The Mentor II: omnidirectional project #2

[dlneubec]

Here are some construction photo's. The first one show the RS52baffle from below, with the 8"PE waveguide. I've treated the WG with plumbers putty to dampen any vibration and resonance in the plastic. There was a dip of about 3db in the RS52 response centered at about 1.7khz that flattened out as a result. If you look at the previously posted RS52 measurments, you can see where the plumbers putty application filled in. The plumbers putty treatment was also done to the ND20 waveguide (MCM6.5").

The second photo is of the top box with bracing pictured from the woofer end. For now, the baffles simply screw on to the structure, but they will eventually be connected with hurricane nuts and cap screws, probably combined with a layer of anti-vibration material of some type. I'm debating whether to make the baffle out of hardwood for aesthetic reasons. If I do mdf, I will most likely have to paint the baffles. I'm not too keen on trying to apply veneer on those chamfers!

The remaining two photo's show the same box, but with the top baffle set on top. I have trimmed the top diffuser down and added a top chamfered piece to finish it off and hide the hardware. I think it looks nice enough like this to finish it without covering the top. The top (RS52) baffle will eventually be remade. The holes you see that were originally for dowels to hold up the diffusers, etc., are no longer needed with the way I've mounted them. I do plan on making the bolts that hold the ND20/WG shorted, and to use cap nuts that will set in the recessed area rather than stick up as they do now.

Finally, there is a plot of the ND20(blue) and RS52(green). I was concerend that the RS52 FR would be substantially changed due to cutting down the concave diffuser from it's original 13" dia. to 8" dia. Also, my original measurments of the ND20 in WG was with it sitting on about a 10"x18" wide box top/baffle. I was concerned that going to a round baffle with the WG recessed into the RS52 diffuser would also result in some negative results. The responses are different, but not as much as I feared. Interestingly, both the RS52 and ND20 have a more ectended low end than they did in previous meaurements. The smaller diffuser for the RS52 and round baffle for the ND20 did seem to result in some loss of flattness to the FR's at their upper ends. I would imagine that some of the dips you see are due to diffraction from the round baffles, but it doesn't look too bad to me.

[dlneubec]

Nothing much new, but thought I'd post the plans I'm following (pretty much so anyway) for the construction. There are a bizillion dimensions due to the angles, so I turned most of them off. I'm hoping my woofer to mid and mid to tweeter distances work out ok with my preliminary thoughts on crossover points, around 600hz and 4.2khz or so.

Believe it or not, I need 8 pairs of binding posts! :E This is because I want to put all three crossover sections in the bottom box. So I have one pair for input that goes to the crossovers, then a pair from each crossover that has to be connected to the top box and then one additional pair that connects to the tweeter lead up above.

I hope to mount the crossovers on the bottom section of 3/4" mdf, which will be removeable. I have about a 9-1/2"x11-1/2" area to fit them all in, so I'm crossing my fingers that my inductors and caps don't get too big.

I have two types of damping sheet material coming from PE to use at the back of the removable baffles, which might end up hardwood, and a bunch of binding posts. I would like to finish up the prototype build and start design measurements this weekend, if I can figure out the best way to take them in this case!

I should have more photo's once I get the prototype build done in the next couple days. I'm still struggling in coming up with a good looking way to transition from the bottom box to the floor. You can see what I'm thinking for now on the plans. If anyone has any thoughts or ideas, I welcome them.

[joecarrow]

Is that eight pairs per speaker?? A total of 16 pairs, or 32 individual binding posts? I hope you got a quantity discount!

I hope it wasn't too costly. This project is looking good- I'm glad to see all the progress.

[dlneubec]

Is that eight pairs per speaker?? A total of 16 pairs, or 32 individual binding posts? I hope you got a quantity discount!

I hope it wasn't too costly. This project is looking good- I'm glad to see all the progress.

Hi Joe,

Yes, that is 8 pairs per speaker! :E If I put the crossovers in the top box, I could have reduced it to 4. However, I had room in the bottom box and could make the crossovers removeable and accessible from the bottom, so I decided that was the best solution for me. That's part of the price you pay for having drivers separated between essentially three boxes (or housings) BTW, I had 7 pairs/speaker in the original omni, but 2 pairs were internal.

Actually I ordered just 2 single red and 2 single black 1-1/2" long binding posts from Madisound when I picked up the Seas and Aurasound tweeters to test, knowing I would need some, but I had not though about how many I would need at that time. They are only $1.55 each, but the surprise was they sent me 4 pairs, not 4 singles. I think someone at Mad messed up. At any rate, I ordered another 6 singles, enough to complete one prototype speaker. If they mess up again and send my 6 pairs, then all the better. :W Even $3.10/pair is about 1/2 the price of similar ones at PE.

[JoshK]

As a entertaining tangent, you might have a look at the "walsh F' thread on diyaudio.

here

Starting on about page 6 BudP (Bud Purvine of Onetics Audio, who made my output transformers for my tube projects) lends some of his expertise on the bending wave theory and boundary value problems. The concept supposedly works for monopole direct radiators as well, but in this case they are describing it in context of an omipole speaker.

I'd like to see our dark master wrap his force around the concept.

[dlneubec]

As a entertaining tangent, you might have a look at the "walsh F' thread on diyaudio.

here

Starting on about page 6 BudP (Bud Purvine of Onetics Audio, who made my output transformers for my tube projects) lends some of his expertise on the bending wave theory and boundary value problems. The concept supposedly works for monopole direct radiators as well, but in this case they are describing it in context of an omipole speaker.

I'd like to see our dark master wrap his force around the concept.

Hi Josh,

I have been watching that thread, but not really understanding the bending wave and boundary value issues. I don't know if there is something there I can use or not. ops:

I'd like to see some musings from the Dark Master myself in regards to those concepts and my humble omnipole attempts here as well!

[dlneubec]

For anyone still interested, here are some pics of the nearly finished prototype. I only have to add the trim/legs around the base, which will raise it up about 1/2" or so.

I got the idea to try solid acrylic rods in place of the hardwood dowels between the two boxes. I like how it looks and may even use acrylic tubing in place of the aluminum tubing where I currently have it.

I also added a combination of 1-1/4" PE sonic barrier to all 4 sides of both boxes and then filled about 75% with acousta-stuff for the remaining areas. Finally, I added a layer of mineral filled viscoelastic polymer damping material between the woofer baffles and the 3/4" mdf behind them. It has a self adhesive backing and was simply cut and stuck in place. I've also added the binding posts to the back and connected the internal wiring between them.

I have done a bunch of measurements and there will be more on the effect of stuffing and wall lining in subsequent posts.

[dlneubec]

Here are some comaprisons of one of the RS225's (bottom box) before and after adding sonic barrier, acousta-stuf and the mineral filled vinyl constrained layer behind the baffle.

