Zilch's AK Design Collaborative - Econowave Speaker

[Andyman]

Here's a quick shot of the stained side on one of my cabs. hopefully, I'll be able to lay some veneer on the motorboard in a day or so..

FWIW, that's birch plywood stained with a non grain raising Light Reddish Brown Mahogany, then with American Walnut, oiled with boiled linseed oil and top coated with 5 coats of gloss Deft brushed on and sanded smooth.

Kegger loves it

 

[mrTT]

Wow! That looks so beautiful!

 

[Wayne Parham]

Whether you provide padding via fixed resistors or variable L-Pads, you still have to dissipate the same amount of power, given the same drive level and the same tweeter load. The padding reduces current to the tweeter, and it also shunts some away. It sends about 10% of the input power to the tweeter. It isn't that 90% of the power is dissipated by the resistors, because they don't just divvy up the power, they also impede its flow. But it's still pretty significant. If you have a single 10 watt resistor in positions R1 and R2 and a 10 watt L-Pad, I'd only really be comfortable with about 50 watts input power.

If you want to know how much power is dissipated by each resistor at various drive levels, evaluate the circuit in Spice. Just take the models you have and enter various input voltage levels and look at the voltage across each resistor. Then you can easily calculate the power dissipated using Ohm's law. You'll also see it as a chart that shows frequency on one axis and voltage on another, so you can see what frequency range the power rating is most critical.

You can always do what you're doing too, that's cool, but I think that 100dB/M to 110dB/M is not much power. That's probably less than 10 watts. I wouldn't expect a problem at that level even with only one resistor. I was talking about what happens when you sent the speakers ten times that much power.

Now, maybe ten watts is all you'll ever need, and that's fine. Lots of guys run SET amps and that's all they ever need. But since you are probably using drivers that handle waaay more power than that, the speakers can be used for heavier duty. That's when you'll need to consider the thermal specs of your resistors. Just something to think about, that's all.

As for the R1/R2/C1 chart, my reason for posting it wasn't necessarily to suggest the L-Pads be removed. I don't use them, but that doesn't necessarily mean I think you shouldn't. You've said you want the variable control and that's fine. I just saw some of the problems you were running into, trying to use 16Ω compression drivers, running out of L-Pad scale, etc. So I threw in my 2¢ worth.

My main reason for posting the chart is some had said the two versions for your schematic were just out of range for them. One had too much attentuation, the other, too little. One needed the L-Pad set for maximum, the other for minimum. I suggested a few more values be considered than what you had. Then the centerline value could be selected from a broader range. If you use an L-Pad, you have more base levels to start from.

I wasn't trying to "talk you out" of L-Pads, but wanted to point out the reasons for the problems some of you were running into. Varying driver impedance means you have to vary R1/R2/C1 values. The L-Pad is part of the circuit, and changes some of its parameters. So you have to consider that. It also can be a problem when using a high efficiency woofer, because you need all the attenuation available to be part of the R1/R2/C1 response shaping network.

Remember that the tweeter circuit doesn't just provide padding but also response shaping. We need at least 8-10dB fixed attenuation just to get the response shape right. So if you are using a woofer that's only 10dB lower than the tweeter, this becomes an issue to consider.

Think about it. The augmentation we need is 6dB/octave, starting around 5kHz. There's two octaves to 20kHz, so the augmentation curve spans 12dB. If you have a tweeter that's 107dB and a woofer that's 97dB, you only have 10dB difference. You need every bit of that just to get the top-octave compensation you need. That's the main reason I choose fixed resistors. They tie the response shape and padding together.

How would you configure the tweeter circuit when using a 97dB woofer and a 107dB tweeter? Would you use R1/R2/C1 values that provided 6dB padding and top-octave compensation, so you could follow that with an L-Pad that provided 8-10dB span? That setup will allow you to set padding right, 6dB in the R1/R2/C1 shaping network and 4dB in the L-Pad. But it cannot provide the necessary EQ because attenuation will be fully removed by 10kHz, leaving the whole top-octave under-represented.

