[Duke]

Hi Scott F.,

Wow, what an in-depth post!  Thank you so much for taking the time and sharing your speaker vision with us.  I suspected that there were some closet speaker designers out there with some excellent ideas.

I have built speakers with and without impedance compensation circuitry, and my experience doesn't entirely parallel yours.  Note that you can't just slap a Zobel into a circuit - the circuit has to be designed as a whole (without cutting corners), then measured, then refined, then measured, then refined... and then listened to with fresh ears.  And if it's not right, find out why and start from there.  Adding a Zobel will indeed suck the life out of the sound if the whole circuit isn't re-designed taking everything into account. 

Now it's pretty easy to build a speaker optimized for a given tube amp with a high output impedance, but unless that speaker's impedance curve is quite smooth the tonal balance will be wrong with a different amplifier.  I found this out the hard way.  Designing a speaker that retains the same tonal balance with a wide variety of amps is not easy.  If a multi-way speaker is going to work well with a wide variety of amplification, I think that some sort of impedance compensation will probably be called for - even it it's just in the component values chosen rather than in an obvious Zobel.

For the record, my 92 dB "tube-friendly" speakers are being used with a zero negative feedback 2a3 SET amp by one of my customers, and he's coming from hotrodded Klipsch LaScallas.  I showed with a very low feedback, 30-watt OTL amp in one room and a 5-watt zero-feedback SET in another at last year's RMAF.  In the latter room we did manage to clip on very loud solo piano passages, but other than that no problems.  So I think impedance smoothing can be overall beneficial if done right.

I have built several speakers with first order crossovers, and here is one hurdle they present that is seldom mentioned:  It's difficult to get a smooth power response and a smooth on-axis response at the same time with a first order crossover.  Personally, I place the higher priority on the power response. 

Here's the problem:  Assume we have a driver that's behaving as a piston.  Above the frequency where the driver's diameter is equal to about 1/3 of a wavelength, the driver's power response begins falling at 6 dB per octave (assuming flat on-axis response).   Now if we only used drivers below the frequency where their diameter was equal to 1/3 wavelength, hmmmm.... I think there might be a window of opportunity here.

One possible issue is this:  With a first-order crossover, the drivers will have output way above the crossover point, so any narrowing of radiation pattern up there will still have an effect on the power response.  Still, I think this approach has promise.

Regarding enclosure design, I have nothing against sealed boxes but the woofers whose midrange performance I like tend to be woofers that work best in vented boxes.  Also, with a vented box it's possible to tailor the shape of the low-end esponse curve to take into account the speaker's acoustic environment.   For example, I incorporate variable port length into my designs, and as a result the bass tuning can be optimized for a very wide variety of acoustic conditions.  Personally, I prefer tuning well on the low side, trading off some extension for a more natural sound with better pitch definition.  Offhand I don't know of any 92 dB efficient woofers that will give you decent bass extension in a sealed box, though there are probably some out there. 

I would rather not use multiple drivers over the same frequency range as it causes the power response to start falling sooner because the drivers increasingly cancel one another as we go up in frequency.  I also dislike the lobing. 

It will be much easier (and much less expensive) to do what you are talking about in a 90 dB speaker rather than in a 92 dB speaker.   But I'll see if I can find 92 dB drivers that would work.

This is project is, in my opinion, a definite possibility if you can live with a vented box.  If not, then we'll have to juggle a harsher set of tradeoffs as far as efficiency, bass extension and box size goes.  Briefly, expect to at least double your box size requirement when going from vented to sealed, while at the same time losing some optimizing flexibility.

Duke

[Duke]

Hello Hmen,

Interesting idea you have - a line array that doesn't need a subwoofer.  I can think of three ways to do it. 

One way would be to equalize the midwoofers (or fullrange drivers).   IDS uses equalized fullrange drivers in their line array speakers:  http://www.ids25.com/

Another way would be to find small woofers that can go deep, and just build a big enough cabinet that they actually do go deep.   

Finally, we could use a built-in powered subwoofer section - perhaps a rear-facing line array of small high-excursion woofers.   

