Read my analysis in the attached text file. I didn't even consider anything about treble. Earl Geddes was right. The treble was not so important. Not even touches the bigger picture of what makes a great speaker
Treble is very important. From what I've gathered Earl says the top octave isn't so important, meaning the top octave dips associated with wave guides are of least concern when compared to the advantages they bring. But you still need that top octave to be balanced with the rest of the system and the rest of the treble octaves below it are very important.
I'm a fan of Dynaudio tweeters. They can go flat to 40kHz. And I don't even use zobel to compensate the rising impedance like suggested by Dynaudio and implemented by some high end speakers.
There is little instrument and recording [only Chesky in my collection] where the beauty of the highs will be heard. Most of the time [with my lousy implementation] it only produces a lot of HF garbages. There's benefit tho, for blind listening, as it will show any component or part changes in the signal chain.
Once, I build a speaker and I was sure it was the best sound, especially midrange, I have ever heard. I measured it and I found it to roll off at 8 kHz. For the sake of Science, I had to say good bye to the crossover [but I always want to go back].
The next time I need to purchase a tweeter, it would be any of the dual ring radiators...
In order to guess the mystery driver, I downloaded the Faital Pro mp3, and listened all the runner up [FaitalPro, Peerless2, Peerless3, SBA]. I haven't heard the FaitalPro before [didn't joint the first round]...
I only listened to a few part of the opening music of the AC/DC [coz it was easy], but I had an impression that the FaitalPro had very good treble. I liked it so I compared with TC9 [the winner] to see if I would want to buy the drivers but TC9 is better
Are the Dynaudio Ferro fluided ? I was asking myself if it worths to change the FF after 20 years, e.g. if you buy a second hand (not saying if able to find the right thickness of the FF the manufacturer used to choose !).
Without minimum treble, less impact in the mid-low, less transcient everywhere, no ? Maybe it's usefull to let start the treble above > 7 Khz for that, that's why such little 4" wide bandand less are maybe interessant ! While I really ask myself what could be the price to pay (a XO above 200 Hz and before ?? 500 Hz? Maybe it's Worth : first octaves of many instruments like voice !)
On one side we have : better to play a wide band between 300 hz and 6000 hz
On an other side some says : no life if you cut between 300hz & 800 hz where all the energy & life of music often are !
On a last side we have also the old school which says : no way if you filter the human voice too high and above the region where the ear beginns to be able to "see" the directivity (80 hz to 150 hz according the source/or the design)
It seems hard to make a choice with a wide band : be it 3" or 10" ! But at least I'm not sure we are able even after 40 to avoid the 15k hz + range ! (transcients, sometimes low harmonics... maybe too much because the recordings and the drivers : 5 th, 7 th not always usefull especially if we speak about life events reccordings ??? I don't know, because readings of some books of acoustical instrument specialists, ears are able to hear more than 8 octaves in relation to a fundamental (and rare) 40 hz with an acoustical instrument (iirc 7 string bass lower note is 45 hz, 8 strings around 40 hz but not so often seen iirc !). Here we are around 15 K hz... I read some can hear one octave more (30 Khz : hurk!)..... But chamber music, is it really needed to go higher than 15 Khz, at least for a better medium and bass transcient ? I can't find answers in relation to those questions : often speccialists of ears in the medecine dpt are only interrested about the reproduction of the voice to save social relations : so they often talk to a 8 khz High end...
Any ideas about those two questions of trebles and ideal XO with wide bands drivers ??? Only a question of good designs in relations to the drivers ??? (respect of the phas in the nowadays designs, etc) ?
To stay on the level of the floor, I don't have the subjective feeling to suffer of a 2600 Hz XO LR2 with my mid (a 4" aluminium driver) and a LR4 treble with drivers made by Boston Acoustic for the best design they ever made: The lynnfield serie (300L/400L/500L-mine is a improved 400L) (some defaults noticed by Atkins in Stereophile about the 500L but easily fixable as I did - though no phase accurate because a notch passive filter I added at 4.5 Khz).
In which band a wideband has to sing stays a question for me ! (and even more whenclimbing in efficienty and diameter). All simpli is in the design (couple between th XOs and the choice of drivers + loads) ? Like here did Xr with the Harsch filter of the eponym thread!) ?
