playback of high resolution formats such as SACD, for which high quality digital signal processing is somewhat problematic.
My investigations included 2nd, 4th, and 6th-order L-R crossovers, on-axis only. No simulation of driver weaknesses, such as cone break-up, was employed (obviously an important consideration).
3. Did you have any specific drivers in mind that would match your needs?
Yes its true, no single driver quite covers the frequency span if high listening levels are desired. Dang! A number of options exist, though all seem to require a first-order crossover somewhere between 200hz and 300hz, and additional attenuation (significant) at lower frequencies to avoid over-excursion of the wideband driver. The options that may suit my requirements include:
a. An electrostatic loudspeaker for the midbass and high frequencies along with a lower-midbass assist from an array of high quality midbass cone drivers for below the first-order crossover (these must have good high frequency extension) and an additional true "woofer" below that using a 4th or 6th order crossover.
b. A wideband driver such as a Jordan JX92s or Fertin for the midbass, along with a tweeter crossed above 6khz (Neo3pdr or ESg2) as a substitute for the electrostatic above. Slightly lower crossover points may be considered.
c. A Manger driver as a substitute for the electrostatic above.
The above designs are complex multi-driver systems that require careful integration, and an electronic crossover would be helpful. The complexity alone should be quite sufficient to convince most DIY builders to simply not worry about phase distortion. Such designs will be costly and take years to design and build. For this reason I encourage others to experiment with phase distortion so that energy is not needlessly wasted for those who are not as affected by phase distortion as I am. Most people do not perceive sound the way I do (a generally true statement for everyone, I believe).
Regarding the wonderful drivers from Scan Speak and Excel, they do have a problem at the upper extreme of their range, and so a compromise in the upper crossover frequency (certainly placing the crossover well below 6khz) would probably result in a better design if these drivers are to be used. These would be good designs, but I am spoiled by my experience with nearly-full-range electrostatics.
2-way crossed below 300 Hz
It was asked about a 2 way with crossover below the 'no fly zone'. Check out the Jordan JX53 plus the 150. It can be crossed at 200 or a bit higher.
http://www.ejjordan.co.uk/drive_frameset2.htm
Brendon
It was asked about a 2 way with crossover below the 'no fly zone'. Check out the Jordan JX53 plus the 150. It can be crossed at 200 or a bit higher.
http://www.ejjordan.co.uk/drive_frameset2.htm
Brendon
Maybe it would be a good idea to find out what makes a crossover audible, and what can be done about it.
The main culprits I know (without claiming to know them all) are:
1.) Transient distortion
2.) Systems that measure flat on-axis can have a very bumpy amplitude response off-axis, resulting in differing frequency responses for the direct- and the reverberant- soundfield.
some remedies I know:
1.) use of transient-perfect crossover networks (there we go again ...).
2.) driver placement (i.e. coaxial arrangements, D'Appolito configuration, closely spaced drivers compared to the crossover-frequency's wavelength etc). Avoid using drivers whose polar responses differ significantly around the crossover frequency. The popular two-way configuration, using a cone midwoofer and a dome tweeter, is particularly bad in this respect (although I have to admit that I like the sound of many of those two-ways).
I assume that, if those remedies are taken into account, it should be possible to move the crossover frequency slightly above 300 Hz for a woofer/FR combination. This would give some leeway for the use of small (and reasonably priced) fullrange drivers that perform well in the upper treble range.
This are just some of my thoughts, critisism is welcomed and encouraged.
Regards
Charles
The main culprits I know (without claiming to know them all) are:
1.) Transient distortion
2.) Systems that measure flat on-axis can have a very bumpy amplitude response off-axis, resulting in differing frequency responses for the direct- and the reverberant- soundfield.
some remedies I know:
1.) use of transient-perfect crossover networks (there we go again ...).
