offsetting a tweeter

[jasonc]

I hate to ask a question like this, since I should have the tools to figure it out myself. I've tried and tried to get something meaningful out of BDS, but no luck so far.

Can someone explain, in general terms, the benefits of offsetting a tweeter on a baffle? Specifically, I'm about to build a pair of Lou C's Azurites. (Probably the original version with the TM020J3, but I've not made up my mind for sure yet) I'm wondering what effect that offsetting the tweeter by, say 1", would have on the design. I'd be doing it primarily for asthetic reasons, as I think it looks better that way.

Can it be said that offsetting the tweeter is always a good idea, or does it not always work that way? What about in the case of a three-way: Would you generally offset the tweeter, the midrange, or both? Or is it like a lot of other things in speaker-building, in that you can't make such generalizations? Thanks for any insight!

Jason

[bking]

Off setting the tweeter helps control diffraction effects because the tweeter is not equi-distant from any two cabinet edges. However, if the crossover was designed with the tweeter in a certain position on the front baffle, changing that position could alter the frequency response. My suggestion is to build the system as the designer intended, unless you can measure the change in response of the offset tweeter and account for it in the crossover design.

[curt_c]

What we are really discussing here is altering the path length of the diffracted wavefront from the baffle sides. Offsetting the tweeter (or midrange) will only help the diffraction effects on-axis. If you intend to listen to the speakers toed in so they are aimed at your listening position, offsetting can be beneficial. If you aim the speakers directly into the room, you already have differing path lengths, and centered drivers are fine.

Since an existing design has already taken the diffraction effects into account in the crossover design, I’d be hesitant to change the driver position, -although the difference may be minor. One generally designs for on-axis response, but not always. Centered designs will also have a more symmetrical power response.

C

[dlr]

> What we are really discussing here is
> altering the path length of the diffracted
> wavefront from the baffle sides.

Yes, but it's the detail that counts.

> Offsetting
> the tweeter (or midrange) will only help the
> diffraction effects on-axis.

This is where I disagree. If you study the off-axis response from both the BDS and measurements, within a modest angle of inclusion, an optimally placed off-centered driver will produce better results both on- and off-axis to the side where the tweeter is closer.

> If you intend
> to listen to the speakers toed in so they
> are aimed at your listening position,
> offsetting can be beneficial. If you aim the
> speakers directly into the room, you already
> have differing path lengths, and centered
> drivers are fine.

See comments above.

> Since an existing design has already taken
> the diffraction effects into account in the
> crossover design, I’d be hesitant to change
> the driver position, -although the
> difference may be minor. One generally
> designs for on-axis response, but not
> always.

All good points IMO.

> Centered designs will also have a
> more symmetrical power response.

This may be more important in some situations, especially rooms that are more "live". The reverberant behavior probably has more influence, therefore smoother power response becomes increasingly important.

My experience has been, though, that even for systems closely placed to the listener, an offset is a benefit. The direct wave influences the subtlties of the low level detail and spacial imaging. The more linear, the more faithful the reproduction, since all diffraction is time-delayed distortion.

dlr

[cjd]

> If you study the
> off-axis response from both the BDS and
> measurements, within a modest angle of
> inclusion, an optimally placed off-centered
> driver will produce better results both on-
> and off-axis to the side where the tweeter
> is closer.

How 'bout measurements to back it up, eh? I disagree that the better results will exist on the tweeter side. But I do agree that you can get better results with the tweeter offset off-axis vs. on-axis. Don't have that comparison to show though.

In the pic: top is off-axis to the tweeter side, bottom is the other side. These were taken by rotating the speaker around the tweeter axis (so the tweet was the same distance from the mic at all times). Not only does the response dip a bit less, diffraction is better with the tweeter on the "outside" edge. At least in this study.

I'll be re-doing these measurements one of these days now that I've finally cleaned up all the crap in my measurements system (new sound card and a bit more learning on gating and other such). The general curves are still accurate if you ignore the oddities.

C

[curt_c]

Look at it this way, when listening / measuring with increasing angles from the outer side (tweeter to the inside) at some point the path lengths of the side baffle diffractions will be equal, and therefore as bad as a centered tweeter on axis. If measured from the tweeter side, the differences in path lengths continue to diverge.

Of course this all depends on the position of the tweeter on the baffle, as well, so conditions must be stated (or assumed) before rules of thumb can be applied. For example: IMO, a 0.75” offset on a 9” baffle will work better than a say a 1.5’ or 2” offset.

C

[dlr]

> How 'bout measurements to back it up, eh?

