Opinions Needed on Final Driver Selection

[slhijb]

Been designing a new speaker project for several months now and ready to take the next step, purchasing the remaining drivers. Since driver selection is so critical to success of the final design and these are rather pricey drivers, I’m looking for comments or lessons learned from anyone that has experience or knowledge of this type driver combination (or fractions thereof).

Tweets: Dynaudio Esotar T330D (already have these)
Mids: C-Quenze 15H (5”) w/Kapton
Woofers: 2each (4 total) Scan Speak 21W/8555-01 (allows for smaller box and increased sensitivity for baffle step compensation)

Two box general Watt Puppy style design, using BassBox Pro and Xover Pro for design calculations. Box design is pretty well complete (closed boxes) and Xovers still being worked on. Large room 25’ X 25’, but limited space for speakers. For WAF reasons, they have to be reasonably sized boxes.

These drivers appear a good sonic, crossover and timber match, but am I missing anything before I take the final plunge?

Thanks, slhijb

[slhijb]

No help or comments.... anyone?

[Curmudgeon]

Hookaayy, a cautious reply. (Albeit a lot more lengthy than I'd planned.) I think you have chosen well. You don't want to pad either the mid or woofers to adjust sensitivity, so some care there is needed.

A friend and I are building two pair that are similar.
The Great Project uses one each of the Skaaning C-Quenze 15H, SS Revelator 99000 tweeter, and SS 25W 8565-01. The 15H is the standard unit with concave dust cap, no kapton. I have heard that there is now a version??? with a new surround.

The pro's: timbral continuity is seamless. Dynamic range and transients are in the breathtaking class. Transparency is easily the equal of anything I've ever heard. (I've not heard any of the big 'stats lately though.) I cannot really say that it has a sound of its own; use slightly warm interconnects, and it is a slightly warm system; use slightly analytic interconnects and you have a slightly analytic system. The best recordings are magnificent, and it does not make poor ones sound worse.

The cons: the flip side of the above. You have a tiger by the tail. It happily responds to the very best amplification and crossover components; the clarity is such that there is no doubt whatsoever. Components that I was perfectly happy with in the past are suddenly shown to be less than the best.

We have mounted the mid and tweet in a box which is covered with a rounded semi-teardrop felt covered foam mountain, to reduce diffraction. The woofer is in a 6 cubic foot enclosure, 1" felt lined, tuned to about 17 Hz. (Extended Bass Shelf.) No audible woofer hangover. ("Fast", to the degree that that applies to woofers.) Cabinet has slanted back, shelf brace. Mid subbox is sealed.

We tried to watch xo component cost; only moderately successfully. Tweeter xo uses silver home-wound coils, polystyrene/tin capacitors.

Mid uses a silver foil lowpass (Just discontinued by Alpha-Goerz due to Ag price volatility) and 10 Northcreek 225V metall'd polypro's, 5 each anti-parallel. These may be becoming unavailable as well, VERY good value anti-paralleled. Other values tin/polypro.

The woofer uses an Erse low DCR inductor, metall'd polypro's. House curve LR, for about a -3 dB shelf high mid and tweet.

Mills 12W resistors throughout, Cardas lugs and their patent lower cost "binding posts" . XO moved back to the amp, wired with Cardas bare silver wire; home made speaker cables, Cardas 9.5 ga hookup for the woofers, 5 X Cardas 15.5? "ribbon -+-+-" and 5X Cardas 21Ga silver for the tweeter. XO star ground wired.

Home made interconnects still evolving; currently a variant of the Van den Hul hyperflex design seems to be becoming the design of choice, with Cardas silver and RCA's.

Rest of the system for evaluating all this is Levinson 30 DAC, Audio Synthesis passive pre, (modded slightly) and either a Bryston 4B or Pass X250. We just finished rebuilding the first XO to match the improvements in our second; imaging seems fine (not a major issue for either of us) and room placement seems relatively unfussy so far.

Other random thoughts. We had to put a trap on the woofer to suppress the HF breakup. (At least -50 is suggested.) The xo is basically first order multi-slopes, an idea cheerfully "borrowed" from the Thiele speakers. MLSSA and LEAP were used. (Do you have any measurement capability? We really needed it. )

[slhijb]

Curmudgeon, great reply! Thanks for sharing your project. Sounds absolutely stunning. It’s encouraging to hear your system’s going so well. Please keep us all posted on your progress, I for one am totally interested.