The first plot is the impedance of the RS225 without any fill in the box. The second plot is the impedance of the RS225 after the stuffing, etc. has been added. The stuffing seems to have generally lowered the overal impedance up through 100hz, but resulted in about a 8ohm peak at 55hz.

The third plot is the nearfield FR and phase of the drivers with and without the stuffing etc. The orange line is without and the blue line is with. Not only is the FR smoother with the stuffing, etc. it is more extended on the bottom end, starting at about 50hz, up almost 5db at 20hz in this 25l sealed box. The phase plots are the same color, but the smoother one is the one with the stuffing, etc. added.

[joecarrow]

Ok, I've got to ask- and I know this is premature- how do they sound??

The bass is promising, but I can't help but notice the tweeter response. Didn't you have better response than that before? Or is this measurement more concerned with bass? If this is an honest measurement of what that 4800 hz crossover does, I'd say it would be worth measuring it, and if it's second order maybe increase the slope.

[dlneubec]

Here are some FR comparison measurements for the drivers. I was unsure (still am) of what method to use for the design measurements, so I tried various methods to see how they compared.

The first graph shows the RS52 and ND20 with the mic at both 1 meter and 2 meters and 38" high. There seems to be little significant difference between the two.

The second graph shows the RS225 with the box or mic in various positions and with no baffle between the woofers and with the rectangular chamfered baffle between the woofer as shown in the pics in the post above. One setup was with the box sitting on the floor, the mic up 13.5" setup and 1 meter away. The second setup is with the box raised up so the woofers were centered at 38" high and the mic was at 38", 1 meter away.

The third graph shows the RS225, RS52 and ND20 at 1 meter with the mic at 38". The 4th graph is the same as above, but at 2 meters. Not much difference between these either, so I think I could use either 1m or 2m measurements.

The final graphs shows the RS52 and ND20 on axis and also at 45º and 90º off axis. The 45 and 90 degree off axis FR's track the on axis response very closely, so omni effect and the power response without crossover should be good for 360º.

I did not yet do any vertical polar measurements, but plan to do so.

Does anyone have any suggestions as to what set of measurements they would use for design?

[dlneubec]

Ok, I've got to ask- and I know this is premature- how do they sound??

The bass is promising, but I can't help but notice the tweeter response. Didn't you have better response than that before? Or is this measurement more concerned with bass? If this is an honest measurement of what that 4800 hz crossover does, I'd say it would be worth measuring it, and if it's second order maybe increase the slope.

Hi Joe,

These are nothing but raw response measurements from each driver in SoundEasy using a 4ms gated digital MLS measurements. There is no crossover yet, so there is nothing to listen to.

Heck, I don't really know how to start on the 3way crossover design and I can't seem to get any help with suggestions for crossover points relative to my CTC distances or the raw responses that I have.

The raw response of the tweeter was a bit flatter before, but that was when the waveguide was sitting on a flat baffle of about 10"x16", and now it is sitting in an 8" round concave diffuser. I suspect that some of the changes are due to diffraction from what in effect is a round baffle for the tweeter.

[Paul W]

Dan,
The reason you are not getting suggestions may be that you are creating the DIY knowledge base with your projects

Without vertical measurements (which I believe are critical) I'll venture around 700-800 and 3k for xo frequencies. You may need to "cut and try" to get power response where it sounds best. In other words, the typical "flat on axis at 1M" might not be best when (very high) room contribution is included. It may be a cyclic process between relatively close range MLS (1M-2M, outdoors if possible) and listening window RTA measurements.
Paul

[dlneubec]

Dan,
The reason you are not getting suggestions may be that you are creating the DIY knowledge base with your projects

Maybe you're right about this. I do have what I would think are some basic questions about 3ways that are more generic. Here is a few off the rop of me head, for example:

1) How do you measure CTC distance from the woofer to mid with a WWMT, furthest center to center, closest center to center, or a middle point between?
2) What happens when you exceed the max CTC distance in terms of performance, crossover design troubles, etc.?
3) How important is the 3octaves between crossover point rule of thumb for 3ways, and what happens when you have less than 3 octaves?
4) Is there a "typical" proceedure for a 3way design? I can find nothing in D'Appolito or Dikason or on the web that outlines what steps one should consider and in what order. For example, assuming you have measurements, do you start with the mid, using your target crossover points, network type and slopes and then add the woofer and tweeter nets?

Here is one that is omni in nature:

1) Do I have to make baffle step adjustments in an omni? John K. says no, because the driver stays in 4pi space and never transitions to 2pi space. Jon Marsh says yes and that it only has to do with baffle size. How am I to determine if this is a factor or not? I'm totally confused by this one.

Without vertical measurements (which I believe are critical) I'll venture around 700-800 and 3k for xo frequencies. You may need to "cut and try" to get power response where it sounds best. In other words, the typical "flat on axis at 1M" might not be best when (very high) room contribution is included. It may be a cyclic process between relatively close range MLS (1M-2M, outdoors if possible) and listening window RTA measurements.
Paul

Thanks for the suggestions. 700 to 3k is a little over 2 octaves. Is there a way to evaluate if this is ok, when the rule of thumb is 3 octaves? BTW, what led you to you pick 3k?

When you mention "power response" are you speaking of vertical power response? If not, then I'm confused (not suprise there), since it seems to me that the omnipolar nature of the speaker and the comparisons of 1m, 2m, and 0º, 45º, and 90º off axis measurements suggest that output is equal in all directions. What am I missing? Or are you suggesting that the room contribution could be such that you can't rely on standard measurements? It seems to me that room contribution might be much more uniform and predictable, since the output is the same in all directions, versus a typical speaker which has much different contribtion from the front wall, rear wall and side walls.

What is "listening window RTA measurements"?

[joecarrow]

I can venture a few guesses on these, and give you some basis for why I say what I do.

1) With two woofers and a mid, you measure it from the mid to each woofer. If it's 30 inches from the mid to one of the woofers, you'll have the same bad effects if the woofer is playing up to a frequency that shouldn't be more than 10 inches from the acoustic center of the entire system.

2) When you exceed the "rule of thumb" center-to-center distance, you run into problems with how the response combines. Plotted in the vertical axis, these look like lobes, with different angles off of vertical having peaks and dips at a given frequency. The frequency response for a given point in space will get more ragged, since at some frequencies there is constructive interference (which is what you want- all of the sound do show up at the same time)- and destructive interference.

If you had the woofer and mid coinciding in space, then there'd be none of this effect at all and you would just have to worry about phase issues induced by the driver's natural behavior plus the effect of the crossover.

If it turns out that the second woofer is too far from the mid for an effective crossover, it might be necessary to turn it into a 3.5 system. Fortunately, your preliminary measurements don't seem to suggest that to me. It looks like more of your problems are in the mid/tweeter, with the mid playing too high.

