Any interest in an ASR community speaker project?

[DDF]

The wide baffle design of the Ellipsa or Stradivari is very interesting and I would recommend looking at Troels Gravesen's writings on that approach.

I was interested in some of Grimm's claims so investigated how a wide baffle moves the diffraction peak and affects the in room response. Details attached and here: https://www.diyaudio.com/forums/mul...ros-cons-wide-baffles-study-grimm-ls-1-a.html

Grimm's philosophy: "What we can do is decide where to put the baffle-step frequency...If we can get reasonable, constant directivity above [300 Hz], we have a design spec!”

Some high level observations:

• The differences in power response probably most distinguish the sound of a wide baffle relative to an equally well designed narrow baffle monitor: It puts the power response “hole” in a different frequency range (see below)
• The wider baffle decreases the off axis response from 200 to 700 Hz by ~ 3 dB, relative to the standard narrow baffle. A speaker with excess 600 Hz can sound “hollow”, so this tuning will be the opposite of that (less hollow than pure accuracy?), but can also reduce the clarity of bass lines or thin out guitars.
• Equalized to flat on axis, a narrow monitor will have up to 3dB less power response from 700 Hz to 1.2 kHz than a wide baffle speaker. A speaker with less energy in this range can sound less “punchy”
• The 500 Hz to 1 KHz region produces 35% of the intelligibility, while the range from 1 to 8KHz produces just 5% of the power but 60% of the intelligibility. So the wider speaker may offer enhanced intelligibility

Narrow baffle speaker diffraction off axis, normalized to on axis:

 

[617]

I was interested in some of Grimm's claims so investigated how a wide baffle moves the diffraction peak and affects the in room response. Details attached and here: https://www.diyaudio.com/forums/mul...ros-cons-wide-baffles-study-grimm-ls-1-a.html

Grimm's philosophy: "What we can do is decide where to put the baffle-step frequency...If we can get reasonable, constant directivity above [300 Hz], we have a design spec!”

Some high level observations:

• The differences in power response probably most distinguish the sound of a wide baffle relative to an equally well designed narrow baffle monitor: It puts the power response “hole” in a different frequency range (see below)
• The wider baffle decreases the off axis response from 200 to 700 Hz by ~ 3 dB, relative to the standard narrow baffle. A speaker with excess 600 Hz can sound “hollow”, so this tuning will be the opposite of that (less hollow than pure accuracy?), but can also reduce the clarity of bass lines or thin out guitars.
• Equalized to flat on axis, a narrow monitor will have up to 3dB less power response from 700 Hz to 1.2 kHz than a wide baffle speaker. A speaker with less energy in this range can sound less “punchy”
• The 500 Hz to 1 KHz region produces 35% of the intelligibility, while the range from 1 to 8KHz produces just 5% of the power but 60% of the intelligibility. So the wider speaker may offer enhanced intelligibility

Narrow baffle speaker diffraction off axis, normalized to on axis:
View attachment 68389

The LS1 is really a peculiar design. Unique set of compromises for sure. I'd love to hear it.

 

[dwkdnvr]

I was interested in some of Grimm's claims so investigated how a wide baffle moves the diffraction peak and affects the in room response. Details attached and here: https://www.diyaudio.com/forums/mul...ros-cons-wide-baffles-study-grimm-ls-1-a.html

I actually saw this linked off the 'evidence-based speaker design thread', and it's what made me look at the LS1 as a stepping stone to the Ellipsa idea. A lot to digest, and I need to go through it again.

I do think the graphs you posted are germane to the discussion from the previous post from @617 - I think it shows that one of the big problems with directivity matching is that the baffle effects hit at basically the same frequency that the mid starts to become more directional, exacerbating the problem - it enhances the dip that occurs before the tweeter 'flare' hits. Makes me think that this might be one of the reasons some designs go for tapered or variable-width faceted baffles. Really highlights how difficult it is to do a conventional non-waveguide flat baffle design that maintains top-notch off-axis performance.