2.) driver placement (i.e. coaxial arrangements, D'Appolito configuration, closely spaced drivers compared to the crossover-frequency's wavelength etc). Avoid using drivers whose polar responses differ significantly around the crossover frequency. The popular two-way configuration, using a cone midwoofer and a dome tweeter, is particularly bad in this respect (although I have to admit that I like the sound of many of those two-ways).
I assume that, if those remedies are taken into account, it should be possible to move the crossover frequency slightly above 300 Hz for a woofer/FR combination. This would give some leeway for the use of small (and reasonably priced) fullrange drivers that perform well in the upper treble range.
This are just some of my thoughts, critisism is welcomed and encouraged.
Regards
Charles
Give him what he wants! Critisism!
And a 5" woofer crossed at 180 Hz
This will make a perfect system, if the baffle is very narrow.
Jordan JX53, that is a Car Speaker
mostly used in Peugeot 205
Would tell you what that is for sort of hifi
that about phase accurate - is very much overrated
as far as I know,
and there is probably not a living soul
that know how phase_accurate sounds.
We are only guessing.
Are , you pleased with this Critisism?
You want more? Want more? /bob marley
Rastaman Talk - Get it!
Now you get what you want,
Do you want more? (want more)
Now you get what you want,
Do you want more? (want more)
You think it's the end,
But it's just the beginning
They stab you in the back
And they claim that you're not looking.
But Jah have them in the region
In the valley of decision.
Now: what you want?
Want more? Want more?
Now: what you get?
Want more? Want more?
Now - Now: do you want more?
With a dometweeter crossed at 1800 Hzphase_accurate said:
And a 5" woofer crossed at 180 Hz
This will make a perfect system, if the baffle is very narrow.
Jordan JX53, that is a Car Speaker
mostly used in Peugeot 205
Would tell you what that is for sort of hifi that about phase accurate - is very much overrated
as far as I know,
and there is probably not a living soul
that know how phase_accurate sounds.
We are only guessing.
Are , you pleased with this Critisism?
You want more? Want more? /bob marley
Rastaman Talk - Get it! Now you get what you want,
Do you want more? (want more)
Now you get what you want,
Do you want more? (want more)
You think it's the end,
But it's just the beginning
They stab you in the back
And they claim that you're not looking.
But Jah have them in the region
In the valley of decision.
Now: what you want?
Want more? Want more?
Now: what you get?
Want more? Want more?
Now - Now: do you want more?
halojoy wrote
Maybe you should just ask those who are mad about real full-range speakers ?!
I think nobody would accept an amplifier (not even a very crappy one) that does to transients what an ordinary multiway speaker does !
Since our hearing uses mainly the inter-aural time-delay to recognize the direction of sound, the time-smear introduced by non phase-accurate speakers results in smeared imaging.
So I honestly ask you if YOU have an answer WHY crossovers introduce audible errors ?????
If there is no apparent reason WHY crossover frequencies between 300Hz and 3 kHz (or more) should be avoided, why bother ??
Regards
Charles
Maybe you should just ask those who are mad about real full-range speakers ?!
I think nobody would accept an amplifier (not even a very crappy one) that does to transients what an ordinary multiway speaker does !
Since our hearing uses mainly the inter-aural time-delay to recognize the direction of sound, the time-smear introduced by non phase-accurate speakers results in smeared imaging.
So I honestly ask you if YOU have an answer WHY crossovers introduce audible errors ?????
If there is no apparent reason WHY crossover frequencies between 300Hz and 3 kHz (or more) should be avoided, why bother ??
Regards
Charles
To our Swiss friend phase_accurate
And the others in this thread
Please do not take any greater notice of
my last 2 posts here
I was out of my mind balance
so those 2 are not fully representative
for the average of a person called halojoy.
I do not want to cause any unnessesery misunderstandings.
Thanks and regards
/halojoy
And the others in this thread
Please do not take any greater notice of
my last 2 posts here
I was out of my mind balance
so those 2 are not fully representative
for the average of a person called halojoy.
I do not want to cause any unnessesery misunderstandings.
Thanks and regards
/halojoy
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.