3) This rule of thumb is for passive crossovers. If you build them closer in frequency, then they start to interact in ways that aren't described in texts on crossovers, and instead require serious analysis of the entire global circuit. They become increasingly sensitive to component values, and it can be a huge problem trying to maintain a reasonable impedance- with possibly crazy low/high impedances and wildly swinging phase at low impedance.

If you have just one of the two crossovers in a 3-way as active, this is no longer a concern because the two crossovers don't interact electrically anymore, since the amplifier serves as a buffer. Actually, if you had the two sections bi-amped with totally separate passive filters, that would do it too.

4) Sorry man, I've read this stuff on the internet, and supplement that with my mechanical engineering education. The "right" way or even a "good" way to design a three-way crossover just isn't something I've learned yet. I bet that some of the gurus here could be a big help with that. You're certainly putting in some equity here to show that you're not just another n00b who wandered in wanting everything handed on a silver platter.

And, regarding the omni question- I'm quite certain that you don't need baffle step compensation. Here's my justification.

Starting with the woofers- there's not really any difference acoustically with how the woofers interact with the room. At their frequencies, they're basically omni anyway and their waves wrap around the cabinet easily and reach whatever wall is nearest and are reinforced by its presence.

Next, we come to the tweeters- in a conventional speaker their wavelength is a small portion of the face of the baffle. Their output thinks it's just radiating into half a room, until it gets to the cabinet edge. It's reinforced by that wall it starts out next to, all the way down (the mids, too) to a frequency dictated by the width of the baffle. Your Mentor doesn't have the mids or tweets firing off of a front baffle, so they don't get that reinforcement. Instead, their omni all the way from the lowest low to the highest high.

If that made sense, hopefully you'll agree with me that "baffle step" is more about *gain* above a certain frequency, than loss below it. Your design simply fails to have that gain.

Are you sure Jon Marsh said you'd need baffle step for this one? He's a pretty sharp guy, and I think he'd agree with this answer. It'd actually make me a little nervous if he didn't- at that point I'd feel better if we all got together and re-examined the problem definition.

[joecarrow]

By the way, I agree with a low and steep crossover between the mid and tweeter, in case that wasn't clear. It's hard to say *what* you'd call the spacing for those, so it's probably best to just minimize high frequency interaction.

[dlneubec]

Joe,

Thanks for taking the time and making such a detailed response. It is very much appreciated!

2) When you exceed the "rule of thumb" center-to-center distance, you run into problems with how the response combines. Plotted in the vertical axis, these look like lobes, with different angles off of vertical having peaks and dips at a given frequency. The frequency response for a given point in space will get more ragged, since at some frequencies there is constructive interference (which is what you want- all of the sound do show up at the same time)- and destructive interference..

Hence the need to do vertical polar measurments? I guess that would suggest something about how and where the combining would take place, I assume. If so, here's a question then about taking those measurements. First is it best to take vertical measurments of each driver individually, or should they be taken with all the drivers on? It occurs to me that if they were all on, areas of overlap would show increasing spl quite clearly, I would think. Also, how woud they be taken? Are angles really important, or can I simply raise the mic through a series of heights, while roughly adjusting the distance at each height to keep the acoustic distance reasonably close at each height?

If you had the woofer and mid coinciding in space, then there'd be none of this effect at all and you would just have to worry about phase issues induced by the driver's natural behavior plus the effect of the crossover.

I have to wonder how serious phase issues can be. The drivers, are very carefully aligned to the listening position, so in that way, they are coincident in space vertically relative to the listener. Their sound waves should reach the listening position at the same time, or very close to it and the crossover should be able to easily address this, I would think.

3) This rule of thumb is for passive crossovers. If you build them closer in frequency, then they start to interact in ways that aren't described in texts on crossovers, and instead require serious analysis of the entire global circuit. They become increasingly sensitive to component values, and it can be a huge problem trying to maintain a reasonable impedance- with possibly crazy low/high impedances and wildly swinging phase at low impedance.

If you have just one of the two crossovers in a 3-way as active, this is no longer a concern because the two crossovers don't interact electrically anymore, since the amplifier serves as a buffer. Actually, if you had the two sections bi-amped with totally separate passive filters, that would do it too.

My plan is for this to be all passive, unless it simply can't work out well that way.

I bet that some of the gurus here could be a big help with that. You're certainly putting in some equity here to show that you're not just another n00b who wandered in wanting everything handed on a silver platter.

Exactly, however I'm not holding my breath based on past contributions and repsonses to questions. I'm doing all I can do other than to shut up and go elsewhere.

And, regarding the omni question- I'm quite certain that you don't need baffle step compensation. Here's my justification.

Starting with the woofers- there's not really any difference acoustically with how the woofers interact with the room. At their frequencies, they're basically omni anyway and their waves wrap around the cabinet easily and reach whatever wall is nearest and are reinforced by its presence.

Next, we come to the tweeters- in a conventional speaker their wavelength is a small portion of the face of the baffle. Their output thinks it's just radiating into half a room, until it gets to the cabinet edge. It's reinforced by that wall it starts out next to, all the way down (the mids, too) to a frequency dictated by the width of the baffle. Your Mentor doesn't have the mids or tweets firing off of a front baffle, so they don't get that reinforcement. Instead, their omni all the way from the lowest low to the highest high.

If that made sense, hopefully you'll agree with me that "baffle step" is more about *gain* above a certain frequency, than loss below it. Your design simply fails to have that gain.

Are you sure Jon Marsh said you'd need baffle step for this one? He's a pretty sharp guy, and I think he'd agree with this answer. It'd actually make me a little nervous if he didn't- at that point I'd feel better if we all got together and re-examined the problem definition.

What you say makes sense to me and John K's explanation did also at the time. What Jon M. said is in this thread, https://www.htguide.com/forum/showthread.php?t=25587, hence my confusions and concern. I never got a response to my follow up questions.

[dlneubec]

By the way, I agree with a low and steep crossover between the mid and tweeter, in case that wasn't clear. It's hard to say *what* you'd call the spacing for those, so it's probably best to just minimize high frequency interaction.

Hi Joe,

Do you expect steeper crossovers to help alleviate problems that might occur when pushing the CTC distance toward the maximum?

Are you thinking steeper than 4th order?

[Paul W]

Dan,
From a slightly different angle...

Q 1&2: To visualize what Joe says, you can download ARPE from the FRD Consortium, plug in your drivers, and look at the results around your intended crossover frequencies. ARPE doesn't expect horizontally oriented drivers, but it should get you in the ballpark.

Q3: For series crossovers, I agree with Joe...component interaction will drive you nuts! For parallel, a very narrow passband can still be troublesome. From personal experience, I've been surprised by low impedance and it has been difficult to get narrow filters to sound "right" (sound somewhat like a bell).

Q4: Don't know if it is typical or not, but I start by targeting a response for each driver and XO. Only after I get fairly close to the individual targets do I move on to system integration.