That is the problem, in my opinion. If you need 14dB attenuation or more, the L-pad arrangement is more palletable for me because you can still get all the response shaping you need prior to the L-Pad. But if you are using high-efficiency woofers, most around 96dB to 98dB, then using the L-Pad becomes more problematic. There's only 10dB difference, just enough to provide requisite response shaping. And frankly, the most attractive woofers for a loudspeaker like this are those that have voltage sensitivity of about 96-98dB/2.83v/M. So to me, for these kind of woofers, the R1/R2/C1 component values were the only way to set padding and provide top-octave compensation. There just isn't any more room for padding.

It's your project, and I'm not trying to steer you one way of the other. I'm just telling you the things I've run into when working on my speakers, very similar to yours in most respects. Hope it helps.

 

[mrTT]

Spice

Wayne
Thanks for your post. To be honest, it takes such a long time for me to understand a half of what u post. But please help me out with your spice program. This is what I posted earlier.

I went away trying to learn Spice. It's too spicy for me. I can't handle it. On a serious note. What am I suppose to do with it? Say I want to play around with R1 R2 C1 values. Where do I put the numbers? What functions am I suppose to use? Sorry Wayne. I am a bad student. Even your "student version" is too hard for me.

I think my econowave project is done. But this is for my learning and future project.

Thanks

 

[Kegger]

See exactly what your seeing Wayne, and I'm on the same page.

Your info and thoughts certainly are wanted/welcomed by many here.

L-Pads make it easy to get you up an running if you have woofers of 95db sens
or less, higher then that an L-Pad starts to become an issue with adjustment an
the EQ curve as well can get a little tricky, and myself I want the higher sens to.

 

[Wayne Parham]

Spice models

I went away trying to learn Spice. It's too spicy for me. I can't handle it. On a serious note. What am I suppose to do with it? Say I want to play around with R1 R2 C1 values. Where do I put the numbers? What functions am I suppose to use? Sorry Wayne. I am a bad student. Even your "student version" is too hard for me.

The nomenclature in my Spice models is the same as the nomenclature in my schematics. So you can easily identify R1, R2 and C1 values. They're exactly as shown.

Below are some Spice models.

1. The original 2nd/3rd Pi crossover
2. The newer 3rd/3rd Pi crossover
3. A model of an old Eminence MD2001 compression driver on an H290 horn

The early crossover worked very well, but I optimized it later with help using the Smith & Larson ICD. This improved summing in the crossover region, largely because tweeter circuit peaking is reduced. There is also some additional delay on the woofer, although not as much as you might expect moving up an order. The crossover point was changed, not just the slope, and this resulted in less delay shift.

You may notice that the Spice models of the crossovers only use a simple Re/Le model for the driver. This is because the frequency of diaphragm resonance and the first quarter wave modes are below the crossover frequency. However, the fact that some optimizations improved the results may mean that it would have been good to include the resonances in the model. This is a mute point, since the S&L ICD uses a ZMA file of measured impedance for the load.

The last Spice model shown is of a compression driver on a horn, with resonances included to simulate diaphragm resonance and the first couple quarter-wave modes.

====== 2nd/3rd order Pi crossover =============================

Pi Speaker Crossover

! 10dB tweeter attenuation at crossover frequency
! for three octaves; 6dB/octave augmentation three
! octaves above crossover frequency
!
! R1=16 ohms, R2=16 ohms, C1=0.47uF
!
! Model for standard "2 way" loudspeaker crossover
! with 2nd order low pass woofer filter and
! 3rd order high pass tweeter filter
! with 10dB compensation circuit
!
! Input across nodes 0 & 5, positive to 5
! Tweeter output across 1 & 2, positive to 1
! Woofer output across 0 & 6, positive to 6
!
! component (+)node (-)node value