 

[617]

I actually saw this linked off the 'evidence-based speaker design thread', and it's what made me look at the LS1 as a stepping stone to the Ellipsa idea. A lot to digest, and I need to go through it again.

I do think the graphs you posted are germane to the discussion from the previous post from @617 - I think it shows that one of the big problems with directivity matching is that the baffle effects hit at basically the same frequency that the mid starts to become more directional, exacerbating the problem - it enhances the dip that occurs before the tweeter 'flare' hits. Makes me think that this might be one of the reasons some designs go for tapered or variable-width faceted baffles. Really highlights how difficult it is to do a conventional non-waveguide flat baffle design that maintains top-notch off-axis performance.

Yes, I expect that to be the major challenge with this design.

Maybe later I will experiment with a wide baffle version of the speaker. Any thoughts?

If 3d printing was a bit more affordable in volume, a big curved baffle might be a good option. See the genelec 1022a for example.

 

[dwkdnvr]

Yes, I expect that to be the major challenge with this design.

Maybe later I will experiment with a wide baffle version of the speaker. Any thoughts?

If 3d printing was a bit more affordable in volume, a big curved baffle might be a good option. See the genelec 1022a for example.

Yeah - I started down the 'Ellipsa clone' idea for purely aesthetic reasons. These speakers are going in our living room, and I think the SF speakers are the most beautiful designs I've run across - plus the wife is on board. I had concerns about the 'wide baffles don't image well' idea, but since so much of our listening in the LR is 'ambient' and not sweet-spot, I figured it was ok. Now, having looked at the design trade-offs a bit more, I find the wide baffle idea pretty interesting in it's own right. Whether or not I can pull off the construction of an Ellipsa style speaker and do it justice is of course a whole different discussion
I have to admit I'm a bit surprised that I haven't run across any LS-1 clones though. Maybe 'cloning' commercial designs isn't as popular as it once was with the advent of seriously impressive speaker design tools, but it's a well-regarded speaker with interesting design choices, and not really all that hard to duplicate. (aside maybe from the roundovers - I have some 6" ABS pipe that I'm hoping will work)

Of course, in terms of baffles the other way to go is the LX521/Nao-Note direction - baffle-less so that each driver has it's own diffraction effects. The variable width baffle idea taken to extremes. Pretty hard to go this far without going open-baffle though.

 

[617]

Some normalized directivity sonograms showing the effects of different baffle widths on different radiator sizes. First we have a 225mm woofer (Sd 213 cm2, 8" nominal)

Very narrow baffle - 9" wide.

Wide-ish baffle. Like 12" or so.

Very wide baffle - 250cm x 250 cm:

Moving on to a 125mm midrange (sd around 52 cm2)
Narrow baffle as before:

Wide-ish baffle:

Very wide baffle:

We can see that due to the frequencies involved and the increasing intrinsic directivity of the 8" woofer, changing baffle width has little effect on the woofer response, and almost no effect if a low pass is added. With the midrange, however, there is a 'peak' visible in these sonograms (appears as a little hump in the DI) where dispersion is overly wide.

This situation can be mitigated somewhat by EQin a little dip, sacrificing some on axis neutrality for off axis neutrality (not an unusual approach). Shallower crossover slopes may also help. I am not sure how putting the midrange off-axis would help, I will look at that next. I suspect it smooths on-axis but DI will remain similar but it needs to be examined.

 

[617]

Testing an off center midrange. This is the baffle in question with the three midrange positions:

First we have on center:

Big dip in the frequency response at 2k, but ignore that. The DI curve and the directivity sonogram are both 'normalized' in the sense that the linear response of the system doesn't affect them, they portray the acoustic reality of the baffle and driver size alone.

If we move it halfway to the edge:

The axial dip is a bit more benign as one would expect. DI curve appears to be a bit nicer as well - it goes up but then just sort of goes to the right, rather than going up and then down.

Finally, with the mid pushed all the way to the side:

Perhaps a bit better? Maybe a bit worse? The axial response is much more even now, but the DI is not dramatically better.

Just for fun, what if we move the midrange up to the top left corner?