What do you think about these approaches?  Are you comfortable with equalization or a built-in subwoofer section, or would you prefer the simpler approach even if it calls for a bigger box and/or lower efficiency?

Duke

[Scott F.]

Hey Duke,

Great reply  Good to know you are using amps that are extremely sensitive to speaker selection in your designs.

I know exactly where you are coming from. The battle is going to be finding the appropriate speakers that behave well. It's not going to be easy at all. As you well know, even though the specs on a driver look like they might work for a design, when you listen to them, they just don't cut the mustard. It's been a while since I scoured the countryside for drivers for this type of design but about 4 or 5 years ago, there weren't many to be found. In turn I gave up and started using vintage and true high efficiency drivers. If prototyping drivers and minimum speaker orders weren't so expensive, that would be the way to do it. Then you could custom design the driver so it would behave (mostly) the way you wanted it to.

When it comes to off axis and power response, like you, I'd sacrifice off axis response. The way we listen really is a singular experience. Sure we have people over and listen but then the music often becomes a background rather than a focus. Most of our hard core listening time is done alone. Just us and the stereo. A narrower sweet spot then becomes far less critical IMO. Oh, and the ported enclosure works just fine. Like you, I like the ability to 'tune' a port to a room. I usually use a bit of foam in the port to dial them in for the room.

Hey (not to take us off topic but), I see you handle JoLida. Have you listened to Mike's factory mods yet? In particular, I've got the modified 102b here and it is simply marvelous for the price. He only adds another $100 for upgraded coupling caps and resistors. At $750, this thing is an absolute steal.

Back on topic. It will be interesting to see if you can find suitable, off the shelf drivers for a project like this. Most of the manufacturers and drivers I've seen really won't fit the bill for one reason or another. Shame too, there could be a nice sized market just for that type of speaker. Especially if by using different drivers you could voice the different offerings to be front row or mid hall. Then you would have most everyones personal preferences covered (and sell more speakers  )

[Duke]

An "aside" comment before I go back to responding to your posts...

I got an e-mail from someone who was afraid that my starting this thread indicated I'm discontinuing my Jazz Module loudspeakers.  Nope, they're very much still the headliner in my line-up, and I'll probably bring a pair to the Rocky Mountain Audio Fest next month.  I'm presently working on a larger, somewhat more ambitious project that hopefully will be ready in time for the show - so I might have two pairs of speakers in the room.  And I'm working on several prototype stand-mount speakers, as I'd like to offer something in that category that's not another "me too" speaker with a 6.5" woofer plus 1" dome tweeter, like a hundred others out there (some of which are very good, but there's no point in my essentially duplicating what's already being done).

Another project is a small-room-optimized loudspeaker, but I have to do more research and prototyping to figure out what characteristics a speaker should have to really work well in a small room.  This thread has been so productive, I might start another thread on that subject some time just to get other people's ideas and experiences.  Most of my experience has been in medium sized rooms, so I'm sure there are a lot of AudioCircle members who know more about what works well in a small room than I do.

Duke

[Duke]

Hello WEEZ,

"Easier said than done" is right indeed!

Meeting the impedance requirements you mention (8-ohms minimum and non-reactive) probably quadruples the design effort required.  My current designs do have impedance peaks in the bass region so they don't really meet your non-reactive criterion.  Now I could significantly reduce or even eliminate those bass impedance peaks by filling the box with a lot of acoustic stuffing, but I use those peaks as my ticket to a free lunch. 

You see, a high output impedance tube amp will behave sort of like a current source, and will deliver greater wattage into a higher impedance load.  Take for instance the Atma-Sphere S-30 OTL amlifier.  It's a 30-watt amp into the 8-to-12 ohm load my Jazz Modules present across most of the spectrum, but into the 50-ohm twin bass peaks the amp is putting out closer to 60 watts!  So what I do is tune the cabinet much lower than normal, and use that 60 watts to get a "free lunch" - a +3 dB bass boost which (with the low tuning) gives me bass extension quite a bit lower than I should be getting.  A low-feedback SET tube amp will often have similar characteristics. 