vin 5 0 AC 1.0

! Woofer virtual circuit
R3 6 7 6
L3 7 0 1.0mH

! Tweeter virtual circuit
R4 1 8 6
L4 8 0 0.1mH

! Compensation components
R1 3 1 16
C1 3 1 0.47uF

! Pre-compensation load
R2 3 0 16

! RC damper for woofer
R11 6 15 8
C11 15 0 22uF

! Crossover Network
L2 5 6 1.0mH
C4 6 0 10uF
C2 5 4 8uF
L1 4 0 0.6mH
C3 4 3 22uF

====== 3rd/3rd order Pi crossover =============================

Pi Speaker Crossover

! 10dB tweeter attenuation at crossover frequency
! for three octaves; 6dB/octave augmentation three
! octaves above crossover frequency
!
! R1=16 ohms, R2=16 ohms, C1=0.47uF
!
! Model for standard "2 way" loudspeaker crossover
! with 3rd order low pass woofer filter and
! 3rd order high pass tweeter filter
! with 10dB compensation circuit
!
! Input across nodes 0 & 5, positive to 5
! Tweeter output across 1 & 2, positive to 1
! Woofer output across 0 & 6, positive to 6
!
! component (+)node (-)node value

vin 5 0 AC 1.0

! Woofer virtual circuit
R5 6 7 6
L5 7 0 1.0mH

! Tweeter virtual circuit
R4 1 8 6
L4 8 0 0.1mH

! Compensation components
R1 3 1 16
C1 3 1 0.47uF

! Pre-compensation load
R2 3 0 16

! RC damper for woofer
R11 6 15 8
C11 15 0 20uF

! Crossover Network
L2 5 9 1.5mH
C4 9 0 20uF
L3 9 6 0.5mH
C2 5 4 6.8uF
L1 4 0 1.0mH
C3 4 3 20uF

====== Compression driver on horn =============================

Tweeter reactance

vin 1 0 AC 100.0

! source impedance
R1 1 6 60

! tweeter virtual circuit (Eminence MD2001)
! voice coil reactance
R4 6 10 6.6
L4 10 11 0.1mH
! mechanical reactance (on H290)
C6 11 12 25uF
L6 11 12 2.5mH
R6 11 12 100
C7 12 13 12.5uF
L7 12 13 1.25mH
R7 12 13 100
C8 13 0 6.25uF
L8 13 0 0.625mH
R8 13 0 100
R9 11 0 20
C9 11 0 66uF

 

[Wayne Parham]

See exactly what your seeing Wayne, and I'm on the same page.

Your info and thoughts certainly are wanted/welcomed by many here.

L-Pads make it easy to get you up an running if you have woofers of 95db sens
or less, higher then that an L-Pad starts to become an issue with adjustment an
the EQ curve as well can get a little tricky, and myself I want the higher sens to.

Well, yeah, I agree. An L-Pad is OK as long as your woofer is less than 95dB/2.83v/M or so. But then again, the woofers I think are best used in a system like this have greater sensitivity. Most of the good prosound woofers have voltage sensitivity around 96dB or more, and this makes it important that the R1/R2/C1 network provide the full 10dB tweeter padding, with top-octave augmentation that removes it at the top edge of the band. There's no room for additonal padding, and if you try to put it there, you're robbing your system of needed HF extension.

Honestly, that's the biggest deal, not the power handling. The power handling issue is just that the compression driver and woofer can handle way more power than the crossover if you just use single 10 watt resistors. It doesn't matter if the padding is L-Pad or fixed resistors, you really need more thermal capacity if you want the crossovers to handle as much power as the drivers do. I'm all about keeping things matched.