The difference in opinions on baffle step may be only one of perspective. I agree with Jon in the sense that there is clearly a step from the small horizontal baffle down to the floor. On the other hand, the 360 degree radiation will spread wall reflections over a very wide frequency range. I doubt it matters in the end because your end results will include any appropriate compensation. (You may already have evidence of baffle step/diffraction in the different tweeter measurement results between square and round mounting plates.)

700 & 3k were merely starting points based on what I thought the drivers could handle. "Best" may be higher or lower...which gets to the reason to check vertical response. The results you are seeing at 1M, 2M, 0,45, etc are encouraging, but power response is about the total power radiated in all directions (including vertical) not just radiation in the horizontal plane 38" off the floor. CTC distance plays a role here, and I wonder if the diffusers are large enough to prevent LF "leakage" at the low-end of the respective drivers. For example, might the low-end of the dome mid be bypassing the diffuser and be bouncing off the ceiling? If there are any errant rays (visualize 360 degree cones at different vertical angles) they will influence power response and may impact your choice of XO frequencies.

Since there is an unusually high percentage of indirect/reflected sound (think opposite of waveguide) the room will have a much greater impact on the sound of this speaker. That is (my) reason for suggesting final tweaking be done in consideration of RTA measurements around the room, with higher weighting for the measurements near your listening position.
Paul

[dlneubec]

Ok, vertical measurements.....

I don't know if I did these correctly, but here's what I did. I started with the woofer and did 1m distance, 14" up from the floor with the speaker sitting on the floor. Then I went up in 4" incrments to 38" , while keeping 1m distance from the bottom woofer.

I then repeated this for the midrange, except I went from 18" up to 50". Then the tweeter at 26" up to 50". The colors for the graphs were the same for each graph as follows:

14"=black
18"=dk blue
22"=dk green
26"=pink
30"=dk red
34"=orange
38"=lt blue
42"=lt green
46"=red
50=gray

The plots below are the woofers first, followed by the mid then tweeter. I don't know what to look for here, but they look to me to be pretty consistent throughout the range. The woofer is the most different, with a broadf 5db peak that changes from nearly 1k at 14" to 500hz at 38".

Let me know what these tell you and if anyone has any suggestions as to further measurements. Also, how to interpret these when it comes to making crossover design choices such as choosing crossover types and slopes or target crossover points, I'd sure like to hear them!

[Paul W]

As if I had done several omni speakers... :roll:

It looks to me as if the diffusers are not controlling directivity below 1k. No problem for the tweeter, but there may be a lot of mid-dome signal below 1k spraying the ceiling. To be sure, try holding mic/driver distance constant and sample FR while moving the mic in an arc from 38" high to directly above the speaker (how close is the ceiling?).

If, in fact, there isn't much diffuser directivity below 1k you could either increase the diameter of the diffusers or raise the mid crossover frequency to make the mid more consistent. The woofers don't seem to negotiate the cabinet so easily above 1k so maybe that helps. Tweeter seems fairly well controlled above 3k. Target the best overall woofer response around the listening position.

Joe's suggestion of steep slopes may be the least problematic. Anyhow, that's my 2 cents. Like you, I wish more people would take a shot at this...you've undertaken a really interesting challenge!

[Dennis H]

For measurements, I'd measure each driver on its own axis, i.e. lower the mic even with each driver. Then set it up in the sim to account for that. It's a bit harder than measuring them all from the same point but I think it will be more accurate for your complex speaker.

[dlneubec]

Paul,

Thanks for the anwers above. I'll download ARPE and see what I can learn. I think that zaph suggested that the ND20 be crossed above 3.5k, IIRC, due to lowr end distortion issues. I also recall something about a distortion peak in the RS52 at around 4.2k. I'll have to review those again. Perhaps these issues will drive my decision of where to start initially.

I've no experience with series crossovers and don't anticipate trying one here.

I'll take some more vertcial measurements on the mid to get a better handle on what effect the diffuser is having on directivity towards the ceiling.

[dlneubec]

For measurements, I'd measure each driver on its own axis, i.e. lower the mic even with each driver. Then set it up in the sim to account for that. It's a bit harder than measuring them all from the same point but I think it will be more accurate for your complex speaker.

Hi Dennis,

Maybe I will do a few sets of measurments using different techniques. One with the mic at the actual driver height as you suggest, another with the mic fixed at 38" high and 1m distance with the speaker on the floor and another with the speaker lifted up so that the woofers are measured at the 38" mid level. Pick one set to start modeling with and once I get a decent looking model, plug it into the other sets of measurements to see how the results change.

If they are fairly consistent from set to set, then I might stay with the one I started with. However if the results are very different, then I have some more serious decisions to make.

[joecarrow]

Joe,
Hence the need to do vertical polar measurments? I guess that would suggest something about how and where the combining would take place, I assume. If so, here's a question then about taking those measurements. First is it best to take vertical measurments of each driver individually, or should they be taken with all the drivers on?

All drivers on, and unfortunately- measurements with the crossover in place to see how they combine. This is where a digital active crossover would be a huge help. There might be some tools available to you to simulate the summation of the driver response, but the waveguides and diffusers make the whole thing kind of different from normal assumptions.

It occurs to me that if they were all on, areas of overlap would show increasing spl quite clearly, I would think. Also, how woud they be taken? Are angles really important, or can I simply raise the mic through a series of heights, while roughly adjusting the distance at each height to keep the acoustic distance reasonably close at each height?

I think you've got the right idea here.

I have to wonder how serious phase issues can be. The drivers, are very carefully aligned to the listening position, so in that way, they are coincident in space vertically relative to the listener. Their sound waves should reach the listening position at the same time, or very close to it and the crossover should be able to easily address this, I would think.

If the drivers are all aligned very carefully, then the phase induced by the crossover is just as important- maybe even more so. If the driver phase is matched by the crossover, then all of the sound shows up at the same time and you're in great shape. If the crossover and driver behavior is non-ideal, and two of the drivers aren't in phase throughout the crossover region, then it will mess with listening axis response.

My plan is for this to be all passive, unless it simply can't work out well that way.

I think it probably can be worked out, but it will be a challenge. Part of the challenge is that with this physical configuration we can't be 100% sure of what acoustical output we want from each driver- especially from mid to tweet. This is where it would be really helpful to be able to prototype it; either with a DCX or out of your sound card.

Regarding the crossover slope and CTC spacing for tweeter to mid- I would really think that 4th order would be sufficient. There's a really wide overlap with second order. With 4th order, it's that much better.

Also, regarding the peak with the woofer... can you check to see if there's some energy storage there? I'm curious if there's a bit of cavity resonance there. That could be addressed a couple of ways if it turns out that it's there. I'm with Paul in suggesting that the diffuser's diameter could be increased.

If there's a peak in distortion on the RS52, I'm guessing it correlates to the high frequency metal dome breakup. Everywhere I've read about this driver, it's been said that it's an outstanding driver that needs that one thing accounted for.

