Phase, Impedance, FR Importance

[paul_n_oneal]

What do you consider the most important piece when modeling a crossover. Do you always try to have the phase right on or do you care more about a flat FR or impedance, etc. Long writeups welcome. Please explain why you choose one over the other. Just trying to increase my knowledge of who, what, where, and why...

[kenc]

> What do you consider the most important
> piece when modeling a crossover. Do you
> always try to have the phase right on or do
> you care more about a flat FR or impedance,
> etc. Long writeups welcome. Please explain
> why you choose one over the other. Just
> trying to increase my knowledge of who,
> what, where, and why...

Hmmm, good question. Without a bunch of tecnical gobbleygook, I think isolating any of these three without giving consideration to the other 2 is a recipe for disaster. You already knew that though.

I'm curious to see what people respond. I've heard a lot of speakers that had great phase or great fr plots, but just didn't sound right for whatever reason. I've also heard some speakers that didn't measure so well that sounded pretty darn good.

I'll be very surprised if someone picks impedance. Who can hear impedance? Assuming it's the amp is up to the task...

[paul_n_oneal]

> Hmmm, good question. Without a bunch of
> tecnical gobbleygook, I think isolating any
> of these three without giving consideration
> to the other 2 is a recipe for disaster. You
> already knew that though.

> I'm curious to see what people respond. I've
> heard a lot of speakers that had great phase
> or great fr plots, but just didn't sound
> right for whatever reason. I've also heard
> some speakers that didn't measure so well
> that sounded pretty darn good.

> I'll be very surprised if someone picks
> impedance. Who can hear impedance? Assuming
> it's the amp is up to the task...

Impedance isn't my highest priority either but I do look at it frequently during modeling. The impedance indicates a number of things that I use. If it's a ported design, the impedance will show the tuning frequency and wheter ported or sealed will show approximately where the xover point is. IIRC, I read a paper awhile back that said something like keeping the impedance phase between +/- 30 degress should be a design consideration. For what reason, I have no idea. I usually go for phase first, FR second, and impedance last. I am curious to see what other do though.

[romanbednarek]

> What do you consider the most important
> piece when modeling a crossover. Do you
> always try to have the phase right on or do
> you care more about a flat FR or impedance,
> etc. Long writeups welcome. Please explain
> why you choose one over the other. Just
> trying to increase my knowledge of who,
> what, where, and why...

Impedance is the lowest priority. I usually just check to make sure that it doesn't dip too low. However if you are using a tube amp it is beneficial to have a flatter impedance response so the speaker doesn't react with the output impedance of the amp and influence the frequency response output.

I like to keep the phase aligned at the listening height. If the phase isn't aligned you aren't getting the maximum added contribution from both drivers. This can be an issue because there will probably be a listening height where the phase is aligned and if you designed the response to be flat at a listening axis where the phase wasn't aligned, it may now be aligned at this height and you will get a peak in the response. Also having the phase aligned across a wider range in the crossover region should improve the imaging.

So that's my take on the subject.

[andre]

> Impedance is the lowest priority. I usually
> just check to make sure that it doesn't dip
> too low. However if you are using a tube amp
> it is beneficial to have a flatter impedance
> response so the speaker doesn't react with
> the output impedance of the amp and
> influence the frequency response output.

> I like to keep the phase aligned at the
> listening height. If the phase isn't aligned
> you aren't getting the maximum added
> contribution from both drivers. This can be
> an issue because there will probably be a
> listening height where the phase is aligned
> and if you designed the response to be flat
> at a listening axis where the phase wasn't
> aligned, it may now be aligned at this
> height and you will get a peak in the
> response. Also having the phase aligned
> across a wider range in the crossover region
> should improve the imaging.

> So that's my take on the subject.
Paul , another look !!! After years of DIY speakers I'm sure the best crossover is

"no crossover" . Some manufacturers in the past (SPICA, EPOS, and few others)have made outstanding speakers without crossover or at most a Cap on the tweeter ! I'm assuming you have to match the right tweeter with the right tweeter !I'm still looking for such a DIY desing.