Without minimum treble, less impact in the mid-low, less transcient everywhere, no ? Maybe it's usefull to let start the treble above > 7 Khz for that, that's why such little 4" wide bandand less are maybe interessant ! While I really ask myself what could be the price to pay (a XO above 200 Hz and before ?? 500 Hz? Maybe it's Worth : first octaves of many instruments like voice !)
On one side we have : better to play a wide band between 300 hz and 6000 hz
On an other side some says : no life if you cut between 300hz & 800 hz where all the energy & life of music often are !
On a last side we have also the old school which says : no way if you filter the human voice too high and above the region where the ear beginns to be able to "see" the directivity (80 hz to 150 hz according the source/or the design)
It seems hard to make a choice with a wide band : be it 3" or 10" ! But at least I'm not sure we are able even after 40 to avoid the 15k hz + range ! (transcients, sometimes low harmonics... maybe too much because the recordings and the drivers : 5 th, 7 th not always usefull especially if we speak about life events reccordings ??? I don't know, because readings of some books of acoustical instrument specialists, ears are able to hear more than 8 octaves in relation to a fundamental (and rare) 40 hz with an acoustical instrument (iirc 7 string bass lower note is 45 hz, 8 strings around 40 hz but not so often seen iirc !). Here we are around 15 K hz... I read some can hear one octave more (30 Khz : hurk!)..... But chamber music, is it really needed to go higher than 15 Khz, at least for a better medium and bass transcient ? I can't find answers in relation to those questions : often speccialists of ears in the medecine dpt are only interrested about the reproduction of the voice to save social relations : so they often talk to a 8 khz High end...
Any ideas about those two questions of trebles and ideal XO with wide bands drivers ??? Only a question of good designs in relations to the drivers ??? (respect of the phas in the nowadays designs, etc) ?
To stay on the level of the floor, I don't have the subjective feeling to suffer of a 2600 Hz XO LR2 with my mid (a 4" aluminium driver) and a LR4 treble with drivers made by Boston Acoustic for the best design they ever made: The lynnfield serie (300L/400L/500L-mine is a improved 400L) (some defaults noticed by Atkins in Stereophile about the 500L but easily fixable as I did - though no phase accurate because a notch passive filter I added at 4.5 Khz).
In which band a wideband has to sing stays a question for me ! (and even more whenclimbing in efficienty and diameter). All simpli is in the design (couple between th XOs and the choice of drivers + loads) ? Like here did Xr with the Harsch filter of the eponym thread!) ?
Last edited:
In order to guess the mystery driver, I downloaded the Faital Pro mp3, and listened all the runner up [FaitalPro, Peerless2, Peerless3, SBA]. I haven't heard the FaitalPro before [didn't joint the first round]...
I only listened to a few part of the opening music of the AC/DC [coz it was easy], but I had an impression that the FaitalPro had very good treble. I liked it so I compared with TC9 [the winner] to see if I would want to buy the drivers but TC9 is better
Agree in sound clips TC9 win over 3FE22r16 but i still in future like to buy a pair of the Faital's because i like to investigate a 16 ohm driver because from my guitar gear i have the experience that if amp can swing the higher voltage there is sometimes some magic freedom in sound when driver has that higher load resistance.
If I can get a full range driver to cover 400Hz to 6kHz, I am happy But the XO from the bass to the full range and from full range to tweeter need to be time accurate otherwise transients lose their impact. I looked at that in detail here:
http://www.diyaudio.com/forums/multi-way/277691-s-harsch-xo.html
and B&O "Hole Filler" XO here:
http://www.diyaudio.com/forums/multi-way/88135-filler-driver-ala-b-o-23.html
For a 2-way FAST, a XO as high as 600Hz can still work, but 300Hz to 400Hz sounds better.
yeap, I was very "involved" yesterday in the reading of Harsch Xo link you provided to me before !
<<<<<For a 2-way FAST, a XO as high as 600Hz can still work, but 300Hz to 400Hz sounds better. <<<<
300 hz to 350 hz was often a classical choice of Xo in relation to the front wall and cabinet design if Iundesrtood the classics ... when we are in the low efficienty world and those small " diameters.
<<<<<For a 2-way FAST, a XO as high as 600Hz can still work, but 300Hz to 400Hz sounds better. <<<<
300 hz to 350 hz was often a classical choice of Xo in relation to the front wall and cabinet design if Iundesrtood the classics ... when we are in the low efficienty world and those small " diameters.