Wow that's actually pretty damn good. DI and directivity sonogram are improved, and the axial response is quite civilized as well.

What do we give up when we go off center? For one thing, the speaker's response becomes assymetrical in the horizontal axis. I would argue that better room reflections are more important than a subtle difference in response 45+ degrees off axis.

I feel like I used to have a better intuition about the effects of diffraction on DI, but it's not serving me well right now. Anyway, I hope these diagrams serve to show how baffle step, diffraction and smooth DI are all related.

 

[617]

I was concerned about CTC spacing of a midrange in the top corner of a baffle and the woofer at the bottom, but I think it's not so bad, depending on the crossover frequency.
The baffle in question:

Crossover: LR4 at 200hz:
(note the sonogram is vertical in these simulations to better show the problems with CTC spacing)
I don't have the numbers right in front of me, but a good 125mm midrange can do 200hz pretty loud without too much trouble.

400 hz:

Ouch, big lobes forming. 400 hz is a bit high, however, and things only start getting really bad at 45 degrees. Still, the DI is lumpy and so are room reflections.

Let's look at 800 hz, as a treat:

Ok, this is getting quite bad. Axial response is fine but the DI has scoliosis and neutrality is really suffering off axis. What happens if we blend the drivers with a LR2 crossover? Still at 800hz:

The crook in the DI is reduced, but the issues with the crossover being up so high are still there. Perhaps LR2 at 400hz?

Not bad. How about 200 hz? 200hz LR2 might be asking a bit much of the midwoofer.

Very smooth, as we might expect, but interestingly, if you scroll up, you will see it's a bit worse than with an LR4 slope. Here they are side by side:

LR2 on the left and LR4 on the right. I always sort of assumed that blending drivers over a wider bandwidth always made for smoother transitions acoustically, but this doesn't appear to be the case. If the drivers match up well at a certain frequency, blending them can actually make things slightly worse. Now, neither of these results are bad, but the LR4 version has a wider listening window, a better DI, and has the added power handling advantages.

 

[617]

On to another issue - the small waveguide. I chatted with the designer of the waveguide who said matching the 4" sb21 SDC waveguide to a 125mm mid driver would be good, but I thought I'd look at exactly what the waveguide is doing. See this image comparison slider:
https://cdn.knightlab.com/libs/juxt...html?uid=bae75ff2-ac22-11ea-a30b-0edaf8f81e27
Vcad is kind enough to interpolate the limited polar measurements I have for the SB21 from the product cutsheet so we can make a meaningful comparison. These were measured on a big baffle if I'm not mistaken.

The first point of comparison is lower band. Without the waveuide, the tweeter is basically omni to 3K, where the traces begin to diverge. At 7K they reach another inflection point and dispersion drops off quite fast. With the waveguide, the tweeter's dispersion is never quite so wide, even at low frequencies. The 60 degree trace is 2db down from 0deg (which is minor but interesting.) At around 3K, as with the baffle mounted tweeter, the off axis starts to drop off, but does so somewhat more gradually.

What is kind of interesting is that at low and at high frequencies, the dispersion of the two drivers is quite similar. Above 10K or so, we see similar performance, and from 1-3K we see the same trend in both - unchanging wide dispersion. The difference is that the waveguide gradually and smoothly contracts directivity, whereas the bare tweeter hits an inflection point where the response drops off dramatically.

Let's see what happens if we compare it to the off axis performance of a 125mm mid on an infinite baffle:

https://cdn.knightlab.com/libs/juxt...html?uid=c5f96adc-ac24-11ea-a30b-0edaf8f81e27

If you look at the 60deg trace, things look pretty dire, but ignore that and look at 0-40 degrees, which match within 2db for much of the sensible crossover region. At 2K, the waveguide driver is down 2db at 60deg, and the midrange is down 3 1/2db or so. That's pretty damn close and I suspect a little crossover finesse could mate these pretty well.