There are lots of guys that build my speakers and just use a single 10 watt resistor R1 and another for R2. My speakers are very popular with low power SET amp owners. Many of them get a copy of my plans, then write to me and tell me they're running an 8 watt SET amp or something like that, asking if they can use a single Mills resistor for R1 and R2. I always tell them the same thing I have said here. If you will only ever run the speakers on that SET amp, then you can forgo the quad-block of resistors and use a single Mills for R1 and R2 instead. Same thing with the woofer Zobel resistor, which is even more vulnerable than the tweeter circuit resistors are. You can de-rate these parts as long as you don't need the extra power handling ability.

You don't have to match the power handling capacity of the crossover to that of the drivers if you don't want to. If you'll only ever use less power, I suppose it's fine. But then again, additional thermal capacity is always a good thing. We're talking about a handful of 50 cent parts, after all. It's cheap insurance against overload. That's just how I see it.

When you overheat a power resistor, it usually chars the circuit board. They usually get hot enough to melt solder long before they fuse. You'll literally have solder dripping inside the cabinet. Sure, it won't happen at 10 watts, maybe not even 50 watts. But these speakers sound good at very high power levels, just like they do at more moderate levels. That's as impressive as their sound quality - the fact that they'll really crank and still sound nice. You can fill a small theater with them, backyard events, whatever. But you will need to have enough current capacity in the crossover to do it.

 

[Kegger]

Here's a quick shot of the stained side on one of my cabs. hopefully, I'll be able to lay some veneer on the motorboard in a day or so..

FWIW, that's birch plywood stained with a non grain raising Light Reddish Brown Mahogany, then with American Walnut, oiled with boiled linseed oil and top coated with 5 coats of gloss Deft brushed on and sanded smooth.

Kegger loves it

Yep looking very sweet Andy!

 

[djnagle]

Very cool looking Andy. Are you going to tape veneer the front edge of the box or does the motor board cover it? I'd like to stop by sometime to hear your e-waves. Cheers.

 

[RayW]

If Spice is giving you trouble, you might try Micro-cap. You can find a demo version here. It's limited to 50 components and has a few other limitations but I find it easier to drive than Spice. If you feel really inspired you can pay the $4500 and get the full version. I've been using Micro-cap 8.0 for quite a while and I like it a lot.

Ray

 

[Zilch]

Rubber hits the road:

 

[mrTT]

Andy
Please help.
I've already stained with mine with Light Oak stain. But I like yours much better. What do I need to do? I love those red stains. Beautiful! If it is too much trouble, I won't even try.
Thanks in advance.

 

[Wayne Parham]

Good show, Zilch. I hope you understand what I'm saying, and don't see this in the wrong light. I think what you guys are doing here is great, so I hope you don't think I'm second guessing what you've already done.

That compression driver looks something like the PSD2002, by the way. Same characteristics up high.

I would expect both crossovers to work well. Even the one for 95dB woofers is set with components that allow for enough top-octave compensation, but just. Look at the difference at the top end of the band, above about 15kHz. The highly padded circuit (blue) can still provide augmentation but the one with less padding (green) can't. It has already bypassed the padding completely by around 15kHz, so there is no more augmentation possible above that point. This marks the edge of the utility of this kind of circuit. Any less padding and you cannot provide effective top-octave augmentation with this kind of passive circuit.

The problem is when you need less padding than what you've shown with the green line. The difference in the shape of the green and blue lines is only expressed above 15kHz. But now imagine what would happen if you needed even less padding. The rolloff up high would happen sooner, making the overall curve more like the red line. Instead of drooping at 15kHz, it would start falling off around 10kHz. The more padding is removed, the more the curve approaches the red line because there is less and less attenuation that can be removed for top-octave augmentation.

That's why I've been saying if you have a combination that requires less padding than can be provided by the crossover shown in green, you might not want to use an L-Pad. You can, but you'll lose HF extension in the process. You'll need all the augmentation you can get, and so you'll probably want all the padding to be part of the response shaping network.

