is driver spacing important ?

[busterno1]

I am in the process of rebuilding a set of speaker kit boxes in to new boxes of slightly different shape same volume .

because it is a 2 1/5 I have been told I have to keep the drivers at the same spacing in the front baffle . Is this right .

if this is right. What effect will moving the drivers closer together have . If any Thanks

[sreten]

The full range bass/mid should be near the tweeter,
both near seated ear height, spacing these drivers
increases lobing at the crossover frequency.

The 0.5 way bass unit can be near the other unit,
or placed at any hieght, position influences the
smoothness of room gain and on axis floor dip.

Halfway down the cabinet is one compromise.

sreten.

[Hefty]

I have a vague memory of reading something about this...
It had something to do with not placing the drivers further apart than the wavelength of the crossover frequency.

[kelticwizard]

Keep the drivers the same space apart when the crossover will occur at a large fraction of the wavelength of the crossover frequency.

If you had a 3 way speaker and wanted to cross over the bass to the mid at 200 Hz, the wavelength is 67 inches so you could put the bass and mid anywhere on the baffle and it will still be a small fraction of the crossover wavelength. You wouldn't have to worry there.

If you are going to cross over the mid to the tweeter at 2,000 Hz, the crossover frequency wavelength is 6.75 inches. That is the center of one driver to the center of the other. A couple of inches one way or the other can have a large effect there.

There is the phenomenon of reinforcement of sound when the crossover frequency wavelength equals the space between the drivers on the baffle, (center-to-center), and also the issue of negative reinforcement when they are half a wavelength apart on the baffle. So if the crossover is designed for two speakers that are not exactly one wavelength apart at the crossover frequency, and you space them exactly one wavelength apart using the same crossover, it is likely that you will have a "hump" at the crossover frequency. If you put them 1/2 wavelength apart and the crossover was not designed for that spacing, there will be a "dip" in the response at the crossover frequency. There is also the matter of lobing when the drivers are more than a wavelength apart.

The driver center-to-center spacing, especially going mid-to-tweeter, is one of the reasons that ready-made crossovers and simple crossover formulas are not such a great idea. They will probably miss if they don't take driver spacing into account.

One of these days, I am going to conduct some tests to figure this thing out once and for all. I know from sad experience the phenomenon exists, but I never approached it systematically. If somebody knows about some simulation freeware, (I understand LspCAD takes this into account, but it is not freeware), let us know.

A good speaker designer will have taken this account already, so by all means keep the driver spacing identical in those ranges where the space between the center of the drivers is some substantial fraction of the crossover wavelength.

PS: To determine wavelength of any frequency, divide the the wavelength of 1 Hz, (13,500 inches) by the frequency in question. What is the frequency of 500 Hz? 13,500¸500=27 inches. How about 1700 Hz? 13,500¸1700=7.94 inches. Of course, you don't thave to take this to the hundredth of an inch.

[Bricolo]

and how do you design a crossover for a given spacing?

[kelticwizard]

Bricolo:

I wish I had the answer for that. I don't. I don't mind speculating, though.

I think it is all tied in with impedance phase of the driver in addition to the phase effects introduced by the crossover components as well.

I have read that the phenomenon of having the drivers one wavelength apart can be as much as +6 dB difference. However, certain other things I have come across leads me to believe that it is more like +3 dB difference. And the half wavelength spacing works in reverse-3 dB difference at the half wavelength.

I have read that LspCAD actually models this right on the money. Adire Audio has a free version of LspCAD for the AV8, Shiva, etc. However, it does not have tweeters on this version. There seems to be a possiblilty of importing tweeters to the program to find out, but I have not investigated this.

I would specualte that maybe the safe way to do this is to space your mid and tweeter 3/4 wavelength apart, since one wavelength gives positive reinforcement and half wavelength gives negative reinforcement. But that is just a complete guess on my part.

If you look at commercially made speaker systems, a large percentage of them have the tweeter and mid spaced exactly at the crossover frequency, even when they can move them somewhat closer.

I suspect that, on a second order design which is supposed to have the polarity of the tweeter reversed, this might give the designers the chance to actually maintain the same polarity for the mid and tweeter. This might be accomplished by crossing over the woofer and tweeter over at different frequencies than the target crossover frequencies, and taking advantage of the resulting difference in phasing at the crossover frequency and the "fill in" effects of having the drivers one wavelength apart.

I think the best thing for me to do is finally get Speaker Workshop set up, put together some mids and tweeters and start measuring. To be honest, this has been bugging me for some time.

[sreten]

Think KW's going off on the the wrong tack here.

The only interest on axis is the relevant phase offset of the two
drivers. On the vertical axis this offset increases due to the
spacing of the drivers reaching a maximum of the driver spacing
at 90 degrees.

If the spacing is one wavelength you get a single null above and
below when this difference is 180 degrees phase, basically 3
lobes of response at the crossover frequency, the nulls are
approximately 45 degrees above and below axis, but basically
the closer the spacing the wider this null angle is, or the wider
the spacing the narrower this null angle is.

If the drivers do not have planar acoustic centres the nulls can
be assymetrical, but nonsymmetrical crossovers can compensate
for this phase difference.

This also explained the preference for vertical alignment, 3 lobes
horizontally is not a good idea.

If the driver spacing is two wavelengths you get two nulls above
and below the axis, i.e. 5 lobes, the number of lobes increasing
with increasing driver spacing.

This lobing can be a problem with 2 way MTM designs,
compared to 2 way MT designs.

This is only at the crossover frequency, driver spacing effectively
increases as you go up in frequency, causing more problems with
wide overlap crossovers, or multiple driver arrays.

sreten.

[Bricolo]

Originally posted by sreten
Think KW's going off on the the wrong tack here.

The only interest on axis is the relevant phase offset of the two
drivers. On the vertical axis this offset increases due to the
spacing of the drivers reaching a maximum of the driver spacing
at 90 degrees.

If the spacing is one wavelength you get a single null above and
below when this difference is 180 degrees phase, basically 3
lobes of response at the crossover frequency, the nulls are
approximately 45 degrees above and below axis, but basically
the closer the spacing the wider this null angle is, or the wider
the spacing the narrower this null angle is.

If the drivers do not have planar acoustic centres the nulls can
be assymetrical, but nonsymmetrical crossovers can compensate
for this phase difference.

This also explained the preference for vertical alignment, 3 lobes
horizontally is not a good idea.

If the driver spacing is two wavelengths you get two nulls above
and below the axis, i.e. 5 lobes, the number of lobes increasing
with increasing driver spacing.

This lobing can be a problem with 2 way MTM designs,
compared to 2 way MT designs.

This is only at the crossover frequency, driver spacing effectively
increases as you go up in frequency, causing more problems with
wide overlap crossovers, or multiple driver arrays.

sreten.

I'm very interested in this theory, where can I learn more about it?

[sreten]

Originally posted by Bricolo

I'm very interested in this theory, where can I learn more about it?

Its pretty standard acoustic theory, unfortunately I can't point
to a single site that explains it all. Linkwitz's seminal 1978 article
on 3 way design covers some of the points but not all, but I'm
having trouble finding a working link to the article.

sreten.

[5th element]

Funny thing lobing Ive never really understood why it happend fully. Now I do, seems really obvious DUH, its just constructive and distructive interference patterns from two "point" sources emitting sound at teh same frequencies. So yes lobing will be worse in MTM's if not properly designed.