@wolf_teeth you've crossed 3-way at 475 and 3,000, of course 4" W4 can reproduce trumpets with such a narrow range. And I'm still trying to cross a 6" W6 at 3,500 and I wonder if I need to move down to 2,000 instead then just to follow your steps?
This measurement has too low spl. Noise floor (dark brown) runs around 50-40dB. Notice that REW shows distortion at fundamental F
W6 is of course different driver.
To use REW as generator and RTA analyzer simultaneuosly, just click the symbols and set parameters. Then press red dot at RTAscreen and green triangle on the generator.
Below my laptop on it's own speakers and mics. Notice room noise around 30dB. Camera symbol is to take a screeshot of RTA view.
PS. it is not wise to use harmonics in source signal!
Compare to 2-tone
Last edited:
Technically 400 and 4k for mine.
Being they are different drivers, they will be different in approach. However, using a 6" driver to 3.5k is a stretch for most of them. There are some that can, but most are maybe good to 2.5-3k tops. This is because the driver will beam, and the breakup is usually inherently around 3-3.5k or even lower. Coaxes eliminate the CTC issues, and you effectively have a 5" waveguide with a 1" tweeter. I think 2.5-3k is your target area, depending on what HD says measured. This is likely that the tweeter does not have a very large chamber, and the midbass inductance limits it as you have shown.
@wolf_teeth thanks! Just one important clarification for anyone reading this - the "trumpet" distortion is strictly woofer somewhere above 3 kHz. Tweeter is not part of it - I hear absolutely identical distortion with tweeter on, heavily L-padded or completely disconnected. I never heard the sound getting more distorted with whatever I do with the tweeter, it only can get too bright and fatiguing at the extreme.
I am certain there are other drivers with lower HD than the W4. That said, the bass from my 1139 has less HD than the coaxial woofer in the T18, and the mid range is lower in the T18 than my build as shown above. The treble HD however is on close in both builds, and lower in mine than the T18 as frequency increases. I'm certain the TB mids have resonances in the surround, and it is why there are 2 bumps of 2nd order HD, first mainly at about 850Hz.
This above however is not the point I'm trying to make, nor does it plainly state how good this W4 driver sounds, nor the really phenomenal off axis response it gives.
This below is what I'm concerned over...
The Voice Coil measurements show a 15% HD measurement at 850Hz. Anybody viewing that should be appalled. 15% is a lot compared to a large amount of available modern drivers. In my measurements, that same 850Hz HD peak is registering as -35dB. It's not quite low enough to be 1%/-40dB, but it certainly is not 15% for HD2, and HD3 is another step lower at -50dB.
Odd order is pretty low, and even order is acceptably low, and the tweeter is lower still. The VC review is only one viewpoint for this driver, and my measurements contradict what it says.
Now that I feel I have stated enough to save its documented reputation...
It also has a sound that not many in my experience have been able to achieve. It figuratively acts as a scalpel, and dissects the material it is fed, slicing and dicing. There is no smearing, muddiness, or glazing over in its reproduction capabilities. There are also no fatiguing peaks that enter in the mix. It simply does the job and does it well, and hands off the higher excursion frequencies to someting that can do a better job. Frankly, these make it hard for me to build more because I'm thinking I may not top them in sound again. That said, I enjoy the process too much to not build anything else, and maybe I'll stumble into something even better.
Now, i have heard mixed reviews of the W6. Some saying it didn't measure as specified, some saying it doesn't sound good, and some saying it sounds great, but could be applied a better xover and be even better. It seems to be even more so a driver with varying results than the W4.
This above however is not the point I'm trying to make, nor does it plainly state how good this W4 driver sounds, nor the really phenomenal off axis response it gives.
This below is what I'm concerned over...
The Voice Coil measurements show a 15% HD measurement at 850Hz. Anybody viewing that should be appalled. 15% is a lot compared to a large amount of available modern drivers. In my measurements, that same 850Hz HD peak is registering as -35dB. It's not quite low enough to be 1%/-40dB, but it certainly is not 15% for HD2, and HD3 is another step lower at -50dB.