If a woofer is used that is at the very high end of the dB/v/M scale, perhaps because of low impedance or high efficiency or both, then you may want only 8dB-10dB padding instead of 12dB or more. Or if a 16Ω compression driver is used, it will have less voltage sensitivity. In addition to the issue of having to find the right components to set the transfer function for the right response shape, there is the matter that voltage sensitivity is lower, so padding requirements are lower too. If it drops down below about 10dB, then there is the problem of how to achieve 6dB/octave augmentation through the full top octave.

My conclusion is the closer you get to 10dB attenuation, the less attractive L-Pads become. Less than 10dB, and they are really not a good idea, in my opinion. I only pointed this out because some were talking about using 16Ω compression drivers, needing more HF extension, etc. I certainly wasn't critiquing your existing designs.

By the way, same deal on the power handling stuff. I'm not critiquing, just saying. I can sure understand the attraction of using fewer parts. Keep it simple, that's a good rule. Not picking on anyone, far from it. But someday, take those speakers outdoors and give 'em some juice. Crisp, clear and clean. You'll understand why I like to have the loudspeaker equivalent of four-bolt-mains on the crossovers. They'll play surprisingly loud, and still provide excellent sound quality. Audiophile sound quality at concert level SPL.

 

[Andyman]

Andy
Please help.
I've already stained with mine with Light Oak stain. But I like yours much better. What do I need to do? I love those red stains. Beautiful! If it is too much trouble, I won't even try.
Thanks in advance.

What kind of wood are your cabs and what brand stain did you use on them?

Light oak sounds like a pretty mild stain and you should be able to stain over it with the darker reddish stains no problem, unless there's some sort of topcat already involved.

 

[dnewma04]

You can apply conditioner to birch prior to staining to avoid some blotchiness that sometimes occurs without it. Birch and maple seem especially prone to this.

 

[Russellc]

Good show, Zilch. I hope you understand what I'm saying, and don't see this in the wrong light. I think what you guys are doing here is great, so I hope you don't think I'm second guessing what you've already done.

That compression driver looks something like the PSD2002, by the way. Same characteristics up high.



I would expect both crossovers to work well. Even the one for 95dB woofers is set with components that allow for enough top-octave compensation, but just. Look at the difference at the top end of the band, above about 15kHz. The highly padded circuit (blue) can still provide augmentation but the one with less padding (green) can't. It has already bypassed the padding completely by around 15kHz, so there is no more augmentation possible above that point. This marks the edge of the utility of this kind of circuit. Any less padding and you cannot provide effective top-octave augmentation with this kind of passive circuit.

The problem is when you need less padding than what you've shown with the green line. The difference in the shape of the green and blue lines is only expressed above 15kHz. But now imagine what would happen if you needed even less padding. The rolloff up high would happen sooner, making the overall curve more like the red line. Instead of drooping at 15kHz, it would start falling off around 10kHz. The more padding is removed, the more the curve approaches the red line because there is less and less attenuation that can be removed for top-octave augmentation.

That's why I've been saying if you have a combination that requires less padding than can be provided by the crossover shown in green, you might not want to use an L-Pad. You can, but you'll lose HF extension in the process. You'll need all the augmentation you can get, and so you'll probably want all the padding to be part of the response shaping network.

If a woofer is used that is at the very high end of the dB/v/M scale, perhaps because of low impedance or high efficiency or both, then you may want only 8dB-10dB padding instead of 12dB or more. Or if a 16Ω compression driver is used, it will have less voltage sensitivity. In addition to the issue of having to find the right components to set the transfer function for the right response shape, there is the matter that voltage sensitivity is lower, so padding requirements are lower too. If it drops down below about 10dB, then there is the problem of how to achieve 6dB/octave augmentation through the full top octave.

My conclusion is the closer you get to 10dB attenuation, the less attractive L-Pads become. Less than 10dB, and they are really not a good idea, in my opinion. I only pointed this out because some were talking about using 16Ω compression drivers, needing more HF extension, etc. I certainly wasn't critiquing your existing designs.