For the sake of completeness, see the comparison of the bare tweeter and the midrange:

https://cdn.knightlab.com/libs/juxt...html?uid=852e544e-ac25-11ea-a30b-0edaf8f81e27

Looking at the 60 degree traces, we see 2db of mismatch at 1K, 4db of mismatch at 2k, 4-5db mismatch at 3K, and 8 db mismatch at 4K. I include that last figure to demonstrate that simply blending over a wider bandwidth isn't going to help much - 4K is only one octave above 2K, and we have a big increase in directivity mismatch. Perhaps Richard Vandersteen knows something we don't.

 

[TimW]

On to another issue - the small waveguide. I chatted with the designer of the waveguide who said matching the 4" sb21 SDC waveguide to a 125mm mid driver would be good, but I thought I'd look at exactly what the waveguide is doing. See this image comparison slider:
https://cdn.knightlab.com/libs/juxt...html?uid=bae75ff2-ac22-11ea-a30b-0edaf8f81e27
Vcad is kind enough to interpolate the limited polar measurements I have for the SB21 from the product cutsheet so we can make a meaningful comparison. These were measured on a big baffle if I'm not mistaken.

The first point of comparison is lower band. Without the waveuide, the tweeter is basically omni to 3K, where the traces begin to diverge. At 7K they reach another inflection point and dispersion drops off quite fast. With the waveguide, the tweeter's dispersion is never quite so wide, even at low frequencies. The 60 degree trace is 2db down from 0deg (which is minor but interesting.) At around 3K, as with the baffle mounted tweeter, the off axis starts to drop off, but does so somewhat more gradually.

What is kind of interesting is that at low and at high frequencies, the dispersion of the two drivers is quite similar. Above 10K or so, we see similar performance, and from 1-3K we see the same trend in both - unchanging wide dispersion. The difference is that the waveguide gradually and smoothly contracts directivity, whereas the bare tweeter hits an inflection point where the response drops off dramatically.

Let's see what happens if we compare it to the off axis performance of a 125mm mid on an infinite baffle:

https://cdn.knightlab.com/libs/juxt...html?uid=c5f96adc-ac24-11ea-a30b-0edaf8f81e27

If you look at the 60deg trace, things look pretty dire, but ignore that and look at 0-40 degrees, which match within 2db for much of the sensible crossover region. At 2K, the waveguide driver is down 2db at 60deg, and the midrange is down 3 1/2db or so. That's pretty damn close and I suspect a little crossover finesse could mate these pretty well.

For the sake of completeness, see the comparison of the bare tweeter and the midrange:

https://cdn.knightlab.com/libs/juxt...html?uid=852e544e-ac25-11ea-a30b-0edaf8f81e27

Looking at the 60 degree traces, we see 2db of mismatch at 1K, 4db of mismatch at 2k, 4-5db mismatch at 3K, and 8 db mismatch at 4K. I include that last figure to demonstrate that simply blending over a wider bandwidth isn't going to help much - 4K is only one octave above 2K, and we have a big increase in directivity mismatch. Perhaps Richard Vandersteen knows something we don't.

These links aren't working for me.

 

[617]

These links aren't working for me.

Bummer, here are the image comparisons side by side. The slider tool is really handy but it appears not to work.
With waveguide / without waveguide. Notice waveguide constrains directivity even at lower frequencies, and that the differences in dispersion at extremes are not that big.

Here's the midrange / tweeter in WG. They match pretty well in the potential crossover region. I would expect very good matching if an LR4 slope is used. Then again these are infinite baffle simulations.

And finally the midrange / bare tweeter. Even down very low they don't match well, and they match very poorly as frequency increases. Other tweeters may not have as distinct an inflection point as the sb21, which is basically 180 degrees to 7K.

For example here are the manufacturer FR files for the Dayton RS125p (not a vCad simulation of off axis) and the rs28f. Rs28f is a very robust tweeter, you could use it 1800LR2 if you wanted. It looks like you could match these pretty well at 2K, where they are both giving very wide dispersion. The sb21 is cheaper, smaller and gives better top octave off axis, however. Still, not exactly bad performance, and it would enable a 100% dayton design. One of the reasons I love dayton is that they let you download FRD files of all their drivers.

