As I said, i only got as far as the theory! I did a search for midrange drivers that might qualify, and came up with the grand total of zero.
On top of that, the eventual realisation of the impossibility of getting good enough (read "exact") phase matching even on axis, never mind off-axis, for linear phase crossovers was what put me off the whole crazy quest, and eventually landed me on three-way Linkwitz-Riley.
Alex
Joachim Gerhard (who I suppose qualifies as a DIY man here...) has a lovely sloped baffle in his SEAS based commercial Anima speaker. He said he used third order butterworth:
An externally hosted image should be here but it was not working when we last tested it.
I used a sloped baffle in my own three-ways to get time alignment between mid and treble, and my recent measurements seem to confirm that this worked quite well. The drawback in this case, though, was that the listening location is then very much below the tweeter axis, resulting in some quite serious HF loss above 5kHz.
Alex
You are at the difficult end of the Duelund three way spectrum here Alex. Ideal slopes get nearer LR2 and the mid is struggling. Well, that's the theory.
An externally hosted image should be here but it was not working when we last tested it.
I quote from Steen Duelund's paper below.
Attachments
Time aligned ! .. Mr Peabody: set the Wayback Machine to 1979!
Haven't seen 'this' theory for some time.
Clearly if one waits long enough that old frock in the closet may come back into fashion.
Still not sure re Mr Graveson's designs.
I meticulously built one of his crossover creations.. once.
It proved to be inferior sounding to the stock/oem crossover, in my drivers and my ears... at least.
Others also concurred, so not entirely my perceptions alone either.
Definitely took the fan out of fanboy for me.
Haven't seen 'this' theory for some time.
Clearly if one waits long enough that old frock in the closet may come back into fashion.
Still not sure re Mr Graveson's designs.
I meticulously built one of his crossover creations.. once.
It proved to be inferior sounding to the stock/oem crossover, in my drivers and my ears... at least.
Others also concurred, so not entirely my perceptions alone either.
Definitely took the fan out of fanboy for me.
You are at the difficult end of the Duelund three way spectrum here Alex. Ideal slopes get nearer LR2 and the mid is struggling. Well, that's the theory.
I quote from Steen Duelund's paper below.
Yes, but maybe I wan't clear in my post - in this case I use a proper three-way LR4 crossover. I don't think of the midrange as a filler driver, since the crossover points are three octaves apart.
Alex
So why don't you dispose them along an arch ??
Because I hadn't appreciated the drop off off axis at the time, and I can't bring myself to build new boxes…
Alex
I think the whole issue of time alignment is made up by putting a cross over in a frequency range where it should not be !
The engineer created a 'problem' to be able to create a 'solution'
"The phase coherence of harmonics in the vocal formant range, ~630Hz to 4000Hz"
http://www.diyaudio.com/forums/multi-way/223714-phase-coherence-harmonics-vocal-formant-range-630hz-4000hz.html
.
The engineer created a 'problem' to be able to create a 'solution'
"The phase coherence of harmonics in the vocal formant range, ~630Hz to 4000Hz"
http://www.diyaudio.com/forums/multi-way/223714-phase-coherence-harmonics-vocal-formant-range-630hz-4000hz.html
.
Bare, why didn't you send a builders comments to Troels if you thought his filter was worse than your original?
Alex, I looked into the theory of the B&O filler driver, and it's not what Steen Duelund worked toward. The filler driver was an attempt to get linear phase (or good impulse response) by wiring the LR2 bass and tweeter out of phase and filling the null with the shallow mid filter which was actually 90 degrees out of phase, or "in quadrature" as Lynn Olson remarked. It was a second order all pass overall.
Steen Duelund was trying to get all 3 drivers working with the same phase, so it all integrates nicely and has an allpass or flat frequency response. I've been struggling to understand his filters, but they are 4th order at heart. And he doesn't attempt to get linear phase, but rather phase alignment.
So firstly some theory:
I'll apply it in the next post, so bear with me again!
Alex, I looked into the theory of the B&O filler driver, and it's not what Steen Duelund worked toward. The filler driver was an attempt to get linear phase (or good impulse response) by wiring the LR2 bass and tweeter out of phase and filling the null with the shallow mid filter which was actually 90 degrees out of phase, or "in quadrature" as Lynn Olson remarked. It was a second order all pass overall.