Sorry I couldn't reply a bit sooner- this job hunt had a few good leads for me lately. That, and my girlfriend wanted to hang out.

[dlneubec]

All drivers on, and unfortunately- measurements with the crossover in place to see how they combine. This is where a digital active crossover would be a huge help. There might be some tools available to you to simulate the summation of the driver response, but the waveguides and diffusers make the whole thing kind of different from normal assumptions.

I don't have a digital crossover and I don't believe the digital filter emulation capability in SoundEasy will work with the USB Audio device I'm using. I think it takes a card with more input or outputs, IIRC . Someday I'll update my USB soundcard to a unit that can do digital filter emulation right out of SE. That sure would be a handy thing to have, especially in this case!

Also, regarding the peak with the woofer... can you check to see if there's some energy storage there? I'm curious if there's a bit of cavity resonance there. That could be addressed a couple of ways if it turns out that it's there. I'm with Paul in suggesting that the diffuser's diameter could be increased.

How do you check for stored energy? A CSD waterfall plot? I don't know how to do that in SE, though I know it is capable of it.

I was worried when cutting down that diffuser that it might end up too small, but what's done is done. I might be able to create a ring of wood by cutting a circle and then cutting the center out with a circle jig and then fit it over the existing one, and glue them together, sand down, etc. However, that is a lot of work and making it look decent will be difficult. Getting a new one made is no simple thing either, so for now, I will see what results I can get with the one I have and address this issue later when I'm more sure it really is an issue.

If there's a peak in distortion on the RS52, I'm guessing it correlates to the high frequency metal dome breakup. Everywhere I've read about this driver, it's been said that it's an outstanding driver that needs that one thing accounted for.

Sorry I couldn't reply a bit sooner- this job hunt had a few good leads for me lately. That, and my girlfriend wanted to hang out.

No problem. I've been married for more than half my life now, so I know about significant others!

Here is what i know about the RS52 and ND20fb (mostly coming from Zaph). Zaph suggests that the RS52 should be good down to 500hz LR4, 700 LR2. At 500 hz with 30 watts it outputs a max of 106db before reaching xmax and at 70w and 108db with a 700hz crossover. The xmax was reached at 520hz and 480hz respectively, IIRC. There is a peak in the F3 harmonic distortion at about 4.3-4.4khz and the cone breakup peak is around 13khz. It is a very low distortion driver, as I understand it. Also, Dennis murphy has completed a design that has yet to be published (or so I'm told) with a crossover of the RS52 to the RS225 at 600hz, so that would certainly seem to be a safe choice, I would think.

Zaph says the ND20 has good distortion figures above 3.5khz, so I was wanting to cross at 3.5khz or above, probably 4th order.

[dlneubec]

While I'm contemplating what the best scenario for doing the measurements is, I decided to pick on of the sets I already had and try to model up a crossover. I used the measurement set which is 2m distance, 38" mic height, same position for the mid and tweeter. The woofers I used measurements from 2m, mic at 38", but the speaker was raised off the floor so that the woofers were centered at the mic height of 38". All measurements were with 4ms gated windows, 1/12 octave smoothing applied. The woofers were measured as a pair, but the FR was imported 6db down from the tweeter and mid so that when the two drivers were added in SoundEasy, they would sum correctly.

I was shooting for 600hz and 3600hz crossover points, for a 2-1/2 octave spread. These points seem to make sense to best avoid the distortion areas of the mid and tweeter. All slopes were intended to be 4th order acoustic and all are 4th order electrical, except the high pass on the mid, which is LR2 electrical. Below is the results. It took a LOT of iterations and tweaking to get this to work. For those who are not familiar with SoundEasy, it adds both a inductor and resistor component at each inductor location, but the resistance is only to allow you to include the DCR value of the inductor in the model. There is actually 18 components in the crossover despite the list showing 28 at the right.

I had to do some messing around with resistors at either the amp side, driver side or both to get a reasonably flat response with good phase tracking at the crossover points. I'm not sure if this is an acceptable method or not. Also, I added a cap after the driver side resistor on the tweeter to flatten the rising top end of the tweeter. I don't know if that is advisable or not either.

Also attached is the modeled FR of the crossover, the reverse null, the impedance/impedance phase and the phase plot of all three drivers and crossover. For some reason, SE doesn't want to plot sum of the two woofers all the way to the bottom of the graph, which makes it look a little confusing. The one you see for the woofer is the ouput of a single driver. I also have a group delay graph I can post, if anyone wants to see that.

I'm looking for feedback here. Let me know what you think!

[joecarrow]

How do you check for stored energy? A CSD waterfall plot? I don't know how to do that in SE, though I know it is capable of it.

Yes, that's about right.

I was worried when cutting down that diffuser that it might end up too small, but what's done is done. I might be able to create a ring of wood by cutting a circle and then cutting the center out with a circle jig and then fit it over the existing one, and glue them together, sand down, etc. However, that is a lot of work and making it look decent will be difficult. Getting a new one made is no simple thing either, so for now, I will see what results I can get with the one I have and address this issue later when I'm more sure it really is an issue.

Hey, were we talking about the thing between the woofers, or the diffuser up by the mid? I thought this was a woofer issue, and whatever you had between the woofers didn't look round. I'm not sure exactly what's causing the problem, but it seems to exist between 500hz and 1khz- that goes just a little too low to let the mid to woof crossover make sure it's taken care of.

No problem. I've been married for more than half my life now, so I know about significant others!

Here is what i know about the RS52 and ND20fb (mostly coming from Zaph). Zaph suggests that the RS52 should be good down to 500hz LR4, 700 LR2. At 500 hz with 30 watts it outputs a max of 106db before reaching xmax and at 70w and 108db with a 700hz crossover. The xmax was reached at 520hz and 480hz respectively, IIRC. There is a peak in the F3 harmonic distortion at about 4.3-4.4khz and the cone breakup peak is around 13khz. It is a very low distortion driver, as I understand it. Also, Dennis murphy has completed a design that has yet to be published (or so I'm told) with a crossover of the RS52 to the RS225 at 600hz, so that would certainly seem to be a safe choice, I would think.

Zaph says the ND20 has good distortion figures above 3.5khz, so I was wanting to cross at 3.5khz or above, probably 4th order.

It sounds like we're on the same page with our understanding of the drivers' capabilities. This is getting closer and closer. I'm sure it'll all pay off in the end.

[augerpro]

Wow Dan this is looking good! I've been following this thread and admire your willingness to go off the beaten path!

I only have a couple comments since most of this project is WAY out of my league. If you could I would try measuring outside with the speaker up off the ground. That 4ms is really short. There are probably only 3 true data points under 1000 Hz and it's probably causing that wave in the FR there. If it was me I would want to make sure the level and contour of the FR and phase was correct as possible.