By the way, same deal on the power handling stuff. I'm not critiquing, just saying. I can sure understand the attraction of using fewer parts. Keep it simple, that's a good rule. Not picking on anyone, far from it. But someday, take those speakers outdoors and give 'em some juice. Crisp, clear and clean. You'll understand why I like to have the loudspeaker equivalent of four-bolt-mains on the crossovers. They'll play surprisingly loud, and still provide excellent sound quality. Audiophile sound quality at concert level SPL.

Very consise Wayne, thanks for the explaination. Your input is very valuable and with Zilch's graph, I can see what you mean. I have noticed a little of what you are saying come to life when I switched from 97 db 2225H to 93 db 2235H, as well as switching the two cross overs.

Oddly, It seems that the one for <95 db has a little less sparkle, ( with the 93 db 2235H) while it looks to be the flatter of the two...maybe its the slight peak being taller in the > 95 db crossover.

Another quirk I've noticed is when I've closed 2 ports instead of one. It trims down (or seems to, I havent measured) the efficiency of the woofer just slightly, making a little more HF emphasis,( or at least shifting the balance that way, and seems to clear up the mids a little as well) without getting too hot with the > 95 db version. I may be getting to the point of where (despite no measurement equipment darn it) I may be able to successfully try the "chart" method.

Are you saying there would be a little more extension done with out the L-Pad? If so, I may be able to reep a litle benefit there? (refering to the green line)

Thanks for the input Wayne, I think we all know and respect your ability:thmbsp:

PS Whoops, the green line is the > 95 db crossover, I wouldnt be using it anyway, not with 2235h I guess.

Russellc

 

[Andyman]

You can apply conditioner to birch prior to staining to avoid some blotchiness that sometimes occurs without it. Birch and maple seem especially prone to this.

There's all sorts of blotching issues here with birch, maple, pine, and cherry among others.

I spent a lot of time the last few weeks searching online and reading my finishing books and there seems to be as many options as there are finishes; conditioners, washcoats of glue size, shellac, oil, use of gel stains or NGR dye stains, on and on.

I did a couple of tests and looked at the birch ply I had and then looked to other items I've made here, like my display table and aquarium/TV stand and decided to use the NGR dye stains I had, apply the oil to pop the grain, and then topcoat it with Deft for a more protective finish. The run of the mill Minwax stains (pigments) didn't do it for me and the gel stain was much, much too light.

 

[dnewma04]

Yours turned out very nice, Andy. I figured you used one of the many techniques you pointed out.

 

[Zilch]

This marks the edge of the utility of this kind of circuit. Any less padding and you cannot provide effective top-octave augmentation with this kind of passive circuit..

Indeed.

Kegger and I analyzed it previously in Spice, and determined that Jack had utilized its full capability in the high-sensitivity circuit (Green).

http://audiokarma.org/forums/showthread.php?p=2142851#post2142851

There's only 7 dB of padding done by the compensation filter there, as I read it, and, assuming the HF driver/waveguide combination actually produces 107 dB, E'Wave should be good with woofers up to 100 dB sensitivity. As you stated earlier, if this is actually the case, there are very few woofers that won't work with E'Wave.

I just added the full spectrum of curves from your chart there, (at somewhat higher resolution in the center thumbnail,) as well (#2251). It's clear why Jack used 1.5 uF to maximize the performance. :yes:

I previously showed the behavior of the L-Pad back on Page 5:

And here's the measured variability among four D220Ti drivers:

http://audiokarma.org/forums/showthread.php?p=1722475#post1722475

Finally, at least with the two pair I measured in post #1, D220Ti is 3-4 dB "hotter" above 10 kHz than D210Ti:

http://audiokarma.org/forums/showthread.php?t=150939

 

[Kegger]

Hey Zilch is there something you did any different for the red and blue ones?
(2nd graph, post 2259)
As that extreme top end really looks nice not dropping off like I usually see it.
Every other graph I've seen shows that UHF at say 17k on up dropping like a rock.