With a solid state or push-pull tube amp I don't get the same free lunch, so in that case I shorten the port and thereby tune the speaker a bit higher. 

I have nothing against acoustic suspension, but recognize that a 1 cubic foot acoustic suspension box will probably need a subwoofer (unless the speaker's efficiency is pretty low).  Now the plate amp on most subwoofers has a 12 dB per octave low-pass filter, and theoretically at least this will give a better transition with an acoustic suspension speaker than with a vented speaker. 

One of the speakers I'd really like to build is a one cubic foot, 90 dB efficient, tube-friendly stand-mount speaker.   I've done a lot of modelling and some prototyping, but probably won't get back to it until after RMAF.  I guess you can tell that I lean towards vented boxes, because user-adjustable tuning really expands their horizons imho.  Unfortunately... I really dislike grilles!  I know they are sometimes necessary, but stilll...

I'm curious about the main reason behind the grille cloth requirement - is it for cosmetics alone, or do you want something that will protect the speaker from exploring little fingers? 

I don't like grille cloth filtering the highs, and grille frames can also have a significantly detrimental effect.  Let me show you:  Cup your hands around your mouth and talk.  Hear that coloration?  That's what a grille frame does (though to a lesser extent - grille frames are smaller than your hands and farther from the driver).  Now let me show you the solution:  Move your hands 1/2 inch out from your mouth so they are no longer touching your face, and talk.   See - the coloration is almost completely gone!  So the principle for grilles is, if at all possible, you want them to sit out away from the baffle a half inch or more.  Whether or not this gives adequate protection to the drivers is another question.

Thanks for your suggestions.  I'll try to get over my aversion to grilles.

Duke

[Duke]

To Kevin Haskings,

Are you still plagued by those bitchy laws of physics?  What you need my friend is a Marketing Department!

A Marketing Department does not fall under the jurisdiction of those so-called "laws of physics".  You want 20 Hz bass in a 100 dB, one-cubic foot box?  Easily done with but a few keystrokes!!

Pumping your hand in my plaid suit and checkered bow-tie, fin on my back and a toothsome grin,

Duke

[Duke]

Hi Jeff/TONEPUB,

Hope things work out well with your family, and of course you'll be missed at RMAF.

I'm still building Jazz Modules and Stormbringers, but have discontinued the Cheetah that I think you saw a couple of years ago.  I hope to show a new (larger and more ambitious) speaker at RMAF.  I'm also working on a couple of smaller designs, but nothing that will be ready for RMAF.

I'm not planning on exhibiting in Las Vegas next January, as I'm not recruiting dealers - which is the main focus of that fairly expensive show.  I'll probably go to gawk and schmooze, though.

Duke

[Audiovista]

This thread turned out to be very illuminating.

Duke, 

Are you a step closer to the new product definition now? Any open questions?

It would be so cool to see a new product specification being developed through this exchange of ideas. 

Boris

[Duke]

Hi Scott F.,

Thanks for writing back.

Driver specs are usually in the ballpark, but published curves are often over-optimistic.  So part of the process would be buying and testing drivers. 

Just for the record, the actual on-axis acoustic rolloff is imho what we're talking about when we say "first order crossover".  There may be no actual filter element, or the filter may be more complex than a single cap or coil.   And depending on the driver and crossover frequency, the power response of a given driver with a "first order crossover" may roll off at 6 dB per octave, 12 dB per octave, or not at all (what's happening to the driver's radiation pattern has an exaggerated effect on system power response with first order crossovers).  Also in my opinon, getting a first order rolloff for one octave on either side of the crossover frequency would have to be "good enough" - it's just not practical to expect to consistently maintain a true first order rolloff beyond that. 

One thing I would not be able to do is use the simple crossover topology that you suggest.  Unless we address driver resonance, we'd have audible peaking at resonance and power handling would be a joke.  So I'd have to use resonant peak filters (which would not be in the signal path).  I don't think this spoils the purity of the concept, but it does make for a crossover that's several times more complicated.  Also, it looks to me like setting sights on 92 dB rather than 90 dB would more than double the driver cost. 