As for the resistor placement I beleive there is some info on this in the LDC. My results on a 3 way XO I did did not mirror what was in the LDC though. I have no idea what guidelines there are about this. I just did whatever "looked" good.

I think you'll like that ND20. I finally listened to my MTMs I made and the bells and chimes have a nice weight. Pretty smooth too. I was really surprised! Roll it off hard though, at higher volumes the distortion becomes fairly noticeable. A very good tweeter for the $5 though.

[dlneubec]

Hey, were we talking about the thing between the woofers, or the diffuser up by the mid? I thought this was a woofer issue, and whatever you had between the woofers didn't look round. I'm not sure exactly what's causing the problem, but it seems to exist between 500hz and 1khz- that goes just a little too low to let the mid to woof crossover make sure it's taken care of.

It sounds like we're on the same page with our understanding of the drivers' capabilities. This is getting closer and closer. I'm sure it'll all pay off in the end.

I've done a few minor tweaks and removed that last cap on the tweeter net and ordered some parts (since I did not have enough in my parts bin) so I can mock this thing up and see what it sounds like. I might get to it this weekend if the parts show up tomorrow. I usually get stuff from PE the next day, even though its shipped ground!

I thought you guys were talking about the mid diffuser. ops: So you are talking about the way the woofer response looks betwen 500 and 1k, correct? The woofer diffuser is easy to redo at most any size and shape. What were you thinking? I did recently try the round diffuser that the tweeter uses for just one woofer and the rectangular baffles response looked a lot better. However, a different shape/size could certainly be tried. I do think that a diffuser is needed to avoid the effect of each cone cavity has on the other in terms of resonance.

[dlneubec]

Wow Dan this is looking good! I've been following this thread and admire your willingness to go off the beaten path!

I only have a couple comments since most of this project is WAY out of my league. If you could I would try measuring outside with the speaker up off the ground. That 4ms is really short. There are probably only 3 true data points under 1000 Hz and it's probably causing that wave in the FR there. If it was me I would want to make sure the level and contour of the FR and phase was correct as possible.

As for the resistor placement I beleive there is some info on this in the LDC. My results on a 3 way XO I did did not mirror what was in the LDC though. I have no idea what guidelines there are about this. I just did whatever "looked" good.

I think you'll like that ND20. I finally listened to my MTMs I made and the bells and chimes have a nice weight. Pretty smooth too. I was really surprised! Roll it off hard though, at higher volumes the distortion becomes fairly noticeable. A very good tweeter for the $5 though.

Hey Brandon,

I may try outside, but it certainly won't be until the weather changed around here! BTW, I plugged in mesurements I took on the woofers from 2m with the mic up at 38" and the woofers at normal height, and also a set of measurements from 1m into this crossover. The phase wasn't as perfect as it looked on the original, but it only took a few minor component adjustments to get them to look good. So I'm a little less worried about the phase matching, etc. with the woofers. Measurements of a mocked up crossover might tell me something as well.

I tried to go up to at least 3.6k for the ND20 with LR4 to try and address the concern about dstortion at higher levels. I don't know if that is high enough or not. On the other hand, as you go higher, you bring up that F3 distortion peak of the RS52 at 4.3k or so. BTW, not only does it look like a nice tweeter for $5, it is one of the only ones that would work in this design. The one Seas neo I tested might also have worked and I think it would fit into the bottom of the diffuser. It has lower distortion but the FR was not nearly as good looking as the ND20. The Aura neo has better distortion, but it FR did not look good in my application.

[joecarrow]

I would think that just bringing that woofer diffuser out to the edges of the baffle would do it. I don't think that the wavelengths involved would care all that much about the actual cavity of the woofer; rather, I think it would be the overall spacing of the top and bottom enclosures. If you separate them with a diffuser, then each woofer basically sees a path to the outside that's too narrow to really support a resonance within the bandwidth. That's just a guesstimate; the other part of my thought model for this would require energy storage and a more-or less fixed frequency where it is worst.

Your variable frequency peak also makes me think of a diffraction or path length difference effect. That also may be affected by a larger diffuser- though I'm not sure if it would be for better or worse.

[dlneubec]

I'll definitely be doing some more testing on woofer diffuser options and/or using none at all. I'll probably try a larger rectangle (out to the baffle edges) and perhaps a larger circular one.

I can tell you that measurements with and without the smaller rectangular diffuser are not that different, at least in the frequency range the woofers are used in, but I can hear a difference, with a definite cavity type resonance sound when the diffuser is removed. What I hear is probably higher up in the FR, where you indeed do see significant differences in the measured FR. However, that may not be significant with a finished crossover.

[dlneubec]

I decided to do my measurements at 2 meters, with the mic at 38" and kept at that height. All measurements are with a 4ms gated window with no changes in mic or speaker position between measurements. Adjustments for the difference in driver to mic distance are then done in SoundEasy by setting up the driver positions relative to each other in the "front panel layout", which is intended to account for any differences in acoustic offset from the driver to the mic position.

When I took the measurements, I tested several diffusers between the woofers. No diffuser, the current rectangular diffuser and a 10" round diffuser were tested and none seem to have any effect on the FR below 1khz or so, at least with the short gated window I'm using. Above 1khz, in the natural roll off area of the driver, there is a difference. I don't think the diffuser makes a difference for the crossover design, so I will do more tests after the crossover design is completed to see if it can have any positive effects on the sound I hear, or on the measurements at that time. That will include nearfield tests (at the front of the speaker/baffle edge) to determine if there is any difference between the diffusers.

I was concerned about the rising FR in the midrange and peak around 3k with the first crossover posted above and also with the peak in the tweeter output at around 9-11khz. The crossover attached is an attempt to address these issues. I simplified the woofer net by going to a 3rd order electrical for a 4th order acoustic. I also was able to mess around with resistor placement and values, amongst other adjustments in the mid net and get the slope reversed, so that there is a little bit of a BBC dip now in the 3k area. An LCR trap was added to tame the peak of the tweeter a bit at 9-11khz.

Included is the xover, FR plot, reverse null plot, impedance/impedance phase plots and a group delay plot.

The results looks pretty good to me, though I see that there are some funky looking things in the group delay plot that were not present before. I don't know how big a concern that should be.

Anyone have any thoughts?

I decided to mock this one up and see what it sounded like. I will post that next.

[dlneubec]

I mocked up the crossover above and hooked things up. The drivers are all new, except the miniscule time doing tests, so there needs to be a lot of break in, but all in all I think it sounds quite good.

I'll wait for further break in and tweaking before posting a more detailed review, but I will mention a few things I noticed. The speaker is amazingly consistent across the room. You can walk across the room from +45º to -45º and the sound is extremely consistent. The speaker is harder to localize and even bigger sounding than my original omni. The bass is very good for a sealed f3 of 50hz, but it sure doesn't have the deep end or impact that the 12" powered subs in the bottoms of my first omni. However, with one of the old omni's hooked up, with one 12" sub playing on one side and the Mentor 2 on the other, the balance is pretty good. The single 12" fills in the bottom end. If I turn off the top end of the old omni so jus the 12" sub plays with the Mentor 2, I can simulate what these will sound like with a sub kicking in around 40-60hz.