Last edited:
Personally I wouldn't mess with ferro fluid. I imagine every detail about it makes a big difference. I'd be particularly careful what you believe when "someone says" anything. Most of what I hear "audiophiles" say is wrong. Most aren't aware of all the variables at play.
On the issue of the high freq. end, FM radio, which I've always thought sounded plenty good enough, has bandwidth limited to about 15kHZ (due to the subcarrier being just above that). Going beyond that to 20kHZ or more can make a speaker sound more interesting short term (more "air"), but maybe more fatiguing long term.
I think the more important thing about a tweeter is how it decays. I'd rather have a tweeter that only goes to 15kHZ, but has a fast and clean decay (minimal ringing at any frequency), than one that goes to 30kHZ, but has a poor frequency-selective decay response.
If I was going to use a tweeter, I'd probably go for a Fountek 1.5 inch ribbon (Zaph says it's one of the best ribbons), and cross it no lower than 8kHZ (to a TG9 Peerless). It appears to decay significantly better than any dome tweeters, and goes way above 20kHZ without the usual resonant peaks in the 20's that all metal domes seem to have (which is said to cause listener fatigue). Fountek claims it provides a resistive (very non-reactive) load to the power amp too.
The biggest criticism of using a 3 inch driver for midrange and treble is likely the off-axis response. Above about 8kHZ, treble will drop off axis. The good news, and why I don't mind the rolloff, is because it happens gradually, rather than abruptly, like can happen with crossing from a 6 inch mid or woofer driver to a 1 inch tweeter driver at 3kHZ, for example. Any time the FR has an abrupt change, it will draw attention to itself, and color the perception of everything else. The speakers I've built that use the TG9 Peerless 3 incher for 500HZ and above sound stunningly good to my ear.
Eldam,
My exercises for XO points for FAST system being higher than 400Hz had a turn when i tried FIR filters for second and fourth order filters that was amazing. My 10F8424/SPH-250KE minimum phase plot now looks like one full ranger driver with a system bandpass 50-18kHz and XO is upped to 500Hz LR4 (link http://www.diyaudio.com/forums/full...3in-5in-drivers-round-2-a-46.html#post4397112). In reality drivers have IRR LR4 electric XO and are EQed relative flat also outside their pass band to get the electric LR4 XO slopes replicate in acoustic domain, delay to align acoustic center, and a FIR filter that neutralize 360º phase turn cause by IRR XO slopes.
In past it had been said those phase turn is not audiable, well yesterday made two new FIR filters that replicate a IRR LR4 XO phaseturn at 100Hz often used in reality for a sub and a 1000Hz one too. Then ticked them on and off on the fly when listening to these small Edirol full ranger monitors (link http://www.diyaudio.com/forums/multi-way/277691-s-harsch-xo-4.html#post4407466) and think its a night and day difference for instruments attack and realness. Request if any want the IR.wav filters for testing themselves, they will work in JRiver and Foobar audio players DSP engine. Turn them on and a full ranger will have IRR minimum phase XO phase turn added as it was a two way speaker.
Barleywater share some nice data here http://www.diyaudio.com/forums/full-range/276349-fast-tc9fd-dsp-performace-testing-platform.html with very steep XO slopes, low Q system highpass, recommendation for less than quarter wave ctc spacing at XO point, higher XO points and distortion analyzes.
Add some step and square wave response below which think tells a lot about IRR XO and transients. The minimum phase system bandpass examples 20-18kHz could actual replicate as wesayso's line arrays will perform by itself as a IRR minimum phase system (link http://www.diyaudio.com/forums/full...er-full-range-line-array-127.html#post4391729).
My exercises for XO points for FAST system being higher than 400Hz had a turn when i tried FIR filters for second and fourth order filters that was amazing. My 10F8424/SPH-250KE minimum phase plot now looks like one full ranger driver with a system bandpass 50-18kHz and XO is upped to 500Hz LR4 (link http://www.diyaudio.com/forums/full...3in-5in-drivers-round-2-a-46.html#post4397112). In reality drivers have IRR LR4 electric XO and are EQed relative flat also outside their pass band to get the electric LR4 XO slopes replicate in acoustic domain, delay to align acoustic center, and a FIR filter that neutralize 360º phase turn cause by IRR XO slopes.