As a long time audiophile, but newbee speaker builder, I’m perplexed with 3-way Xovers. After 4 books, emails to DIY experts, reviewing web sites / diy forums, and working Xover Pro software, I think my head’s gonna explode. The 3-way’s very tough to get right. The first order multi-slope approach sounds interesting. Any insight or recommendations will be much appreciated. Would love to go 1st order if practical, little or no padding on the tweeter with cross @ about 2200 (Esotars can handle a large HF areas very effectively) . Woofers aiming at around 190 - 240 Hz. Shooting for a large gaps between Xovers and keeping them out of key vocal areas as much as possible.

Also, not clear on placing a trap on the woofer to suppress HF breakup? Help.

In response to your comments. I narrowed mid driver selection to the Skaaning C-Quenze 15H, 18H, or the Audio Technology 5” Flex unit. To help break the deadlock I emailed Per Skaaning and described my project. He responded that the 15H will be the best fit with its “new cone” (had kinda expected him to recommend the Flex unit). David Gatti said he got the same response when designing “the Delta“. Exactly what the “new cone” changes are, I don’t know. Sounds to me like you may have the newer cone. Maybe someone can chime in.

You discussed that your new system as very revealing, and that quality amplification and crossover components are a must. I currently have Merlin VSM Millennium speakers. They use the Esotar tweeter & SS 18W 8545mid woofer - superb speakers. Will use these to A/B compare the diy speakers once at testing phase. My Core system is: Amp - McIntosh MC-352 SS (300W/ch), PreAmp - McIntosh C-2200 tube, Cardas Golden (earlier version of the Golden Reference) and Cardas Neutrals interconnects. LAT international SS-1000D speaker cables, Electrocompaniet DAC and sundry other components.

Xover components planned: Caps - Either AudioCap Theta PPTor Hovland Musicap, Inductors - Alpha-Core and Goertz, and resistors - Axon/Mills. Will also look hard at silver wiring and solder since the McIntosh & Esotar tend to be on the warmer side.

Test equipment: Eyeballing: -- Wallin Jig, Behringer ecm8000 mic, and ub802 preamp, have the M-Audio soundcard and possibly Speaker Workshop software.

Plan to continue Xover design through further research and Xover Pro. Expect to have Madisound also design Xovers VS my designs using their chamber measurements and LEAP. Once speaker boxes are assembled, will build prototype Xovers using cheap materials and tweak the system. Once it’s right, will build final Xovers. Xovers will be in separate boxes outside the speaker cabinets for future adjustments. At least that’s the plan today. May change as I go.
Jim

[Curmudgeon]

I'll try to expand a little on some of the questions. We found out (usually the hard way) that a great deal of time and patience, and excellence at all stages is required.

Some of our learning process. (I had designed and sold commercial speakers, so I was surprised at how large the next step of quality was.)

Do not EVER, for any reason whatsoever, at any time, in no way, use clip leads for breadboarding xo versions or changes. We use Wonder and Cardas solder (the older lead containing) now.

Forty hours is a good rule of thumb for breaking in components before listening evaluations, although more for drivers is a good idea. (Inductors don't seem critical, caps certainly do.)

You will need your measurement and design tools.

Multi-slope crossovers. First order crossovers place a great burden on the drivers. Very few drivers are suitable. The problem is that the drivers need to be able to handle out of band material well. Peak to average ratio for classical music, uncompressed, is easily 20 dB. At 6dB per octave, that's 3 octaves; so you can have out of band peaks 3 octaves out equal in level to the average inband level. Not a reasonable thing to ask of drivers.

Thiele loudspeakers have used the multislope principle for years, and I happily "borrowed" from them. I spoke with Jim (?) Thiele at a CES once, and he told me that a second pole about an octave away from the crossover point, and a third another octave away, helped with the out of band power issues, without too great a phase error at crossover. I'd used the same approach. I've not seen a good how-to multi-slope design article for the amateur. As ever, these are the final acoustic slopes, resulting from the crossover and driver rolloffs summed, so the specific design will depend heavily on the driver out of band response. That's a big reason as to why your measurement equipment will be invaluable.

I like the Vance Dickason "Loudspeaker Design Cookbook". The title is misleading, it is far more than a cookbook. Traps are covered well there. We found that our choice of woofer (like many stiff-coned drivers) breaks up well out of band, (the 2500 Hz region) but that the distortion products (non-signal related on the whole) are very audible, and must be strongly suppressed. For comparison purposes, that area is now about 60 dB down. Because the mid is so transparent, it will not mask the woofer's misbehaviour at all.

