I'm in the process or trying to put together a design for a WMTMW and have a question about the optimum method of physically time aligning the drivers.
I think there are 2 main options, I can place a woofer and mid in the same plane and tilt them relative to the tweeter or offset the tweeter against the woofers and tilt only the mid (by about 10 degrees)
Which would be the best option please?
I think there are 2 main options, I can place a woofer and mid in the same plane and tilt them relative to the tweeter or offset the tweeter against the woofers and tilt only the mid (by about 10 degrees)
Which would be the best option please?
Attachments
OK, thanks MaVo.
I haven't found any (free) software that will give a complete full simulation of what I'm trying to design. I haven't seen anything that will deal with angled front baffles such as I'm proposing at all.
Have you got any suggestions of something I can use please?
I've used The Edge, Unibox, TPSD Multi and WinISD to put together what I have so far.
I haven't found any (free) software that will give a complete full simulation of what I'm trying to design. I haven't seen anything that will deal with angled front baffles such as I'm proposing at all.
Have you got any suggestions of something I can use please?
I've used The Edge, Unibox, TPSD Multi and WinISD to put together what I have so far.
I am also just learning about this stuff, but maybe you can find something helpfull here: http://www.pvconsultants.com/audio/frdgroup.htm
Much to expensive, but this will do it probably: http://www.linearx.com/products/software/LEAP5/LEAP5_01.htm
Much to expensive, but this will do it probably: http://www.linearx.com/products/software/LEAP5/LEAP5_01.htm
MaVo said:
The time delay to which you refer is not associated with the frequency-dependent delay. The latter is the minimum-phase response of the driver related to the transfer function of the driver as is the delay introduced by a crossover. The time alignment related to the acoustic center is a separate issue and as used in most CAD software is a fixed point, not frequency related, assumed to be the origin and specified in the model as such.
I believe that SoundEasy has the capability to create a model that incorporates driver axis rotation if desired.
Dave
MaVo, I'd love a copy of LEAP but I think $1500 is pushing it a bit for someone with my low level of understanding of all this (at the moment) I wouldn't make the most of it by any stretch of the imagination 
Thanks dlr. So it's "just" a case of physically aligning the drivers using the baffle shape that'll give me either the smallest amount of deviation of response or the deviation that is most easily compensated for?
I'll look at SoundEasy, thank you.
Thanks dlr. So it's "just" a case of physically aligning the drivers using the baffle shape that'll give me either the smallest amount of deviation of response or the deviation that is most easily compensated for?
I'll look at SoundEasy, thank you.
quickshift said:
Yes, the alignment is an offset to align the two acoustic centers. The summed response target can be achieved much more easily when aligned. The individual drivers are designed to a target for each driver. When combined, the phase response of each leg of a crossover pair should be very close to the ideal so that little or no slope alteration is required to achieve the target summed response.
BTW, SoundEasy is not free software, but it's excellent for the price and is a complete package for almost any DIY efforts.
Dave
Let me suggest the following reference for some useful info:
John K's non-commercial Web Site
There's more than one alignment that may be desired. Time alignment (of acoustic centers) is not necessarily the single best goal. There can be reasons to align a bit differently, such as trying to align drivers with a crossover such that the phase at or in the crossover area is closer to optimal. This might be useful if there are limitations such as a difficult driver or a desire to minimize crossover components.
Most often I find that "getting close" is good enough because small deviations from ideal are almost not even noticeable in measurements. The exception to this would be transient-perfect types where the goal is to get close to accurate square wave reproduction.
All of this applies only on a single axis in any case.
Read John's pages, they should answer a lot of questions.
Dave
John K's non-commercial Web Site
There's more than one alignment that may be desired. Time alignment (of acoustic centers) is not necessarily the single best goal. There can be reasons to align a bit differently, such as trying to align drivers with a crossover such that the phase at or in the crossover area is closer to optimal. This might be useful if there are limitations such as a difficult driver or a desire to minimize crossover components.
Most often I find that "getting close" is good enough because small deviations from ideal are almost not even noticeable in measurements. The exception to this would be transient-perfect types where the goal is to get close to accurate square wave reproduction.
All of this applies only on a single axis in any case.
Read John's pages, they should answer a lot of questions.
Dave
I have seen it recommended in the Loudspeaker Design Cookbook and elsewhere that the acoustic center is approximately where the the cone meets the voice coil.