I have done that in the past. IME it has been the case that on average the trend is better off-axis on the near side. As I said, within a modest angle of inclusion, say within the typical angle from on-axis to maybe 10 degrees which will cover most situations, possibly even those pointed straight ahead (which seldom do that I've even known), the measurements generally are better on the tweeter side.

Beyond 15 degrees (and I would even argue 10) its not an issue since this is in the power response region, not the first-arrival region.

We can go all we want to in arguing measurements (of which I've made a few), but the BDS has been shown to be very reliable as a predictor, sometimes extremlely accurate. All you need do is try a number of typical baffles and look at it just a few degrees off.

> I disagree that the better results will
> exist on the tweeter side. But I do agree
> that you can get better results with the
> tweeter offset off-axis vs. on-axis. Don't
> have that comparison to show though.

All I have at hand at the moment are the measurements I made comparing two different boxes, one small with an offset tweeter, the other larger with one centered. The key difference in those is that the one centered has, of course, symmetrical off-axis responses, but the graphs only show the 15 degree measurements, a rather extreme angle for a listening axis. 5 degrees is much closer to reality for typical listening.

<A HREF="http://www.speakerdesign.net/g17d27/d27l_all.gif">http://www.speakerdesign.net/g17d27/d27l_all.gif</A>

> In the pic: top is off-axis to the tweeter
> side, bottom is the other side. These were
> taken by rotating the speaker around the
> tweeter axis (so the tweet was the same
> distance from the mic at all times). Not
> only does the response dip a bit less,
> diffraction is better with the tweeter on
> the "outside" edge. At least in
> this study.

One thing I would like to see is a better start time marker. There's way too much "hash" in that due to some reflections.

Since every case is different, a comparison of a few selected points may seems to indicate one side better than the other, but if you look at an integated average, the near side is better. If you stay within a 5 degree window, what I think of as a typical amount of toe-in change to point the drivers somewhere a bit behind the listeners head, the near side is the better one.

Another issue is that an offset driver needs to be offset in an optimal manner. It is always possible to offset a tweeter and worsen the situation on some particular axis. In this case the near side could, I suppose, be worse. I guess that I would just call this a bad layout, since the BDS can be used to determine the good from the bad offsets.

So with that caveat included, then the near side can be worse for some particular measurement angles. But even in that case, I think that an averaged response on the near side is smoother, just not optimal from what's possible, especially within the typical listening window.

But if one is "shooting in the dark" and not using the BDS or some other software to design the off-axis, then odds are that the near side is better. The "golden ratio" is always a good start.

None of this impacts the original question. I think that Curt's answer covers it by saying that the system was designed without an offset. I agree with him, just follow the design.

dlr

[curt_c]

> This is where I disagree. If you study the
> off-axis response from both the BDS and
> measurements, within a modest angle of
> inclusion, an optimally placed off-centered
> driver will produce better results both on-
> and off-axis to the side where the tweeter
> is closer.

Ah, but at some angle off axis, wouldn’t the baffle edge path lengths for a centered tweeter be the same as an offset tweeter on axis? And for small angles around that axis wouldn’t the diffraction effects be very similar? I've not verified this with BDS, just thinking out loud...

C

[cjd]

> As I said,
> within a modest angle of inclusion, say
> within the typical angle from on-axis to
> maybe 10 degrees which will cover most
> situations, possibly even those pointed
> straight ahead (which seldom do that I've
> even known), the measurements generally are
> better on the tweeter side.

Ahh... I must have missed that part. I was thinking over a broader range, and you're right. But heck - within a ten degree range, even a horizontal MTM can be good.

> One thing I would like to see is a better
> start time marker. There's way too much
> "hash" in that due to some
> reflections.

Yup. These were some of the first measurements I took, ever. I've got things working much better now.

> None of this impacts the original question.

Agreed. It's just an attempt to hijack the thread to further my edification. It worked.

C

[dlr]

> Ah, but at some angle off axis, wouldn’t the
> baffle edge path lengths for a centered
> tweeter be the same as an offset tweeter on
> axis?

For the same baffle dimensions, no. For any given baffle, the difference in path length from any one axis to the other changes if the driver position is moved nearer or farther from the edge in question. The nearer the driver is to the edge, the greater the degree of change of path length for a given angular change.

This has the effect of spreading the diffraction signature in the freqency domain. Thas is, the nearer the edge, the more distributed the effects. It also moves them higher in frequency.

In fact, since the shortest distance to the edge is decreased if the driver is moved nearer, the impact of that point is raised in frequency (as are all points on a given axis.

