"Fundamentals" for a new stand mounted speaker project -what goes around comes around

[JonMarsh]

Or, the circle is a wheel, and does history repeat itself?

12-13 years ago I was building some largish 8" woofer stand mount speakers using Woodstyle W123R cabinets as the starting point; about 48 liters internal volume, 12" by 25" front panel, and 14.5" deep. They're no longer available, but you can snag a PDF of what Madisound sold back then, using Google. That's what I did...

So, here I am, 13 years later, and planning to re-visit that form factor, though with rather different technology. Just think how much has changed since then... SACD was the new wonder kid, I had (have) an SCD-XA777ES and a handful of SACD disks, and I was using a Sony power amp and an Aragon (which I also still have tucked away) with an ES series preamp. I thought I was in heaven... little did I know!


The parts have arrived and will be concept tested soon- this start is a preview of what I'm hoping to get using 30 liters of this new enclosure. First, the model parameters and verification for the 4 ohm SW223BD02 - the system will use one:





In a 30L enclosure with the Wavecor PR using no added weight, just the stock cone mass, for a PR FSP of 15 Hz, we can get a fairly flat alignment down to about 23 Hz. To say that will spank the M8a two way is one of those understatements that beggars the imagination. But then, the SW223BD02 is a bit more power hungry, and features about double the Xmax of the HiVi driver. What's hilarious is that this sub style driver has a higher resonant peak break up frequency- some good engineering there.




From one cabinet we should be able to crowd close to 100 dB anechoic all the way down to the roll off point:




Next week I'm ordering midrange/tweeter drivers to test. Had to raid the piggy bank to do that... it's going to be interesting to see how this plays out.

In the meantime, I'm gluing up panels for side walls and getting ready to transfer testing back to the Livermore homestead- it's a little too industrial around here for GF- can't hardly blame her!

[TEK]

Interesting.
How deep will these bee as you haveto glue up several panels?

[JonMarsh]

More than 12"- I haven't finished all the calculations for things like driver volume, midrange enclosure, etc, but I"m expecting between 15 and 18", er, 315 to 390 cm. front and rear baffle will be doubled up 3/4" LBL, the rear being recessed inside the outer walls. May not be particularly pretty, have to start in the CAD concept soon. Thinking to step the front panel so that only one layer shoes from the outside, and bevel that.

[JonMarsh]

Running around doing errands today, but now I'm getting setup to route some test baffles- for the C18EN001 (which I just missed receiving on Friday- didn't get back home in time) and some of the other drives here to test- the SS 7140 for MEB346, the ScanSpeak 12M4731T00 for Renron, and of course the BSW223BD02 for everyone (myself included) planning a build of one kind or another (Wavecor Ardent Mk1.1, Isiris Stand Monitor) with that driver.

That little Seas Illuminator midrange sure is cute, but it will definitely take some careful rear bevel work to unmask it adequately. BTDT. :W

And I'm setting up the Auralic Vega ("portable" DAC to give it a quick listen and start what some people call a lengthy burn-in before really listing- more about that later in the Digital Audio section.

Should have some good data to post by sometime tomorrow!

[JonMarsh]

I thought I'd posted on this earlier today, but must not have it the "post reply" button after the writing and editing.

Got some decent measurements in after switching back to Fuzzmeasure 3, finally; wrestled with 4 for about an hour, and then wrote a lengthy "feedback" to them about their new piece of software.

Anyway, the SW223BD02 looks just fine, and the measurements correlate well with what I'd expect, based on Unibox modeling.



This is the near field output of the SW223BD02 in one of the 28L test cabinets- the one original designed for a UMAX 10" woofer, but with a built up back panel, and the 10" Scanspeak PR just with the standard weight.

Next is the response putting a microphone several feet off to the side of the cabinet, so we get some summing of the woofer and PR output.




It's about as well correlated as I think I could expect measuring in room, not outdoors or in an anechoic chamber-

I think I'm going to have to build a test cab for the PR223BD02 and SW223BD02, and compare how they stack up. Maybe I need to find a way to shoe horn the SS PR into this project... it certainly does work nice, and should have about 2 dB more peak output- without a significantly larger enclosure. Hmmm...

At the least, this enclosure and the 0.75L PE box for the C18EN001 will allow me to acoustically prototype it fairly quickly... I'm thinking a 2.25" thick LBL front panel, with some bevel work, but 90 degree enclosure walls.

[5th element]

Now that does look interesting. You'll be onto a winner I think with the C18 + SW223. I am looking forwards to your measurements of all the listed drivers.

[JonMarsh]

The C18EN001 should be delivered today- I'm going to be at home, and I've signed the delivery release! I'm looking forward to seeing and measuring them after seeing Troel's measurements. I've got the box and baffle already setup... still hoping to keep this in the range of 26-27" tall, outer dimension about 12" wide, and something like 15-17 or 18" deep. A largish stand mount or shelf mount box... have an idea for a tapered side front bevel, and bevel at the top. Gotta find some time to work on CAD drawings....

[JonMarsh]

Since it's not real on the internet without pictures, voila, Pictures!




These are little gems... I'm sure not inexpensive to make...



Can't wait for the weekend and some time to mount them up and test!

[5th element]

There are many reasons why I would enjoy using a pair of those

[JonMarsh]

I am very curious to see how well they might optimize, as coincident drivers - Troels has already shown that the time/phase alignment is almost perfect, and the TAD's I've heard had a very nice sense of coherency, and a very wide optimum listening position. That's exactly what I need for this system... can always use them for HT mains, too. Though they're almost overkill- ought to be pretty good just in a 2.0 setup for HT!