Odd order is pretty low, and even order is acceptably low, and the tweeter is lower still. The VC review is only one viewpoint for this driver, and my measurements contradict what it says.
Now that I feel I have stated enough to save its documented reputation...
It also has a sound that not many in my experience have been able to achieve. It figuratively acts as a scalpel, and dissects the material it is fed, slicing and dicing. There is no smearing, muddiness, or glazing over in its reproduction capabilities. There are also no fatiguing peaks that enter in the mix. It simply does the job and does it well, and hands off the higher excursion frequencies to someting that can do a better job. Frankly, these make it hard for me to build more because I'm thinking I may not top them in sound again. That said, I enjoy the process too much to not build anything else, and maybe I'll stumble into something even better.
Now, i have heard mixed reviews of the W6. Some saying it didn't measure as specified, some saying it doesn't sound good, and some saying it sounds great, but could be applied a better xover and be even better. It seems to be even more so a driver with varying results than the W4.
With that level of THD and the impedance curve, you should be able to hear something going on at 800Hz. Try a sine wave at 800Hz. Move it around a bit. If you hear yucky stuff, see if it changes if you touch or poke the surround and/or cone with your finger. This means the unit might be faulty. You can do this with the unit out of the box which sometimes makes this 'test' more sensitive
But I don't think this is Andy's problem because it won't be affected by xovers above 1kHz
About audioexpress/Dickason measurement of distortion, it was done at 10cm with rather high voltage, which explains the high % and 115dB fundamental. Sort of stress test.
"This produced the distortion curves shown in Figure 19 for the woofer (6.2V), and the distortion curves shown in Figure 20 for the tweeter dome (1.43V)."
"This produced the distortion curves shown in Figure 19 for the woofer (6.2V), and the distortion curves shown in Figure 20 for the tweeter dome (1.43V)."
Last edited:
The R3 coaxial can't make 200hz at all, it's a midrange, without suspension that allows such movement. It's not the same driver.
While I am still conducting my experiments about cone breakup in W6 please read my posts and then the whole thread about cone breakup and misconception about how many harmonics are important:
In trumpets, harmonics up to the 10th one are making the sound, where the fundamental frequency is actually usually missing - in trumpets ALL the sound is made from harmonics:
From this picture, the most important are H7 and H8, while H10 is probably not very audible. So if you have 10% harmonic distortion in H7, there is a major, MAJOR audible distortion and your speaker will sound like a trashcan. I verified it myself with > 100 hours of listening tests.
Most musical instruments need at...
The teaser - piano has up to 30 audible harmonics and the distortion of all of them sums up together. Trumpets have 10.
In trumpets, harmonics up to the 10th one are making the sound, where the fundamental frequency is actually usually missing - in trumpets ALL the sound is made from harmonics:
From this picture, the most important are H7 and H8, while H10 is probably not very audible. So if you have 10% harmonic distortion in H7, there is a major, MAJOR audible distortion and your speaker will sound like a trashcan. I verified it myself with > 100 hours of listening tests.
Most musical instruments need at...
Harmonics of an instrument are in the source signal (if the mic and recorder are working and producer has not done any tricks).
The reproduction chain (including the speaker) should not add anything. Two different things.
The reproduction chain (including the speaker) should not add anything. Two different things.
@Juhazi I don't think we are disagreeing about anything, it's just I'm trying to explain complex matters with awkward words. Please read my post about piano harmonics:
So piano harmonics are not the exact multiplier of the base frequency. Look at the 440 Hz note:
The H8 and H9 are not at exact multipliers of 3080 and 3520, they are more like 3110 and 3550. If the cone playing 440 Hz has any H8 and H9 distortion at all it will be very audible because it will play in fact at 3080, 3110, 3520 and 3550 at once, with double peaks.