Yes I'm a JoLida dealer, but I haven't tried the factory-modded JoLida 102b yet, though I bet it's sweet.  I have a stock unit that's out on loan right now, and it improved significantly with higher quality output tubes.  I appreciate your comments on the unit - soon as I sell my demo unit, I'll have to get one. 

Just curious - you mentioned zero feedback amplifiers, and I'm under the impression that the 102b has about 7 dB of global negative feedback.  I could be mistaken though, and maybe that's one of the factory mods.  Do you have any information on that?

It is still too early for me to say whether I can find a set of drivers that will be suitable for an all-first-order speaker.  I've already eliminated quite a few based on specifications, but the search continues.  You may be right in that it would take custom drivers to actually pull it off.

Duke

[WEEZ]

Hi Duke,

Well, the grill cloth comment was mostly due to spouse approval....most wives don't like to look at speakers . For me, it matters not a wit .

I thought I read somewhere that a SS 8535 would achieve 45 (F3) in a 20L sealed box. Assuming that's true; along with the more gradual roll-off of acoustic suspension loading...wouldn't you have 'usable' bass to the low 30's? 

Anyway, I just 'muse' sometimes; longing for a more modern, 'up-to-date' version of a speaker like the Large Advent... with a friendly impedence to use with tubes.

I missed your room at RMAF last year. I'll be sure to stop by this year....

WEEZ

[Duke]

Hi Boris,

Thanks for your words of encouragement!

Yup, what I'd like to see is a loudspeaker project that evolves from ideas proposed here.  It might end up as a commercial product and it might not, but if there's interest in the process I'll at least take it up to the prototype stage. 

At this point I see four separate ideas that might be within my capabilities:  Your augmented "fullrange" driver idea; Scott's tube-friendly first-order speaker; Hman's line array that doesn't need a sub; and Weez's tube-friendly small sealed-box speaker. 

At this point I've been focusing on the feasibility of Scott F.'s tube-friendly first-order loudspeaker concept. 

Earlier today I thought I had a set of drivers that would work with first order filters, but when I looked closer (at driver x-max requirements) it turned out that x-max would be exceeded at some frequencies at about five watts input.  Rats.  Unfortunately, x-max figures are hard to find for many dome drivers - so there may be some promising units out there that I don't know about yet.  I could push crossover frequencies higher, but that compromises the acoustic performance.  It'll probably end up being a juggling of tradeoffs - so what's new.

Even assuming I come up with drivers that look promising, I won't really know what the impedance curve is doing until I built a prototype and measure it.  So it could go bust at that stage as well, if I can't keep the impedance curve friendly while using first-order crossovers.

I have ideas on how to solve some of the problems my earlier augmented fullrange speaker had, if the problem wasn't flux modulation.   

I haven't looked into Hmen's line array yet; hoping he'll write back with his thoughts on the three possible approaches I described.  I haven't looked into Weez's speaker yet either.

One thing I'm definitely seeing - a trend towards tube-friendly speakers.  Three of the four explicitly or implicitly intended for tube amps, and the fourth (Hmen's line array) might as well be if that's what I end up going with.

Far as I'm concerned, there's still room on the table for more ideas, if anybody wants to make some suggestions. 

Duke

[Scott F.]

Duke,

You are running into the same problem I was when I was trying to do the same thing a few years back. When you start looking at the frequency response curves (forgetting power for the moment), it seems that most of the divers go into a heavy cone breakup just after they start to roll on the high side. This really forces everybody's hand in using steep(er) XO slopes if you are going to push a medium efficiency driver to its upper FR limits. There were one or two of the old Focal line that looked interesting but unfortunately, they pulled the best drivers from us DIY'ers.

One of these days, I need to sit down and read about driver design. I've picked up some info from the Cookbook but I'm sure there are some more in depth papers and books out there. I'd be interested in learning more when I can find some extra time.