Steen Duelund was trying to get all 3 drivers working with the same phase, so it all integrates nicely and has an allpass or flat frequency response. I've been struggling to understand his filters, but they are 4th order at heart. And he doesn't attempt to get linear phase, but rather phase alignment.
So firstly some theory:
I'll apply it in the next post, so bear with me again!
Attachments
It's better to walk before you run, so what I did here is try to design a Steen Duelund two way. 6" paper bass plus 1" soft dome tweeter. It looks like this with 4th order slopes:
And we're time aligning it with a 3cm. recess on the tweeter:
Now Steen always said you need to correct impedance, so I've done this with correct Zobels. I've also tried to take bafflestep out of the equation with a 6 ohm resistance approach, if you follow.
I pretty much let the Boxsim optimiser take it from there, and what it came up with was interesting. It was fourth order on bass and tweeter and it liked the Zobels. And it didn't like notches at all with time alignment!
So we seem to have a new design to try. It looks very good to me on phase, albeit the bright midrange frequency response needs work, but we can refine that later.
An externally hosted image should be here but it was not working when we last tested it.
And we're time aligning it with a 3cm. recess on the tweeter:
An externally hosted image should be here but it was not working when we last tested it.
Now Steen always said you need to correct impedance, so I've done this with correct Zobels. I've also tried to take bafflestep out of the equation with a 6 ohm resistance approach, if you follow.
I pretty much let the Boxsim optimiser take it from there, and what it came up with was interesting. It was fourth order on bass and tweeter and it liked the Zobels. And it didn't like notches at all with time alignment!
So we seem to have a new design to try. It looks very good to me on phase, albeit the bright midrange frequency response needs work, but we can refine that later.
Attachments
I've tidied up the design a bit here. I couldn't help noticing that the second section of the 3cm. time aligned tweeter filter wasn't doing much.
Here's the drive units:
W 170 S - 8 Ohm
G 20 SC - 8 Ohm
The 4th order electrical filter is very nice indeed for the bass. But clearly the Fs 1.9kHz tweeter has some natural rolloff which we can exploit for a simpler filter. It still looks good on phase and approximates the LR4 curve that is our guide here:
Steen does say he doesn't like 6" plus 1" tweeter. And the Fs resonance is higher than we would like ideally for low distortion. So I'm thinking that this 29mm SEAS jobbie would be closer to his ideas here:
H1318-06 29TFF/W
Being flatter lower, it gives a better overlap that would probably need a 4th order electrical filter and be nearer some sort of Duelund "perfect phase alignment at all frequencies".
An externally hosted image should be here but it was not working when we last tested it.
Here's the drive units:
W 170 S - 8 Ohm
G 20 SC - 8 Ohm
The 4th order electrical filter is very nice indeed for the bass. But clearly the Fs 1.9kHz tweeter has some natural rolloff which we can exploit for a simpler filter. It still looks good on phase and approximates the LR4 curve that is our guide here:
An externally hosted image should be here but it was not working when we last tested it.
Steen does say he doesn't like 6" plus 1" tweeter. And the Fs resonance is higher than we would like ideally for low distortion. So I'm thinking that this 29mm SEAS jobbie would be closer to his ideas here:
H1318-06 29TFF/W
Being flatter lower, it gives a better overlap that would probably need a 4th order electrical filter and be nearer some sort of Duelund "perfect phase alignment at all frequencies".
Attachments
Last edited:
Like Troels' other classic three way design, this one is very much worth building. You're getting a nice time aligned/phase aligned 2nd order design using very well designed drivers. Low distortion, a 3kHz xover used with a relatively small diameter mid and with an 8" driver on the bass.
What this design does is what you want a three way to do. Exploit the advantages of a small diameter mid range whilst using a bass driver larger than your typical 2 way.
What this design does is what you want a three way to do. Exploit the advantages of a small diameter mid range whilst using a bass driver larger than your typical 2 way.
I'm planning to order this kit also.
I'll order it with the standard cap and the Superior cap. I what to do a blind test and see for my self what is the big talk about "Super Caps".
By the why, it seems that the kit from "Jantzen Audio" is better then any other supplier. (from a cash point of view).
I'll order it with the standard cap and the Superior cap. I what to do a blind test and see for my self what is the big talk about "Super Caps".
By the why, it seems that the kit from "Jantzen Audio" is better then any other supplier. (from a cash point of view
).
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.