There is a good two part article on WMTMW transient perfect speakers in the last two AudioXpress issues. Were these articles your inspiration? I don't remember what was recommended for aligning acoustic centers in that article.
I think the flat recessed option is necessary to get the mids to line up with the woofer.
JJ
jupiterjune said:
The junction is a good estimate. I experimented for quite some time and concluded that it's somewhere slightly in front of the junction, but no one has been able to measure it conclusively, not even the AES.
The article you mention is not from any inspiration of mine. Speakerbuilder, a precursor to audioXpress, published an article of mine on relative acoustic offset in 2000, though. It's posted at my web site.
Dave
dlr said:
Very useful, thank you.
This will be a fully active system which should make things slightly simpler with a bit of luck. I still have a fair bit of work to do before I actually put saw to MDF though I think.
I have read a bit on the audibility thresholds of driver offset and it does seem like there is a little leeway. I'm hoping these speakers will be "keepers" so really do want to get it right.
jupiterjune said:
Yes, that was what I was going to base the acoustic center on. The bulk of what I've read agrees with that.
I haven't read AudioXpress but it sounds like I should try to get hold of the last two issues.
I think either option will work, I've got a cardboard version of one side of the speakers built just so I could see how everything fits together; just the mids are angled and all the drivers line up OK.
dlr said:
There is some very useful info on your site too, I've read through it before and am running through it again now as it's certainly very helpful.
The goal is to match the acoustic group delay (including crossover response plus driver and enclosure response) across the crossover region.
The best proof of matching is a wide and deep notch on-axis across the crossover region when the polarity of one of the drivers is reversed.
The easiest way to achieve this is by going active and using a low-cost adjustable digital crossover, like the DCX2496, to tune the system.
Optimum delay has not much to do with acoustic center alignment. Every driver has its associated propagation delays and group delays
The best proof of matching is a wide and deep notch on-axis across the crossover region when the polarity of one of the drivers is reversed.
The easiest way to achieve this is by going active and using a low-cost adjustable digital crossover, like the DCX2496, to tune the system.
Optimum delay has not much to do with acoustic center alignment. Every driver has its associated propagation delays and group delays
Eva said:
I'm not sure why one would be concerned with matching group delays. The easiest method is to achieve target phase response of each leg of the crossover. The summed response would then be correct.
In addition, polarity reversal notches do not occur for every crossover. For even-order Butterworth and Linkwitz-Riley, yes. However, for odd-order Butterworth the inverted connection would show absolutely no change, they would simply invert the lobes. The summed response would still be flat on-axis. Other crossover types (not often used) have similar issues.
All of the above hold only for aligned drivers as well.
Dave
Sorry, you don't seem to get the point.
The purpose of alignment is to get optimum summing on axis ensuring some attenuation off-axis. This rules out any crossover type that does not result in acoustical outputs in phase across the crossover region.
The only (analog) filter types suitable for alignment are Bessel and Linkwitz-Riley.
First order, Butterworth and the rest always gives a misaligned system (by 45º or more) with a poor polar response (maximum response happens off-axis) where driver position does not matter much.
The purpose of alignment is to get optimum summing on axis ensuring some attenuation off-axis. This rules out any crossover type that does not result in acoustical outputs in phase across the crossover region.
The only (analog) filter types suitable for alignment are Bessel and Linkwitz-Riley.
First order, Butterworth and the rest always gives a misaligned system (by 45º or more) with a poor polar response (maximum response happens off-axis) where driver position does not matter much.
Eva said:
Please don't tell me what I don't get, I understand it just fine. Your definition of optimum summing is simply your preference of a design goal. Your first post was incomplete and incorrect as presented, since it did not clarify the crossover. There are perfectly valid reasons to use crossovers other than your preference. Alignment is important for all crossovers, that is alignment of the design axis.
In addition, if the drivers are not physically aligned for their acoustic centers, systems such as the DCX2496 that use a time-delay for one or more drivers that makes their off-axis response deviate from flat with lobing as well. They sum flat only on-axis.
Dave
MaVo said:
Here is a great paper on phase response and it's effects by John Kreskovsky. Dave linked to his site earlier.
Regards.
Dennis
http://www.geocities.com/kreskovs/Phase-B.html
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.