On the modified JoLida 102b, I may have not been overly clear on that one. Yes, it does use feedback. I thought it was 6db but 7 is close enough 

I sure hope you can find some suitable drivers. Trouble is you may not. Of if you do, it may only be one or two which will really limit voicing of the speaker. Good luck in your searches

[Duke]

Hi WEEZ,

Wow, thank you!  You didn't wait for me - you found a driver that will do almost everything you're asking for!  That's great!

I modelled the Scan-Speak and it's an honest 86.3 dB woofer, and in a 1 cubic foot sealed box will give us an f3 of 48 Hz and an f6 of 36 Hz.  That is pretty good.  It won't be all things to all people, but then we both knew that going in.

Here comes the possible fly in the soup:  The DC resistance is only 5.8 ohms.  According to the published impedance curve the impedance minimum is about 7 ohms around 200 Hz.  I can't promise that a crossover wouldn't drop that lower - hopefully not, though.  So if you can live with a minimum impedance around or maybe a bit under 7 ohms instead of at least 8 ohms, we're still in the game. 

You mentioned the Large Advent.  Well, I think there's room for improvement, and what I'd like to do if we go with this project is build it as a three-way.  Here's my angle:  I place a very high priority on the reverberant sound, and if we build it as a two-way with a 1" dome tweeter, there would be a large discontinuity in the reverberant energy in the crossover region because we'd be going from a beaming 7" woofer to an ultrawide pattern 1" tweeter.   

Also, once upon a time I was a dealer for a company called Cliffhanger, and they made a lovely-sounding little speaker called the Bulldog.  It was a 6.5" three-way, and used the LPG 2" aluminum hybrid dome tweeter.  It was the midrange naturalness of this speaker that won me over - I had no intention at the time of becoming a dealer for a stand-mount speaker but this little beast was really something special.

Nominal cone diameter for that 7" Scan-Speak woofer is about 5.5 inches, so either a 2" or a 3" dome midrange should work well.  The Morel 2" soft dome comes to mind, as does the LPG 2" unit from the Bulldog; Madisound is presently listing it as "sold out" and I hope that doesn't mean it's been discontinued.  Then for the top end I'd go with a .75" tweeter instead of a 1" tweeter, and there Hiquphon comes to mind. 

Here are links so that you can take a look at these drivers:

http://www.madisound.com/catalog/PDF/morel/mdm%2055.pdf

http://www.madisound.com/catalog/product_info.php?manufacturers_id=138&products_id=547

http://www.hiquphon.dk/page13.html#Technical%20data%20OW%20I%20(D2094-8708-03)

Let me know what your thoughts are so far.

Duke

[Russell Dawkins]

Anyway, I just 'muse' sometimes; longing for a more modern, 'up-to-date' version of a speaker like the Large Advent... with a friendly impedance to use with tubes.

WEEZ

Jeez, WEEZ, that's pretty much my impression of the Jazz Module, it's even a similar price, once you adjust for inflation. Sounds like Duke has factored tubes into the design, too.

[Duke]

Hi Russell,

Thanks for the plug!

Actually, it was probably the original stand-mount Stormbringer that came closer to the Large Advent in size - though mine probably weighed around twice as much (90 pounds each).  The Jazz Modules on the other hand are 42 inch tall floorstanders.

One lesson I learned is, big stand-mount speakers are hard to sell.  So my next stand-mounts will be quite a bit smaller. 

Duke

[WEEZ]

Russell,

As Duke said, actually it's the Stormbringer that's similar in size. But it's ported . (Also, the SpeakerArt SuperClef.....)

Duke,

Far be it for me to suggest what to do about speaker design...I only listen to 'em...not design 'em. (which is a good thing....) All I know is, that an acoustic suspension 2-way is a rare thing nowadays. But there have been some good ones......and I liked most of them.

WEEZ

[Russell Dawkins]

Weez and Duke, what I meant was the Jazz Modules relate to other speakers of this time as the Advent did to the other speakers of its time, not that the size was similar.