 

[DDF]

I feel like I used to have a better intuition about the effects of diffraction on DI, but it's not serving me well right now. Anyway, I hope these diagrams serve to show how baffle step, diffraction and smooth DI are all related.

To avoid having a cockeyed driver mounting on baffle, they can be designed for no toe in or excessive toe in and that spreads out the diffraction signature similar to the way off center mounting would.

BTW/FYI johnk's page had good sims of directivity and power response http://musicanddesign.speakerdesign.net/Power.html

Any idea if anyone's validated Vituix Cad's diffraction model? I compared the Edge and the popular tools from the Madisound/Parts Express forums with measures from an anechoic chamber, but haven't had the chance to check Vituix CAD.

 

[Lbstyling]

I was concerned about CTC spacing of a midrange in the top corner of a baffle and the woofer at the bottom, but I think it's not so bad, depending on the crossover frequency.
The baffle in question:
View attachment 68431

Crossover: LR4 at 200hz:
(note the sonogram is vertical in these simulations to better show the problems with CTC spacing)
I don't have the numbers right in front of me, but a good 125mm midrange can do 200hz pretty loud without too much trouble.
View attachment 68434
400 hz:

View attachment 68435
Ouch, big lobes forming. 400 hz is a bit high, however, and things only start getting really bad at 45 degrees. Still, the DI is lumpy and so are room reflections.

Let's look at 800 hz, as a treat:
View attachment 68436
Ok, this is getting quite bad. Axial response is fine but the DI has scoliosis and neutrality is really suffering off axis. What happens if we blend the drivers with a LR2 crossover? Still at 800hz:
View attachment 68437
The crook in the DI is reduced, but the issues with the crossover being up so high are still there. Perhaps LR2 at 400hz?
View attachment 68438
Not bad. How about 200 hz? 200hz LR2 might be asking a bit much of the midwoofer.

View attachment 68439
Very smooth, as we might expect, but interestingly, if you scroll up, you will see it's a bit worse than with an LR4 slope. Here they are side by side:

View attachment 68440
LR2 on the left and LR4 on the right. I always sort of assumed that blending drivers over a wider bandwidth always made for smoother transitions acoustically, but this doesn't appear to be the case. If the drivers match up well at a certain frequency, blending them can actually make things slightly worse. Now, neither of these results are bad, but the LR4 version has a wider listening window, a better DI, and has the added power handling advantages.

Nice.

Could you simulate LR6 and LR8? And see if you can raise the crossover point without making it worse?

 

[dwkdnvr]

To avoid having a cockeyed driver mounting on baffle, they can be designed for no toe in or excessive toe in and that spreads out the diffraction signature similar to the way off center mounting would.

BTW/FYI johnk's page had good sims of directivity and power response http://musicanddesign.speakerdesign.net/Power.html

Any idea if anyone's validated Vituix Cad's diffraction model? I compared the Edge and the popular tools from the Madisound/Parts Express forums with measures from an anechoic chamber, but haven't had the chance to check Vituix CAD.

Yeah, I wonder whether these are a bit misleading due to the model. You can see from the comparison that @617 posted in the last post that the theoretical 5" driver has far worse off-axis response than the measured response of the RS125. The degree to which this impacts the analysis isn't entirely clear though - do the same conclusions hold with slightly lower artifacts, or will in some cases the conclusions actually change? I think we need to get to the point of measuring one of the simulation scenarios to compare, which takes time.

 

[617]

Thanks to everyone so far for their attentive reading of these very theoretical explorations into driver sizes and crossover points. As has been pointed out, there are some conspicuous gaps between simulation and reality, particularly in the directivity of actual drivers. I believe VituixCAD's model of piston directivity is useful, but as we can see with the rs125p, the reality is a bit more complicated, and perhaps better, than VCad indicates. Perhaps the phase plug is a factor - perhaps breakup. The biggest uncertainty at this point is the behavior of the WG on the baffle. I've explored what the waveguide does on an IB, and it seems to create a seamless transition to a 4" midrange over a wide bandwidth, but how the actual cabinet diffraction plays into this must be examined.