I'm acquainted with the 8545 driver; it is very good indeed, but the Skaaning is a whole 'nother thing. The C-Quenze units are the newer frame design. We bought our drivers 3(?) years ago, so they must be the older surround/cone design. I have also heard rumors of silver voice coils....

Silver wire. To our ears, silver is the warmer and the cleaner of the two. I know many feel otherwise, but silver (for us) was always hash free in the upper mid, and all the copper we tried was not.

I don't think Madisound carries the 15H, so you probably cannot use their LEAP service. And I do think the 15H is the unit you want. Very good idea though if it is possible.

Diffraction effects were very audible to us, and we went to a lot of effort to avoid them.

Sadly, I cannot endorse the idea of using inexpensive components to breadboard your crossover. As you know, with systems of this level, you are spending a lot of effort to NOT hear things. Colleague and I are both unusually sensitive to upper mid hash and grain, and that is one of the things you pay (quite a bit) to not hear. We spent a lot of time and money, learning and relearning that we just could not reliably learn anything using inexpensive components. I cannot speak to your capacitor selection as we have not tried those. We used North Creek as the basic capacitors (woofer, some conjugates and Zobels) because they sounded fine to us and they were on sale that summer. From what I have read of the Theta's, they should do well. Mid and tweet as I indicated above require as good a capacitor as you can stand to pay for. We did not try Axon resistors; we found the MIlls to be fine, and did not look further. And because your system is different, (and room, and music sources, and ears, and cables, and, and, and) your component choices may well differ from ours. For example, I suspect that your cables are "better" than anything we have.

And to reiterate, capacitors, capacitors, capacitors. We cannot afford the exotics, copper or silver foil, teflon dielectric, but do use the tin foil polyprop and tin foil polystyrene (sparingly!) And we follow Serengetiplains' adventures with great interest.

Oh, and the crossovers are built on the white polyethylene 1/4" pegboard from McMaster. Spiffy.

PS With regards to the wire and other issues, both of us, design engineers, started out as skeptical indeed about some of the things we found out. However, our ears ruled, and we'll let the theorists and the material physicists sort out why we hear what we do someday.

[v-bro]

You people seem of the like I can learn from, true "audiophiles".

If I may ask, isn't making "the box with everything in it" (drivers, filters etc.) merely a consideration of ease?

I have very good though not very extended experiences with multi-amping and filtering them on the inputs. I filtered them with an active filter in the past (very expensive device from marchand) but wanted to make a less complex "routeing" and decided to make a passive line level filter (pllxo). Marchand also sells a complete filter like this, wish I had the schematic and component list.

Isn't haveing no components between the amp and the speaker an advantage to the amp being able to have better "control" on the driver and profit better from it's dampening factor?

Or are there any disadvantages on pllxo-ing besides insertion loss and an impedance shift on the amp's input?

How can the split outputsignal from the source be "decoupled" towards each filter? And how can that be done? Or do I really have to calculate every value taking in account the values from the paralleled filter for the other driver? This would mean that I can't make it a 3-way system in the future without changing all the filters again

My asking this is because I like to learn instead of buying some ready-made one...(and don't have much money these days)
And I have noticed a difference in sound when parallelleing the filters.

[v-bro]

this is the one..http://www.marchandelec.com/xm46.html
are the ceramic caps used in some way to decouple?

Anyone, thanx in advance...

[Curmudgeon]

I can't help much on the pllxo issue, as I do not have experience.

I thought I'd mention a couple more stray items that have occurred to me that might be helpful.

The mid box also has a slanted back, and is fully lined with ~ 1/2 felt. Also, the small flat rim of the magnet is felted to cut down reflection. For the mid, the mounting cutout is beveled at a 45 degree angle, leaving 6 (?) small islands for the mounting t-nuts. And the bevel is, surprise, felted.

[slhijb]

V-bro, sorry but I must also plead ignorance here. Just don’t have experience with these concepts. I’m considering bi-amping, but just in the early research stages myself.

Crumudgeon. Thanks for another great response. I took a little time to absorb your comments. Have come to the conclusion that the first order multi-slope Thiele / Crumudgeon crossover design is a bit advanced for me, especially without a good how-to or multi-slope design article to draw from. I’ll probably go down several Xover roads before a final selection is made, but for now am researching the following:

Concept #1. Bi-amp the 3-way speakers using both passive and active networks. Use a 2-way active low level crossover in front of the amp to filter the mid & high frequencies, from the low frequencies. The active filtered low frequencies will be sent directly to the woofer (driven by one amplifier). The mids and highs are actively filtered before the second amp and then directed to the passive 2-way crossover.
Note. At this point tri-amping is not an option. Not sufficient room in the audio component rack or sufficient green in the wallet for a 3rd amp with all the trimmings.