Now look at H9 I just measured with naked W6 woofer, no XO:
Look at the distortion peak at 490 Hz. The H9 of this is 4410, look where it is at the FR - exactly at the cone breakup downslope after the 4200 bump. So the piano will try to play its shited H9 at 4460 and the woofer will add its own distortion at 4410, the woofer will create a double 4410 & 4460 because we are trying to make it to play in the area it cannot.
And it is very audible because I know, I've put several hours of piano on the naked woofer and it sounds really bad.
The solution is to low pass this 4200 bump completely as low as you can, and make the tweeter to play that. My plan it to try some variations of a cross at 3 khz, suppressing this little bump at 3000 by 3 Db and killing the 4200 by as much as can. I already tried this with 4th order XO and it sounded quite well, but I don't want to settle on 4th order, it's too much.
To understand what harmonic distortion can do to the piano read this thread about piano producing harmonics not exactly at the exact harmonic frequencies: https://music.stackexchange.com/que...harmonics-not-being-integral-multiples-of-the
As you see Fourier of the 440 Hz note up to H5 at 2640 Hz it goes exact, but the H6 is not at 3080 but at something like 3110, and so one.
So if your speaker is producing H8 distortion at 3080 exactly, it will be double note, it will play both 3080 and 3110, and it will sound horribly...
As you see Fourier of the 440 Hz note up to H5 at 2640 Hz it goes exact, but the H6 is not at 3080 but at something like 3110, and so one.
So if your speaker is producing H8 distortion at 3080 exactly, it will be double note, it will play both 3080 and 3110, and it will sound horribly...
So piano harmonics are not the exact multiplier of the base frequency. Look at the 440 Hz note:
The H8 and H9 are not at exact multipliers of 3080 and 3520, they are more like 3110 and 3550. If the cone playing 440 Hz has any H8 and H9 distortion at all it will be very audible because it will play in fact at 3080, 3110, 3520 and 3550 at once, with double peaks.
Now look at H9 I just measured with naked W6 woofer, no XO:
Look at the distortion peak at 490 Hz. The H9 of this is 4410, look where it is at the FR - exactly at the cone breakup downslope after the 4200 bump. So the piano will try to play its shited H9 at 4460 and the woofer will add its own distortion at 4410, the woofer will create a double 4410 & 4460 because we are trying to make it to play in the area it cannot.
And it is very audible because I know, I've put several hours of piano on the naked woofer and it sounds really bad.
The solution is to low pass this 4200 bump completely as low as you can, and make the tweeter to play that. My plan it to try some variations of a cross at 3 khz, suppressing this little bump at 3000 by 3 Db and killing the 4200 by as much as can. I already tried this with 4th order XO and it sounded quite well, but I don't want to settle on 4th order, it's too much.
Look where H9 distortion is very low - at 400 Hz. 400 * 9 = 3600, at the downslope of the 3000 peak. Nice, so the 3000 by itself is not so bad, we can keep the rolloff from 3000 to 3600 as part of the low pass.
The distortion at 320 Hz is probably unavoidable because H9 is 2880 and the tweeter resonance is 1 kHz, and I want to have enough space to cut the tweeter down or I damage it.
Also what matters is how many instruments are distorted at once. If one is playing at 320 and another at 480, at least I'll have only one of them distorted and another is not. But if it still sounds bad I will have to cross at 2880 somehow.
The distortion at 320 Hz is probably unavoidable because H9 is 2880 and the tweeter resonance is 1 kHz, and I want to have enough space to cut the tweeter down or I damage it.
Also what matters is how many instruments are distorted at once. If one is playing at 320 and another at 480, at least I'll have only one of them distorted and another is not. But if it still sounds bad I will have to cross at 2880 somehow.
androxylo, what is the crossover scheme that you have now with W6 coaxial? I have the feeling that you have shown only measurements of the woofer without any eq or xo.
@Juhazi right now I'm playing something temporary, even though it sounds great, just not great enough. The next one I'll try with proper parts:
It will be different when I actually measure it. And I understand that I have to add an inductor in the serial notch filter parallel to the driver to tame the electro-mechanical driver resonance at 3 kHz, the resonance that was giving me distortion grief for 2 months already.