Now, onto some news. SIY has provided measurements of the Dayton amp, and the results are not that good. Actually, for the price I think this is still an amazing product, but I'm not sure I'd use it in a speaker with the level of performance we are attempting.

See the following AP captures:




The power output is not that good; the supplied power supply only puts out 100W, with which we are powering 200W of amplifier boards. This might be rectified with a bigger power supply.

Distortion performance is not impressive. SINAD at 5W appears to be 48-58 db, which is pretty terrible for a hifi amp. Distortion seems to be more of a factor than noise.

When I get the amps I will listen to them myself and make a determination of their usability, but even if I decide they sound great, they severely limit the appeal of the project to audiophiles, perhaps for good reason. Who knows, maybe the amps in the JBL 708 are just as bad! Probably not.

So, this leaves us with a decision. Since the existence of these DSP amps was the inspiration for the 3 way design, we have to revisit some of our earlier project assumptions and choose a path forward.

1. We can find another 3 channel amplifier/dsp solution. One logical solution would be an IcePower 200ASC and a IcePower 50x2 module. Unfortunately the 50W module cannot be run off the supplied power rails of the 200ASC (it has its own power supply) so you'd be wiring them both to the same iec socket.
   • For DSP, Dayton sells a little preamp module with the same DSP processing as the amps. With the 200asc and 50asx2, this would be a total cost of $278, not including an iec socket, input jacks and wiring harnesses. Call it 300 dollars per channel.
   • Alternatively, it might be more appealing to use the caseless minidsp module. This costs $50/$75/$165 more, depending on if the standard/balanced/HD version is chosen. I would favor the balanced version as it has screw terminals instead of RCA jacks, and a balanced input would be a good feature for an active speaker.
   • With the balanced MiniDSP, total cost per channel for a 3 way balanced icepower implementation would be $353
2. Alternatively, most of the advantages of a 3 way active design can be achieved with an active/passive hybrid design.
   • The options for stereo amps are the 50asx2 (50x50) which might be a bit underpowered, the 125asx2, which is only available on ebay (I built one a while back) or a 200ASC+200AC module, where the 200AC runs of off the power supply of the 200ASC
   • In addition, a basic crossover network would be needed just for the mid low pass and the tweeter high pass filter; maybe budget 30 dollars for these parts.
   • A dsp unit would be required, but in this case you'd only be using 2 out of the 4 channels, which seems a bit wasteful given the fact that one of these dsp units could do the whole job.
   • For the three stereo icepower amp options, prices are $118, ~$140 and $201.
   • Another alternative would be to use the miniDSP plate amps, which of course include icePower and DSP in an easy to install package - these are $340 and $430 for the 125/250W versions. This isn't a bad deal at all - for one thing it's very easy to install, and the cost of a balanced minidsp/icepower 125asx is around $240, so you're paying about a hundred dollars to get a bunch of connectors and a huge amount of convenience.
3. A bigger change of direction, but a sealed 8" two way with the icepower 125W amps could be really good. I recently got a 3d printer, and I think I could print a big 8" WG which, combined with a good woofer and minidsp plate amp, work really well.

Anyway, please provide your thoughts moving forward. The most important decision is whether to make an active 3 way (expensive), a hybrid 3 way (somewhat cheaper but requires soldering), an active two way (relatively easy to build but requires a big WG) or whatever other option.

 

[617]

In other news; I tried doing some measurements outside last weekend and it was just a mess. My measurement set up works but it lacks precision and is a major PITA to use. Since @hardisj pointed out the existince of these 'tic' stepper controllers, I have decided to revisit the idea of a computer controlled measurement platform. At the most basic implementation, I should pretty easily be able to control the tic using either its windows gui, or a series of windows shortcuts which can be assigned to move the stepper certain distances, but I am also trying to get a program working which allows ARTA to do measurements fully automated. My friend who knows python is currently working on this.