[paul_n_oneal]

> Paul , another look !!! After years of DIY
> speakers I'm sure the best crossover is

> "no crossover" . Some
> manufacturers in the past (SPICA, EPOS, and
> few others)have made outstanding speakers
> without crossover or at most a Cap on the
> tweeter ! I'm assuming you have to match the
> right tweeter with the right tweeter !I'm
> still looking for such a DIY desing.
JohnK built an MTM with a single cap and inductor using vifa woofers and a seas 27TDFC.

[andre]

> JohnK built an MTM with a single cap and
> inductor using vifa woofers and a seas
> 27TDFC.
Paul , sorry but I'm not able to find the reference you gave me ! Is it on his site?

[curt_c]

IMO:

#1 is easily flat response. Who cares about the rest if the sound is colored? But of course it isn’t just the on axis response, but the power response that is important. Consequently driver capabilities, driver spacing, transfer functions etc. get thrown into the mix. Sometimes you will have to compromise a flat on axis response to get the best power response, but obviously we really want both.

#2 Is phase tracking. Although many of my earlier ‘tuned by ear’ designs sounded good, but obviously had phase all over the place. I generally find that designs that don’t track phase well will not have pinpoint imagining and soundstage, and the instrument locations will drift around depending on the frequency played. Of course this ‘only’ occurs through the transfer bands of the crossover networks. If you don't care about soundstage, -well then your job just got a lot eaisier.

#3 Impedance. Since most of us use solid state amplifiers, and those are voltage sources, impedance variations within reason do not affect the resultant sound. Sure, you can easily ‘optimize yourself into a corner' and get phase and response perfect, but then you find out you are left with a 1 ohm impedance. The trick is finding the combination of topology, crossover frequency, drivers, etc., to get all three al close as possible.

C

[Andy_G]

>"The trick is finding the combination of topology, crossover frequency, drivers, etc., to get all three al close as possible."

then why the heck aren't more of you using series x-os !! ;-))))

[Andy_G]

[Andy_G]

[paul_n_oneal]

So does power response indicate off axis response?

[paul_n_oneal]

> Paul , sorry but I'm not able to find the
> reference you gave me ! Is it on his site?

Here ya go.
<A HREF="http://www.zaphaudio.com/audio-speaker16.html">http://www.zaphaudio.com/audio-speaker16.html</A>

[paul_n_oneal]

> Here ya go.
>
> <A HREF="http://www.zaphaudio.com/audio-speaker16.html">http://www.zaphaudio.com/audio-speaker16.html</A>

Wrong link sorry. Look in his blog.

[unknownuser]

> So does power response indicate off axis
> response?
as it was explained to me, it refers to driver integration - can you tell which driver the sound is coming from, or is the sound just _there_?

I understand little of this stuff, but sometimes remember what I hear.

j.


(Originally posted by: Joshua Layne)

[dlr]

> So does power response indicate off axis
> response?

Yes, but maybe not the way you might think. Power response is essentially the response integrated over a sphere that encapsulates the system. The more important aspect is probably the front 2-pi. The more reverberant a room or the closer reflecting surfaces are, such as floors, walls, etc. the more important it is.

The off-axis is important, some say more so than the direct response, but not in my experience. But it depends on what is affecting the on-axis response.

If you go to my web site, you'll find a very old section on "Windowed averaging". This was one way I worked some years ago. I found that if I used a raw measurement that was a simple average of three axes (-15, 0, +15 degrees), that I can a more acceptable result. This left some non-linearities in the on-axis response, but it "averaged" out in the power response. You'll see reference to this in things such as Stereophile's plots of normalized off-axis plots, most importantly the in the horizontal plane.

This was before I used any diffraction control. What this did was balance the on-axis and off-axis response do some degree. Was it the optimal way? I don't know, it just worked for me.

Now that I use extensive diffraction control I have found that I don't need this. The off-axis is smoothed as well as the on-axis with diffraction control. This lets me concentrate on the on-axis only and know that the off-axis should follow reasonably well.

The caveat here is that there's still the aspect of driver directionality that you can't account for in the on-axis. Choice of drivers and crossover are key to this. Integration is still a bit art, especially for a DIYer, since we can't conduct large scale auditioning for feedback. But if you choose drivers and crossover and cross at or below the point that the lower driver is about 3db or less off-axis at Fc, then you'll be in the ballpark.