Concept #2. Tony Gee’s Andromeda 3-way crossover. Essentially designed as a 2-way system with a sub-woofer. He uses a 2nd order series filter for the tweeter & mid, and a parallel 1st order filter for the woofer. Schematics are attached and below.

http://www.speakerbuilding.com/conte...6_scematic.jpg

I emailed Madisound, but haven’t heard back from them. Have spoken to them before and they have always been super accommodating. Hopefully, they will support.

Just starting to research more closely Xover component quality issues. Hadn’t really looked at the prices of some of the high quality capacitors and such… ouch! However, until the Xover design is complete won’t know what the $ damage is, but your advice will certainly serve as a template. Xover component quality will be very high, commensurate with that of the drivers and cabinet design. For the tweet & mid crossovers I‘m considering Mundford Supreme Silver/Oil Caps and at low-pass filters as mentioned earlier. Again, will depend on the Xover results.

Your point is well taken on not building prototype Xovers with cheaper components to final tweak the design, then going final with high quality components. I’m sure you’re right that the subtle (and possibly not so subtle) nuances will be missing and the ultimate design will suffer. Wow, hope the initial LEAP design’s pretty accurate or things could get expensive.

Agree with silver wiring. The way to go for this project.

Had not thought of using felt for the magnets and beveled / chamfered backs, but I’m getting a roll of felt anyway and will definitely attach it at these locations.
Thanks again, any ideas on the two Xover concepts?

[Curmudgeon]

From a theoretical standpoint, active crossovers have a lot of advantages. Very attractive.
The problem I have with active crossovers is the difficulty in implementing the active stages; it is not easy or inexpensive to make a preamp equal in quality to the amps we've been discussing. And an active crossover can have quite a few stages. Coupled with the cost of an additional amp at this level of performance....

If I were to undertake an active, I'd keep it as simple as possible. A passive RC high pass, and the output of an opamp (integrated or discrete) taking the difference between the passive's input and output will sum to the exact input waveform. The opamp's output will only be first order however. (Been quite a while since I messed around with that. ) There is nothing to prevent you from using a blend of both methods.

We felt that excellent results have been achieved by others using very conventional methods, and, rather than reinvent the wheel, or try for a stunning breakthrough, we'd simply try to use classic methods implemented very carefully. And that alone was enough project for us! A bit more than we expected; the driver transparency makes hitherto inaudible differences perfectly plain.

One approach to multislope is to use modern optimizing design tools. I can only speak of LEAP, but I know other tools have worked the same way, as I used one (XOPT?) pre-LEAP. Basically, you construct a guide curve, for each driver, with the desired response. Then you give the program the driver response, the guide curve, and a reasonable circuit. (That is, you don't ask for a fourth order response and feed it a second order circuit. And the originating values should bein the ballpark as well.) The tool will show the initial response, and you can tweak values as needed to get somewhere close to the guide curve; then activate the optimizer. It will change the values as required for the final response to match the guide curve as closely as it can. When satisfied, you repeat the process with all stages for the final result. With some, depending on the effectiveness of the convergence algorithms, you'll get different results by modifying the initial values slightly. (LEAP's convergence algorithms are excellent incidentally.)

It takes some practice, and understanding of what you want, to use one of these tools effectively. This is really the most practical way to develop the base design. This sounds simpler than it is in practice, as you will listen, measure, modify, recalculate, and relisten as needed.

LEAP may be overpriced and overkill for your needs; but it is extremely well designed software and very good value in its league. There has been a lot of discussion about other design tools more in line with amateur budgets.

The Mundorfs, from what I've read, do seem very well worth trying. And yes, experimenting with components can get expensive.

A lot of effort should be spent on the basics; diffraction, internal reflection, (reflections off the rear wall, and back through the mid cone, delayed, are EVIL) physical alignment of the acoustic centers, star wiring, connectors, and whatever I've forgotten are all important. If I had infinite resources, I'd experiment with the stacked layer cabinet techniques. And more cap types.

You should, of course, be well acquainted with live music, and measurement equipment that allows you to take waterfall measurements is a big help. And again, I do recommend Vance Dickason's book highly.