> And for small angles around that axis
> wouldn’t the diffraction effects be very
> similar?

For a given included angle, moving the driver closer means that the distance to the edge has a greater degree of variation, but it also inludes less of the length of the baffle edge.

The counter to this is that the distances are increasing on the other side (as are those to the top and bottom edges). There's not really a single answer to any question when one must consider four sides. It takes measurements or something such as the BDS to take all aspects into account simulataneously.

Then you have to add in the dispersion characteristics of the driver in question. You end up with too much to contemplate without a tool such as the BDS.

dlr

[dlr]

> Of course this all depends on the position
> of the tweeter on the baffle, as well, so
> conditions must be stated (or assumed)
> before rules of thumb can be applied. For
> example: IMO, a 0.75” offset on a 9” baffle
> will work better than a say a 1.5’ or 2”
> offset.

Exactly. In fact, my first offset baffle was 9" wide with a 0.75" tweeter offset. I think that I used the golden ration to determine that, but my recall is fuzzy (it's been maybe 6-7 years). I still have the baffle, however.

dlr

[johnastockman]

> I hate to ask a question like this, since I
> should have the tools to figure it out
> myself. I've tried and tried to get
> something meaningful out of BDS, but no luck
> so far.

> Can someone explain, in general terms, the
> benefits of offsetting a tweeter on a
> baffle? Specifically, I'm about to build a
> pair of Lou C's Azurites. (Probably the
> original version with the TM020J3, but I've
> not made up my mind for sure yet) I'm
> wondering what effect that offsetting the
> tweeter by, say 1", would have on the
> design. I'd be doing it primarily for
> asthetic reasons, as I think it looks better
> that way.

> Can it be said that offsetting the tweeter
> is always a good idea, or does it not always
> work that way? What about in the case of a
> three-way: Would you generally offset the
> tweeter, the midrange, or both? Or is it
> like a lot of other things in
> speaker-building, in that you can't make
> such generalizations? Thanks for any
> insight!

> Jason

Jason, you're right about generalizations, as there's alot of variables. Go ahead and use the tweeter Lou used...I've built a few of Lou's designs and haven't been disappointed yet. A few have been built for friends, and after listening to them for awhile, I wanted to keep them for myself! Greg S. here on the board was very helpful, giving me some great advice regarding cabinet volume/port dimensions. My Azurites turned out fantastic! I'm keeping these...they sound excellent; great imaging, detail/resolution, smooth mids and highs, and good bass from 2 4" woofers. If you use a different tweeter, you'll have to change the XO. If you need any other info, just let me know. Good luck and let us know what you decide...and pics when they're done! I have some pics of the cabinet construction, etc. at my host site if that would help, also.

John A.

[jasonc]

Thanks for all the informative replies! I'll probably just go ahead and build the speaker "as designed". I *could* take measurements and revise the crossover, but I was looking forward to this one as an easy project

Those speakers look nice. Much the way I hope mine will turn out. Is the tweeter flush mounted? On Lou's site it looks like his is not, but I can't tell for sure.

[curt_c]

> For the same baffle dimensions, no. For any
> given baffle, the difference in path length
> from any one axis to the other changes if
> the driver position is moved nearer or
> farther from the edge in question. The
> nearer the driver is to the edge, the
> greater the degree of change of path length
> for a given angular change.

I agree. My statement ignored the changes in the frequency domain, and was too simplistic. Thanks for clearing up my misinformation.

C

[dlr]

> But heck - within a ten degree range,
> even a horizontal MTM can be good.

And I suppose that it's the power response that makes many like them and others strongly dislike them, which supports the importance of all aspects, both on- and off-axis.

dlr

[johnastockman]

> Thanks for all the informative replies! I'll
> probably just go ahead and build the speaker
> "as designed". I *could* take
> measurements and revise the crossover, but I
> was looking forward to this one as an easy
> project

> Those speakers look nice. Much the way I
> hope mine will turn out. Is the tweeter
> flush mounted? On Lou's site it looks like
> his is not, but I can't tell for sure.

Jason, Lou didn't flush mount the drivers in his design, but I did it anyway, because of the great advice I got from Greg S. The woofers are a little tricky to flush mount because of the very slim lip at the edge of the flange. I got a perfect fit with my router/Jasper jig. I would still flush mount the tweeter, if you have the set-up to do that. Surface mounting is still very acceptable, though, and Lou's comments seem to agree. I really wish I had ordered more of those little woofers! I know alot of people that would love to have me build them a pair, which I would gladly do...it was a fun and fairly easy project to do. Good luck with yours and let us know how it's going.

John A.