Here the crown jewel will be the 14 gauge 2 mH inductor that is supposed to add clarity as it will be the only part at the signal path (right now I have several). Yes a lot of parts but not at the signal path.
And I have yet to invent something for the 3.7 kHz dip, unfortunately both drivers have a coordinated failure there, probably diffraction at the tweeter boundary. My thinking is that the large inductor effect might be nonlinear and the actual curve will differ from what Xsim thinks. Also off axis response should be better, it cures diffractions.
But I am anticipating this to sound great even with the dip.
It will be different when I actually measure it. And I understand that I have to add an inductor in the serial notch filter parallel to the driver to tame the electro-mechanical driver resonance at 3 kHz, the resonance that was giving me distortion grief for 2 months already.
Here the crown jewel will be the 14 gauge 2 mH inductor that is supposed to add clarity as it will be the only part at the signal path (right now I have several). Yes a lot of parts but not at the signal path.
And I have yet to invent something for the 3.7 kHz dip, unfortunately both drivers have a coordinated failure there, probably diffraction at the tweeter boundary. My thinking is that the large inductor effect might be nonlinear and the actual curve will differ from what Xsim thinks. Also off axis response should be better, it cures diffractions.
But I am anticipating this to sound great even with the dip.
Last edited:
I was finally able to achieve some very spectacular sound quality with my TB W6 project, and I want to share some instructions for noobs, written by a noob. Someone planning a project with W6 2313 may find it useful, more experienced builders please correct or improve my notes.
This drive is quite hard for a beginner and it took my 2 months of experimenting and 500 hours of listening to come with something decent.
This drive requires a proper passive crossover. I was playing with active equalization and found it a dead end, for the reasons that will be clear later.
Here is the frequency response of the naked woofer measured at the computer desktop, with a monitor, table and wall reflections - the speaker is less than 1 ft from the backwall. The tweeter response will be added separately.
Here I want to stress that this is a desktop project, speakers are 2 ft from my ears, and a 38" monitor in between, and yet it worked well. My room is here: https://www.diyaudio.com/community/threads/tb-new-line-of-coax-drivers.324159/post-7820521
The most important information about W6 is that the woofer starts to break up at 2.5 kHz and the resonance peaks at 3 and 4.5 kHz are producing a major audible distortion. The only way to achieve this driver capabilities is to build the proper passive crossover crossing somewhere in the 2 - 2.5 kHz area, and additionally suppress the resonance with a parallel notch filter.
I would attribute a limited popularity of this driver to the fact that the Tang Band itself posted an absolutely sloppy crossover at its website. I've tried 20 different crossovers including many posted in this thread over years and none of them sounded to my satisfaction. Even though myself I came with a 2 - pound crossover with huge parts count it beats anything posted so far by sound quality. However, please feel free to come with a simpler version.
To tame the resonances, on the woofer sound path are huge 14 gauge inductors that preserved the punchy and very fast bass. The parallel notch filter is very important part of taming the cone breakup, and please read this thread explaining that the serial notch filter will not work. DSP will not work either, only large inductance with thick wire placed in series with the driver.
This is how this (might not be exact version as I've made like 10 variants of it) crossover is suppressing the 3 kHz and 4.5 kHz peaks. Perhaps more could be done but the difference is huge, completely different level of clarity and distortion level. Here is important to remember that the tweeter has its resonance frequency at 1 kHz and must not be allowed to play that low. Remember that to work with this driver the woofer comes first, the way it sounds not distorted, and the tweeter as added to fill the gaps. The curve that I finally soldered and is good to keep:
The tweeter diffractions are mostly from the baffle. There was one bad tweeter peak probably related to the rubber ring that makes it reflective - I've added a notch filter for it. I plan to recondition this particular speaker by smoothing the baffle and add some felt pads - but it's beyond the crossover to fix.