I took just a couple on axis measurements of the speaker. These were taken at 2m, 38" mic height, 5ms gated window, 1/12 octave smoothing. The response below 500hz or so is not going to be very accurate, so you can probably ignore that. Attached is the measured FR of each driver along with the overall response. Also attached is the impedance/impedance phase plot.

The one thing that has me confused is the results of the phase matching. the speaker seems to match phase at the upper crossover, now around 4khz, but the low end crossover, a little under 600hz does not seem to show good phase matching. The model shows good phase matching, so either I have some raw measurement problems, or I'm not setting things up right in SE to account for the acoustic delays due to driver offsets. Anyone have a suggestion here?

I'd also like to see that peak at 850hz or so controlled a bit. The impedance looks good at about 6 ohms, with a low of 5 ohms at 8khz.

[augerpro]

Hi Dan. What's with the resistor in the bandpass? Try using a T-bandpass layout for kicks. Basically C7 and L11 are in a row then L8 and C13 after that. JonM mentioned it once and it really helped a low impedence/high sensitivity bandpass that was causing me problems. In my particular case it also lowered the Q of the HP section of the bandpass, so that hump at 850Hz may drop a bit too.

Also what method did you use to measure the two woofers? I used John K.'s method for doing my MTMs and the measured results of the actual speaker vs. the model are kind of strange. Sort of like yours, lower woofer level than expected and slightly different phase. Luckily the addition of 1.5 Ohm resistor to the tweeter solved this. Very flat response and they just sound great! But I'm still trying to sort that method out...

[dlneubec]

Hi Dan. What's with the resistor in the bandpass? Try using a T-bandpass layout for kicks. Basically C7 and L11 are in a row then L8 and C13 after that. JonM mentioned it once and it really helped a low impedence/high sensitivity bandpass that was causing me problems. In my particular case it also lowered the Q of the HP section of the bandpass, so that hump at 850Hz may drop a bit too.

Hi Brandon,

The bandpass section was originally made up of an LR2 electrical for a 4th order slope on the high pass, combined witha LR4 electrical/4th order slope on the low pass section. Then I added the resistors and second cap for additional control that seemed necessary. I'm not sure if this is an acceptable approach or not.

I'm not sure I'm following your suggestion on a T-shape. Do you mean move L11 so it is adjacent to C7 and then place the two legs of L8 and C13 adjacent to each other in the center?

Also what method did you use to measure the two woofers? I used John K.'s method for doing my MTMs and the measured results of the actual speaker vs. the model are kind of strange. Sort of like yours, lower woofer level than expected and slightly different phase. Luckily the addition of 1.5 Ohm resistor to the tweeter solved this. Very flat response and they just sound great! But I'm still trying to sort that method out...

I set the speaker on the floor. Set the mic up at 2 meters, 38" high. All measurements were taken at that point. Woofer impedance was measured with one woofer. FR measurements were taken with both woofer on, but set 6db below the mid and tweeter when imported, so that when two drivers were added in SE, they would sum properly. This is the method John K. suggested when I did the first omni MTM project. Then I set up the "Front Panel Layout" for the speaker in SE, where the vertical heights and offsets between the drivers was input and acoustic distances is calculated byt eh software. Again, this was as John had recommended before. It is my understanding that this accounts for the different acoustic distances between the drivers and the mic. I wonder if there is a problem with my setup here however. One thing that occurs to me is that I entered each woofer individually in the front panel and each has a different offset. Yet, each uses the same measured response. It would seem to make more sense in this case, to measure each woofers output separately and then the front panel adjustment would make more sense. What do you think?

Also, I set the horizontal offset in the drivers positions. For example the mid and tweeter are at 0", the top woofer is at -3.9cm (closer than the mid/tweet) and the bottom woofer is at -5.8mm. The bottom woofer is at a height of 33.2cm, the top woofer, 45.9cm. the mid at 93.6cm, the tweeter at 99.1cm and the mic at 96.5cm. I used the location of the baffle each speaker is mounted on to enter these heights. On other approach might be to use the approximate voice coil height of each driver, rather than it's baffle. The heights were input in the Y column and the offsets in the Z column. The Acoustic Offset column was left at 0 for all drivers. SE does not recognize up or downfiring drivers, so I don't see any way to ue the acoustic offset column, unless I use acoustic offset instead of Z location to input the relative offsets of the drivers. Any thoughts?

My thinking is that first, the woofers either need to be measured indivually and locations input individually, or measured together, as I did, but entered position-wise at a common point, perhaps at a point centered between them vertically and horizontally. In this case, that would mean enter them both at an average height of 39.6cm and average offset of -4.6cm. I think I'll try this and see what it does to the phase of the woofers in my model.

Another possibility would be to load the horizontal offset between the drivers as an acoustic center distance change, rather than a Z distance change.

[augerpro]

Hi Brandon,

The bandpass section was originally made up of an LR2 electrical for a 4th order slope on the high pass, combined witha LR4 electrical/4th order slope on the low pass section. Then I added the resistors and second cap for additional control that seemed necessary. I'm not sure if this is an acceptable approach or not.

I'm not sure I'm following your suggestion on a T-shape. Do you mean move L11 so it is adjacent to C7 and then place the two legs of L8 and C13 adjacent to each other in the center?

Yep just like that. The usual layout is called C-bandpass because the filters are Cascaded. The T-bandpass was popularized by Bullock I think. Here is mine:



Try removing R17 from the bandpass and try this layout. I don't want any resisitors in series with a mid or woofer just because of the wattage issue. This layout worked well for my problematic mid with low impedence. You'll have to rework the XO to get the desired FR again. I actually needed smaller components after that, don't know if that's normal but i suppose it has to do with the higher impedence. So it's economic layout also

I set the speaker on the floor. Set the mic up at 2 meters, 38" high. All measurements were taken at that point. Woofer impedance was measured with one woofer. FR measurements were taken with both woofer on, but set 6db below the mid and tweeter when imported, so that when two drivers were added in SE, they would sum properly. This is the method John K. suggested when I did the first omni MTM project.

That sounds like what I did. Under the SPL Measurement you entered -6 like this:



? Then saved the driver file? That's what I did. I noticed the HBT doesn't seem to save to the file unless you have the Driver Editor open to the HBT tab and have the responses graphed. Maybe this was why my responses didn't match? Here is the modeled vs the actual, purple is modeled, red is actual and is just a farfield measurement:



It's close but not quite right. I modeled the XO for smooth offaxis response and surprisingly the match is closer around the XO:



Pretty good other then the tilt caused by the woofers too low or the tweeter too high. Again I have to troubleshoot this. My center channel XO will be built this weekend so I'll see if that sheds some light on this. Otherwise I'm going to build a simple MTM just to measure and model with until my model matches the actual.