The mechanical side of the measurement platform is more in my wheelhouse. I plan on using a NEMA 17 or 23 motor, controlled by the tic, to drive the platform using a GT2 belt. I plan on 3d printing a big 225 tooth gear for the platform, which combined with the 25T gear on the motor, gives me a torque increase of 9 to 1. Just as importantly, it breaks down the stepper motor's steps into smaller increments. Steppers generally use 200 steps to rotate 360 degrees, which means they increment at 1.8 degrees. This makes it impossible to use a stepper directly to rotate 1 or 10 or 15 degrees. By using a 25/225 gear reduction, the stepper's rotation of 1.8 degrees is reduced to .2 degrees, which enables any whole-degree measurement I want.

The whole platform consists of a 36" square plywood base with a sort of triangular frame made of 2x6 lumber. See following pictures;

There are four threaded studs on the top of the carraige that allow you to attach a tall platform of whatever height you want. I have one which is around 6', ideally I'd like to come up with something taller, but accessing the top becomes difficult. A telescoping design might make sense.

 

[somebodyelse]

Given the cost of those Dayton/Wondom DSP amps the performance isn't entirely surprising. It's one reason I'd like to see the internals of something like the JBL 305 measured.

Some other DSP options:
https://www.3e-audio.com/dsp/adau1701-2in4out/ - should be comparable the the balanced non-HD miniDSP, but needs programming like the Dayton/Wondom board
Hypex FusionAmps - the FA122 is broadly equivalent to the PWR-ICE125 and cheaper, at least on this side of the Atlantic. The FA123 is a little more than the PWR-ICE125, but gets you an extra channel of amplification for an active 3-way.
https://auverdion.de/aurora-dsp/ - not been measured so far as I know, but an interesting feature set, and unlike the others mentioned so far it doesn't require Windows.

 

[DDF]

Yeah, I wonder whether these are a bit misleading due to the model. You can see from the comparison that @617 posted in the last post that the theoretical 5" driver has far worse off-axis response than the measured response of the RS125. The degree to which this impacts the analysis isn't entirely clear though - do the same conclusions hold with slightly lower artifacts, or will in some cases the conclusions actually change? I think we need to get to the point of measuring one of the simulation scenarios to compare, which takes time.

These types of models where you enter in driver size and mounting and baffle dimensions model two things, diffraction and driver radiation off axis (dispersion). The error you mentioned is mostly due to the model of driver dispersion, not the diffraction. It assumes the driver radiation is the solution to the wave equation for a rigid pistonic disc. A real speaker cone enters into break up and emits like a smaller cone at higher frequencies, broadening radiation there. There are no physical parameters specified for speaker drivers that lets anyone model this break up except the driver's designers that have access to the FEM model.

To have a hope of an off axis simulation useful for DIY design, the actual driver radiation on an IEC baffle needs to be used (available from the manufacturer or retailer), the IEC baffle's diffraction "backed out" using the tool's diffraction modeler, and then the diffraction of it mounted in the target box modeled and added back in. This has to be repeated at every observation/measurement location. I wrote this up here: http://audio.claub.net/software.html -> http://audio.claub.net/software/DaveDalFarra/Simple Loudspeaker Design ver2.pdf I wrote this up quickly for a DIY meet and Charlie and Jeff then went one better and later created their splice tool.

What I'm curious about is the second part, how well Vituix Cad models diffraction. It doesn't need to get this "wrong" by much to create audibly misleading results because the error will exist over a wide frequency range, so even small dB errors will be noticeable. Here's how the other tools fared compared with anechoic measurements. The Edge vs Paul's BDS vs Jeff Bagby's DBS (based off Paul's BDS so no surprise those two are close):

 

[DDF]

Distortion performance is not impressive.

Thanks for posting those results. Agree that that amp would be a bad candidate for a high res design. No way that splash of really high harmonics will be consistently masked.

 

[617]

Thanks for posting those results. Agree that that amp would be a bad candidate for a high res design. No way that splash of really high harmonics will be consistently masked.

Right. I will probably make an outdoor speaker with them as a gift for the guy who wrote the ARTA program for me. If he succeeds. He's fortunate enough to not be an audiophile.