[Duke]

Hi WEEZ,

Yeah, I have rather unorthodox ideas about speaker design.  I build two-ways with big woofers and relatively high crossovers, but when it comes to a small woofer I want to build a three-way.   What I'm trying to do is keep the radiation pattern as uniform as is practical over as much of the spectrum as I can. 

I hope you don't mind if I rattle on a bit about sealed and vented boxes.  I recently learned something that I'm still excited about (it's not really earth-shaking, but it is interesting to me at least).

Earlier this year I did some testing of sealed vs vented enclosures.  I built two identical enclosures, which were about the right size for a sealed box with that woofer.  Each enclosure got the exact same drivers and crossover.  The ported enclosure was ported to the rear (that's my standard practice for a variety of reasons), and tuned very low.   What I wanted to compare was midrange only. 

My expectation was that the sealed enclosure would have the cleaner sounding midrange, because if nothing else some distorted midrange energy would emerge from the port (imagine talking through a cardboard tube).  I conducted a series of blind tests - the listener didn't know which box was which, and in fact didn't even know what the difference was until all the tests were completed.   I played vocals with and without light acoustic instruments, and instructed the listener to only listen to the midrange.   In every trial, the listener chose the ported enclosure has having cleaner midrange.  I changed stuffing, I used different types of damping material, I did everything I could to help the sealed box (I wanted it to "win" because I was working on a design that called for a sealed sub-enclosure for the midbass driver).   The ported box won every time.

Now maybe if the port had been on the front of the box, the outcome would have been different.  But as I conducted the tests I could clearly hear it too - there was a slight "wooliness" or "boxiness" to the sealed box that the ported box didn't have.  Even when I constrained-layer-damped both enclosures (I had to do both or it wouldn't have been a fair test), the difference was still audible.  I switched drivers and crossovers in case there was a defective driver or miswired crossover.

All I can say is that I don't know how to build a sealed box that sounds better in the midrange than a comparable rear-ported reflex box.  Maybe there are secrets that I just don't know.   

Irving Fried championed the transmission line as sounding natural in part because it was a "low-pressure enclosure" - that is, it didn't put much backpressure on the cone of the woofer.  He characterized the acoustic suspension box as the worst in this respect, and vented boxes as being better than sealed ones.  Of course at the port tuning frequency the vented box puts enough backpressure on the cone to virtually halt its movement - but maybe that doesn't have much effect up in the midrange region.

Now as far as bass goes, yes I definitely would prefer the gentler rolloff of the sealed box.  If we accept the paradigm that the frequency response curve is a reliable predictor of subjective bass quality (I think I read an AES paper that came to this conclusion but I'd have to look for it), then one interesting possibility is a ported box that has a second order bass rolloff for about an octave or so - theoretically mimicing the bass tightness of a sealed box.

So anyway I still like your basic idea for a speaker built around that Scan-Speak sealed-box woofer.  If we went with a 1.5 cubic foot low-tuned reflex enclosure (tuned to say 24 Hz), we'd be -3 dB at 35 Hz and -6 dB at 27 Hz, with a second-order rolloff down to 20 Hz.  [edit - I haven't checked the power handling on this so can't say for sure that it's practical at this point].  And, if my little series of tests described above is reliable, we'd have cleaner midrange as well.  So... would you consider it blasphemy for the speaker to be sealed-box-mimicing rear-ported reflex enclosure? 

Duke

[Duke]

Hi Russell,

Wow, to say my speakers relate to others of our day as the Large Advent did to others of its day is quite a compliment.  Thank you.   Now I only wish I had some big bucks to spend on advertising!

Naw - I'm not practical enough to do that.  If I had the money, I'd most likey spend it working on new designs. 

Duke

[Audiovista]

Duke,

Glad you wrote about experiments with sealed vs. vented boxes. As a kid I used to make sealed speakers only, at that time have never heard vented box that would satisfy me in the bass region (too muddy, no real control) and I liked to build big boxes. Now that I grew up....out of 7 or 8 pairs of speakers that I have, only one is sealed, but I still had the impression of sealed box superiority. Turns out that the only sealed speakers I have are also the smallest ones and present the most difficult load...how easily we deceive ourselves....


Boris