Then I set up the "Front Panel Layout" for the speaker in SE, where the vertical heights and offsets between the drivers was input and acoustic distances is calculated byt eh software. Again, this was as John had recommended before. It is my understanding that this accounts for the different acoustic distances between the drivers and the mic. I wonder if there is a problem with my setup here however. One thing that occurs to me is that I entered each woofer individually in the front panel and each has a different offset. Yet, each uses the same measured response. It would seem to make more sense in this case, to measure each woofers output separately and then the front panel adjustment would make more sense. What do you think?

Also, I set the horizontal offset in the drivers positions. For example the mid and tweeter are at 0", the top woofer is at -3.9cm (closer than the mid/tweet) and the bottom woofer is at -5.8mm. The bottom woofer is at a height of 33.2cm, the top woofer, 45.9cm. the mid at 93.6cm, the tweeter at 99.1cm and the mic at 96.5cm. I used the location of the baffle each speaker is mounted on to enter these heights. On other approach might be to use the approximate voice coil height of each driver, rather than it's baffle. The heights were input in the Y column and the offsets in the Z column. The Acoustic Offset column was left at 0 for all drivers. SE does not recognize up or downfiring drivers, so I don't see any way to ue the acoustic offset column, unless I use acoustic offset instead of Z location to input the relative offsets of the drivers. Any thoughts?

My thinking is that first, the woofers either need to be measured indivually and locations input individually, or measured together, as I did, but entered position-wise at a common point, perhaps at a point centered between them vertically and horizontally. In this case, that would mean enter them both at an average height of 39.6cm and average offset of -4.6cm. I think I'll try this and see what it does to the phase of the woofers in my model.

Another possibility would be to load the horizontal offset between the drivers as an acoustic center distance change, rather than a Z distance change.

I have the same concerns. I asked about this on the usergroup and got one reply that didn't answer anything. I may not have been clear about what I was asking though I guess. My concern was using the polar plots. If the difference in distances is already in the measured phase it seems to me that moving them around the baffle layout (for doing polar plots in my situation) could only introduce error. I don't have a good answer here. I've been working like a madman and haven't much chance to sit down and think this through. Besides all that I think you can tilt the driver in the baffle layout screen, perhaps this can do what you need?

Anyway hope this helps,
Brandon

[augerpro]

Oops. Just noticed the scale is different on my graphs.

[dlneubec]

Hi Brandon,

I will give that topology a try. Thanks for the info!

On the measurement side, another approach has occurred to me for my project that may be much more accurate in terms of phase correction. It occurred to me to take look at the plan and calculate the direct path that sound would have to take from the VC to the mic. Attached is plan of what I came up with in reviewing the CAD drawings.

Since I calculated the actual acoustic distance from the plans, it seems to me that I should set the front panel up so that both X and Y values are the same for all drivers, but the differences between the calculated distances would be entered into the Z value for each driver.

I figured that each driver had to be set a positive distance over the 2m mic position along the z axis to reflect the increased distance or offset. That mean +.98cm for the tweeter, the mid at +1.53, the top woofer at +4.6cm and the bottom woofer at +5.8cm. This should reflect the fact each driver is further away from the mic. The front panel layout I'm thinking of is attached. Does this approach make more sense?

[dlneubec]

Brandon asked about this project in another thread, so I thought I'd post an update.

I never was able to resolve the problems I was having with my simulated phase match as compared to the measured phase match after the crossover was mocked up. I took to swapping compononets in and out and measuring the changes and finally came up with a crossover that is pretty good. The phase match is good and the crossover points are at about 690 and 4000hz, pretty close to my goal.

Also, I took the prototype of the waveguide omni and played it in mono at the Indian DIY (back when the crossover was not as good and points were pushing the low 500's and about 3900hz). It recieved a good amount of praise, though it was a bit all over the map. Some thought it was amazing and others thought they heard some driver integration problems.

Unfortunately, the waveguide omni is just not as good as the original Mentor omni (to my ear). I think I may need to change the mid dome out and use a high efficiency pro mid, 6.5"-8" in size, with no waveguide. Also, I could stand to get better at 3way design, which might help as well. I have decided to put it aside for the time being to work on a more direct update of the Mentor omni project. I will probably start a thread on it soon. Here is a pic of what the waveguide omni looks like prior to the temporary shelving of the project. Also included are some graphs of the measured response of the last crossover build of the waveguide omni, 7ms gated window.

I hope to start making dust on the new project this weekend.

[joecarrow]

Thanks for the update- I saw the other thread and I wanted to ask how it's doing.

I'm sorry to hear it didn't pan out as well as we all hoped. It sure looks cool, though.

[dlneubec]

Hi Joe,

I've not given up on it. I'm just setting it aside until I have more experience under my belt to understand what I need to do to make it first class. :E

However, I'm excited about my new project. It's a partial omni, like the Mentors (my first omni project). It will have a powered sub (RSS265HF) in a 20:1 taper transmission line cabinet, F3 about 29hz, F10 about 19hz or so, in a cabinet about 31 liters. With room gain, that should be some serious low end output. I plan to cross it over up higher, around 180hz or so, to an MTM using the 2-RS150-4's in series, one up firing, one down and a forward firing RS28.

The cabinet size will be much smaller than the original and is intended to improve on a few things. My wife likes the look a lot more and the RS265 should make the bass both clean and strong.

I plan to start a new thread on this one soon. I'll probably start making MDF dust this weekend. I see a similar crossover to the Mentors, which used the RS180/RS28 combo, but crossing up at 1800hz. I'm hoping the 180hz and 1800hz crossovers, with steep slopes on the MTM will allow for some pretty good SPL's before any strain shows up.

Here is a teaser. More later!

[Dennis H]

Hey Dan,

Sorry the second project didn't work out. Maybe it's just a sign of how well you did the first one. I guess these projects are unconventional enough that you just have to cut some wood and see what happens.

[dlneubec]

Hi Dennis,

Ain't that the truth! You really have to put yourself out there and build something with these experimental concepts before you know what you have. But the reward can be excellent too. :T

I'm not disenchanted though. See my post above about the upcoming project. I'm excited about the potential it has and I'm not treading on too much new ground, since I pulled this basic concept off once already. It should be fun and maybe it will be better!

[joecarrow]

Very interesting, Dan! I have to say, I'm a bit concerned about the transmission line. In my experience with modeling them in Martin King's worksheets, you don't often get a size advantage from going TL. 20:1 also seems like a fairly extreme taper. Do you have everything you need to run a simulation? Just running that box size and geometry could work fine sealed, so I don't think you'll run into ill effects... just not sure if there will be much TL action to make it worth any extra effort.

I'll be sure to chime in on the new thread.