In past it had been said those phase turn is not audiable, well yesterday made two new FIR filters that replicate a IRR LR4 XO phaseturn at 100Hz often used in reality for a sub and a 1000Hz one too. Then ticked them on and off on the fly when listening to these small Edirol full ranger monitors (link http://www.diyaudio.com/forums/multi-way/277691-s-harsch-xo-4.html#post4407466) and think its a night and day difference for instruments attack and realness. Request if any want the IR.wav filters for testing themselves, they will work in JRiver and Foobar audio players DSP engine. Turn them on and a full ranger will have IRR minimum phase XO phase turn added as it was a two way speaker.
Barleywater share some nice data here http://www.diyaudio.com/forums/full-range/276349-fast-tc9fd-dsp-performace-testing-platform.html with very steep XO slopes, low Q system highpass, recommendation for less than quarter wave ctc spacing at XO point, higher XO points and distortion analyzes.
Add some step and square wave response below which think tells a lot about IRR XO and transients. The minimum phase system bandpass examples 20-18kHz could actual replicate as wesayso's line arrays will perform by itself as a IRR minimum phase system (link http://www.diyaudio.com/forums/full...er-full-range-line-array-127.html#post4391729).
Attachments
Last edited:
X, Need Your Help
Actually I have no doubt which one I like best. As you can see in the TXT, my preference driver won in many aspects, including head to head comparison between the two.
What leave me wondering is that something still doesn't make sense. And I need further listening of one of the file, especially with my speakers (or do proper ABX).
There are several listeners who think that two files sound similar, and I agree with them.
From Round2 I found that the hardest files to differentiate were 10F and TG9. I found the difference only in stronger sibilance with TG9.
So based on that, the two files that sound similar in this round should be TG9 and 10F, but something else doesn't make sense.
My prediction between two distinct files could be 10F -- SB65 or Peerless3 -- 10F. What I need to clarify is actually which two files are the similar one (thus it should be 10F and TG9) but I couldn't rely on perceived difference because 10F and TG9 do sound different. I need to ABX the files in order to hear minute difference. But I couldn't do this because one of the files (B or F I don't remember) started faster than others...
It is strange that no one talked about this in the discussion of software analysis. I don't do such analysis but from listening it was too obvious that one file started too fast. This affects ABX as it would be super easy to differentiate them...
What I need to confirm through ABX is that TG9 and 10F should sound close...
And I need XRK (or anyone with suitable software) to fix the file if it is not difficult of course.
Actually I have no doubt which one I like best. As you can see in the TXT, my preference driver won in many aspects, including head to head comparison between the two.
What leave me wondering is that something still doesn't make sense. And I need further listening of one of the file, especially with my speakers (or do proper ABX).
There are several listeners who think that two files sound similar, and I agree with them.
From Round2 I found that the hardest files to differentiate were 10F and TG9. I found the difference only in stronger sibilance with TG9.
So based on that, the two files that sound similar in this round should be TG9 and 10F, but something else doesn't make sense.
My prediction between two distinct files could be 10F -- SB65 or Peerless3 -- 10F. What I need to clarify is actually which two files are the similar one (thus it should be 10F and TG9) but I couldn't rely on perceived difference because 10F and TG9 do sound different. I need to ABX the files in order to hear minute difference. But I couldn't do this because one of the files (B or F I don't remember) started faster than others...
It is strange that no one talked about this in the discussion of software analysis. I don't do such analysis but from listening it was too obvious that one file started too fast. This affects ABX as it would be super easy to differentiate them...
What I need to confirm through ABX is that TG9 and 10F should sound close...
And I need XRK (or anyone with suitable software) to fix the file if it is not difficult of course.
With clip-2 this is what I did to get alignment:
B-trim 7026 samples
A-trim 4974 samples
D-trim 3468 samples
E-trim 0 samples
F-trim 3298 samples
E-clip-2 had least amount of lead in and is left unchanged.
I did similar editing to clip-1; no evaluation was made with clip-3.
Reference track was sample rate converted to 48kHz; it appears to be inverted relative to recorded clips. A small amount of clock drift is apparent between CD player and H4. Perhaps reference clips should have been recorded from CD out to line in of H4.