You may ask why I bring all of this up. Well, it's because it all influences the power response.

dlr

[dthomas]

Provided Link: http://www.tolvan.com/edge


dlr,

Well said and written. Especially the part about selecting drivers based on their off-axis performance specifically in the passband. I see a lot of designs that are based on too high of crossover points for midwoofers especially in terms of off-axis performance.

But to go a step further, I too have become very fascinated by baffle step. More in understanding how the affect occurs. One thing I noticed in using the standard calculation... we all know the one 4560/Width. And then calculate for an inductor. Well I found some real anomolies using this, especially just slightly off vertical axis. I was getting the step correction but also other bumps in response that are not as expected nor as explained in all the white papers I have read.

A baffle typically a rectangle, the BSC formula assumes that your baffle is round. None of the BSC documents discussed this. As a matter of fact the common equation for BSC corrects for only a small area of the baffle. Then I moved on to studying baffle diffraction and it all came together. While the baffle width gives an inductor value it corrects only in narrow horizontal axis or worse yet, it is assuming that your baffle is round. A rectangular baffle has many different distances from the driver to it's edges. I have been using a baffle diffraction program on my last two designs and using the BSC calculator in it. The values are greatly different from what the standard calculations give but the performance is more acceptable. It attempts to account for width and height of the baffle in calculating the correct values for BSC. The inductor values get increasingly larger for increases in height even when the width remains the same. I have had very good luck with this with the only variable being the resistor value which can be varied depending actual in room response.

It is worth a try it is a freeware program.

Dave

[paul_n_oneal]

Wow, I just read that twice. Hehe, I have some measuring to do my friend :-). I like the visuals. The one with on-axis and the L/R off-axis looks like sine waves going through each other and if you look at it just right, you can average by sight. You can also see the dips and peaks and how each axis is affected. Pretty darn cool.

Take this as constructive critism as it is meant. I would junk the frames and go with divs and layers. Much neater looking and fairly easy to accomplish. Thats just me though.

Again, thanks for that. As I said, I'm going to be a measuring fool this weekend. One question. Do you move the speaker and how do you just best guess the 15 degress? I know you're not supposed to move the mic.

[Jeff B.]

> IMO:

> #1 is easily flat response. Who cares about
> the rest if the sound is colored? But of
> course it isn’t just the on axis response,
> but the power response that is important.
> Consequently driver capabilities, driver
> spacing, transfer functions etc. get thrown
> into the mix. Sometimes you will have to
> compromise a flat on axis response to get
> the best power response, but obviously we
> really want both.

> #2 Is phase tracking. Although many of my
> earlier ‘tuned by ear’ designs sounded good,
> but obviously had phase all over the place.
> I generally find that designs that don’t
> track phase well will not have pinpoint
> imagining and soundstage, and the instrument
> locations will drift around depending on the
> frequency played. Of course this ‘only’
> occurs through the transfer bands of the
> crossover networks. If you don't care about
> soundstage, -well then your job just got a
> lot eaisier.

> #3 Impedance. Since most of us use solid
> state amplifiers, and those are voltage
> sources, impedance variations within reason
> do not affect the resultant sound. Sure, you
> can easily ‘optimize yourself into a corner'
> and get phase and response perfect, but then
> you find out you are left with a 1 ohm
> impedance. The trick is finding the
> combination of topology, crossover
> frequency, drivers, etc., to get all three
> al close as possible.

> C

Regarding your #2 - why don't we start by you telling me what you mean by "phase tracking". I want to make sure we are talking about the same thing before I make a few points along those lines.

[curt_c]

> Regarding your #2 - why don't we start by
> you telling me what you mean by "phase
> tracking". I want to make sure we are
> talking about the same thing before I make a
> few points along those lines.

Hi Jeff! Thanks for endeavoring to clear up any misconceptions I may have caused.

What I mean by phase tracking is that the phase rotation of the individual drivers will be aligned through the transition band of the crossover network, as suggested in the phase plot below.

C