I also did not make a separate baffle step compensation to avoid increasing the part count even more. I am kind of ok with this 400-900 dip, the bass instruments in this range are still playing very well, not worth the trouble.
And I want to thank @perceval for suggesting this driver and guiding me through the speaker building process, I never did it before. I think a desktop speaker of this quality is just impossible to buy. If you can suggest any commercial model competing with this desktop please tell me.
This drive is quite hard for a beginner and it took my 2 months of experimenting and 500 hours of listening to come with something decent.
This drive requires a proper passive crossover. I was playing with active equalization and found it a dead end, for the reasons that will be clear later.
Here is the frequency response of the naked woofer measured at the computer desktop, with a monitor, table and wall reflections - the speaker is less than 1 ft from the backwall. The tweeter response will be added separately.
Here I want to stress that this is a desktop project, speakers are 2 ft from my ears, and a 38" monitor in between, and yet it worked well. My room is here: https://www.diyaudio.com/community/threads/tb-new-line-of-coax-drivers.324159/post-7820521
The most important information about W6 is that the woofer starts to break up at 2.5 kHz and the resonance peaks at 3 and 4.5 kHz are producing a major audible distortion. The only way to achieve this driver capabilities is to build the proper passive crossover crossing somewhere in the 2 - 2.5 kHz area, and additionally suppress the resonance with a parallel notch filter.
I would attribute a limited popularity of this driver to the fact that the Tang Band itself posted an absolutely sloppy crossover at its website. I've tried 20 different crossovers including many posted in this thread over years and none of them sounded to my satisfaction. Even though myself I came with a 2 - pound crossover with huge parts count it beats anything posted so far by sound quality. However, please feel free to come with a simpler version.
To tame the resonances, on the woofer sound path are huge 14 gauge inductors that preserved the punchy and very fast bass. The parallel notch filter is very important part of taming the cone breakup, and please read this thread explaining that the serial notch filter will not work. DSP will not work either, only large inductance with thick wire placed in series with the driver.
This is how this (might not be exact version as I've made like 10 variants of it) crossover is suppressing the 3 kHz and 4.5 kHz peaks. Perhaps more could be done but the difference is huge, completely different level of clarity and distortion level. Here is important to remember that the tweeter has its resonance frequency at 1 kHz and must not be allowed to play that low. Remember that to work with this driver the woofer comes first, the way it sounds not distorted, and the tweeter as added to fill the gaps. The curve that I finally soldered and is good to keep:
The tweeter diffractions are mostly from the baffle. There was one bad tweeter peak probably related to the rubber ring that makes it reflective - I've added a notch filter for it. I plan to recondition this particular speaker by smoothing the baffle and add some felt pads - but it's beyond the crossover to fix.
I also did not make a separate baffle step compensation to avoid increasing the part count even more. I am kind of ok with this 400-900 dip, the bass instruments in this range are still playing very well, not worth the trouble.
And I want to thank @perceval for suggesting this driver and guiding me through the speaker building process, I never did it before. I think a desktop speaker of this quality is just impossible to buy. If you can suggest any commercial model competing with this desktop please tell me.
Attachments
Even though the cone breakup is not a harmonic distortion its location and relative value can be measured with H3 harmonic. Mind you, the distortion itself is not measured with it, for me it was mostly about doing 500 hours of listening tests to be sure, but when you know where the distortion is you can have clues:
The H3 bump at 1.5 kHz means the bad resonance is at 1.5*3 = 4.5 kHz, indeed a bump at my chart above. But you can also see that there is absolutely no bump at 1 kHz, which would be corresponding to the original peak, but the parallel notch filter completely fixed it. One more notch filter is possible but as I've said too many parts already.
The H3 bump at 1.5 kHz means the bad resonance is at 1.5*3 = 4.5 kHz, indeed a bump at my chart above. But you can also see that there is absolutely no bump at 1 kHz, which would be corresponding to the original peak, but the parallel notch filter completely fixed it. One more notch filter is possible but as I've said too many parts already.