B-trim 7026 samples
A-trim 4974 samples
D-trim 3468 samples
E-trim 0 samples
F-trim 3298 samples
E-clip-2 had least amount of lead in and is left unchanged.
I did similar editing to clip-1; no evaluation was made with clip-3.
Reference track was sample rate converted to 48kHz; it appears to be inverted relative to recorded clips. A small amount of clock drift is apparent between CD player and H4. Perhaps reference clips should have been recorded from CD out to line in of H4.
no would then be very happy its loops from soundcard but real electric ones, but as you probably can see i have dig my old PCI AP192 up from drawer which has 4x4 I/O 2 of them being SPDIF and looping SPDIF it will loop from DC to 192kHz sample rate limit (see nice clean cut IR/SR gold traces on REW plots) where by analog I/O its limited by analog capacitor filters. But okay both these two used programs can't use the HF bandwith, think REW is limited to 16bit/48kHz will then probably have to look at ARTA for more resolution at HF.Its to put visuals on transients response and the realness of instrument this gives also you express hearing with better transient XO types and which is actual possible with full rangers example wesayso's line arrays without any FIR filters. As said yesterday i did the test on simple full range monitors with LR4 100Hz and 1000Hz FIR phase turn mirror and was little shocked because one driver with only one voice coil should have no problems as real multi ways with acoustic center errors, ctc spacing problems with vertical lobe and uneven off axis response between LF and HF driver at XO point, but both filters was very audio able.
Real measurements with mic mostly i get results as below which was before 10F with a TC9FD and the more hope than reality XO type BW1 slope. Think need to study those IRR kirkeby filters that is possible responsible Barleywaters stunning smooth plots, but then again he also told us how hard it is to get acoustic perfect square waves in room except at a single spot in space.
Attachments
Thanks Bob, BYRTT for your inputs.
Yes, swaping the FF in a tweeter should be a mess ! I was just asking as I have a Boston tweeter which is 20 years old ( fear to have a thicker viscosity while it sounds good enough yet to my ears) ! (Btw the ringing peak is tamed by an acoustical filter : http://www.diyaudio.com/forums/multi-way/240636-acoustical-filter-tweeters-4.html#post3637839 you can see it's the same for the 4" )
Here you can see the acoustical filter alone, notice the damping I putted behind the metal cone was also an hearable great improvement : http://www.diyaudio.com/forums/multi-way/240636-acoustical-filter-tweeters-6.html#post3933192
Yes, swaping the FF in a tweeter should be a mess ! I was just asking as I have a Boston tweeter which is 20 years old ( fear to have a thicker viscosity while it sounds good enough yet to my ears) ! (Btw the ringing peak is tamed by an acoustical filter : http://www.diyaudio.com/forums/multi-way/240636-acoustical-filter-tweeters-4.html#post3637839 you can see it's the same for the 4"
)Here you can see the acoustical filter alone, notice the damping I putted behind the metal cone was also an hearable great improvement : http://www.diyaudio.com/forums/multi-way/240636-acoustical-filter-tweeters-6.html#post3933192
Last edited:
Oh, thanks. I thought it was B or F that is "misaligned" or different from the rest
Then if it is so, ooppps... actually B also have wrong timing (7026 samples)??
What I need is correct timing for B, D and F so I can examine the minute difference between the three drivers using Foobar ABX. I think Foobar doesn't have the tool to change the timing of the samples?
Last edited:
Same here no doubt which is preferred and think that's the important thing so X and the horserace above get some feedback data that reflect what participants prefer. The guessing think is more a kind of competition and fun but got tougher to decode this round where we listen to previous winners, myself had TG9 and 10F right at second round but they don't for me at all sound as back then, now with new better procedures setup and different tracks.
I checked trim values and spotted apparent typo. Trim value for A should be 4947 samples.
Here is link to clip-2 recordings trimmed for alignment:
https://www.dropbox.com/sh/uq2mjm93rjk7ynl/AACTRtiBmAqNQ_ItI3o7dB-za?dl=0
These are 16bit wave files; I didn't want to convert back to mp3.
Here is link to clip-2 recordings trimmed for alignment:
https://www.dropbox.com/sh/uq2mjm93rjk7ynl/AACTRtiBmAqNQ_ItI3o7dB-za?dl=0
These are 16bit wave files; I didn't want to convert back to mp3.