[JonP]

Yeah, beautiful drivers indeed... I got to check one out at THE Show in Newport... Madisound had a room, man, they had $3-4000 in drivers out on the tables for the masses to fondle... Satoris, Illuminators, Revelators, Accutons, up to 10"... and that Seas... very nice.

Also, 3yrs running at earlier shows, have heard the famous TAD References, with those concentric Be mid/tweeters, being demoed by Andrew Jones himself... (guess that's now history) and I agree on the ridiculous amount of coherence those things have... Always one of the best of show.

The Seas may not be made of vapor deposited Beryllium, but it seems to be exceptional good as an example of the concentric driver. Results should be... impressive.

[JonMarsh]

I'm sure hoping so, Jon. I'm going to start back on this today- route another test front baffle, and start putting together a 2nd test low frequency box- it occurs to me I can do a lot of crossover development using some test cabinets before I can get the nice pretty LBL one's put together (though that fabrication process has already started, too, at least for the cabinet side panels).

I've seen Madisound at the RMAF in the past, but haven't been there in a while. OTOH, Josh and Adam do figure I'm something of a regular customer... can't imagine why...

Yes, I guess what I'm shooting for is not SOTA, but a applying those practices to some more affordable components. Of course, we'll put caps in the crossover like they WERE Beryllium domes... :W

[JonMarsh]

It's almost 2 PM but I can say I've gotten a bit done today.

First path of attack was the cabinets, I unpacked one of the two extra Maple 0.75 cu. ft PE cabs I keep in the shoe closet, and masked it off (literally) for some inexpensive Dayton binding post cups.



Pretty easy to cut out without messing up,



and what do you know, but they just drop right in. (truly not what I would call challenging wood working...)



Because it occurred to me that I might actually want to prototype a pair of these and evaluate the crossovers before building the final cabinets, I went ahead and routed another front panel up...




Naturally, I saved the best part until after lunch...




Measurements!


Not any big surprises here, either good or bad. All measured at 4.25 VRMS, a bit hotter than the normal 2.83 VRMS. Scope monitored.




On axis, 15 degrees, 30 degrees, and 45 degrees, midrange.




Distortion on axis. A bit higher 2nd than 3rd, to be expected.


Tweeter measured the same way, from 0 degrees to 45 degrees.



Pretty similar to Troels' results, but with above 20kHz also.


Tweeter distortion up last, note that by graphing with dB, you get a lot more relative difference at the bottom end of the scale-




Really not bad, though not astoundingly good, either. Perhaps "quite serviceable" would be a good expression?

What will be interesting to see is how nice a crossover can be put together, and how the final result looks. Certainly looking like some kind of acoustical 4th order at roughly 2 kHz.

[JonMarsh]

Of course, we're not done collecting data. But this gets it a little closer:


SW223BD02




C18EN Midrange





C18EN Tweeter



Definitely going to need a full zobel around that fundamental tweeter resonance- it's been ages since I've seen one up that high!

I'm thinking 350-400Hz for the woofer to midrange transition, though a bit higher would allow a less expensive crossover; I wouldn't go over 500Hz; that would be just 2 octaves for the midrange, with a 2K or so crossover to the tweeter. I don't see the mid working well for a crossover point beyond 2200 Hz.

[5th element]

Looks like the standard SEAS performance. 3rd harmonic dipping to around -60dB and then hovering at -50dB for the most part, with the 2nd around -50dB. Excellent performance really but not quite as good as the best. I am glad you did distortion measurements, Troels' always does % scales that squash out the fine detail, these let you see everything as it actually is.

With the midrange breakup causing the peak in the third order you most certainly want to use a 4th order 2kHz minimum on it. Luckily the tweeter, with the wave guide loading of the mids cone, should be able to handle that without issue. It even has a decent amount of output down to 1kHz, quite surprising for a tiny dome like that. Your measurements at a higher drive level are nice to see because they give you more of an indication of what the tweeter can do. Troels' lower level distortion measurements show rising 2nd order below 2kHz, but at the higher drive level it starts to rise far higher.

Your measurements also show somewhat nicer 2nd order performance on the mid down at its lower end, it should easily be able to cope with a 150Hz 4th order no problem if you wish to go that low.

Linear distortion is very nice for both, nice and smooth, especially the tweeter, amazingly it's smoother, both on and off axis, than some standard tweeters! The breakup is fairly benign too. I really want to use a pair of these.

[JonMarsh]

Once more, unto the breach...

Well, believe it or not, I didn't just dive into whipping up some typical crossover... no sir, not me!

Let's "phase it", metal cone drivers can give anyone a case of "audiophilia nervosa", and looking at the response curves of the C18EN001 made me realize it was likely a bigger challenge than even the estimable Dayton RS-180 with it's hideous breakup peaks- that, and the high Fs and limited low frequency capabilities of the tweeter...

BTW, I made an "executive decision" to use the 15 degree axis as being the most typical of the averaged forward response, in this way, I think the forward averaged power response will be more accurate than using the direct on axis curve.


Hmmm, seems like a job for the Dark side.. reaching back into my older, more evil bag of tricks. Can you say, "elliptic filter"? How about, "high order LR acoustical targets"? Well, LR6 acoustical for the mid and tweeter.

The one thing particularly working in favor of that sort of approach is the coincident time aligned configuration- in fact, that makes goes a long way to making something that is possibly border line feasible quite realizable, if not quite a walk in the park. :twisted:


All I've been working on is the midrange to tweeter network. I figure a plain old LR4 will do nicely for the mid to woofer; for now I'm thinking around 300-400Hz, to keep the component cost in check- yes, we could take this down lower, but then that would also lead to cost issues with the woofer crossover.



So, here we have "Exhibit A".



Believe it or not, this IS an effort to keep the component count down. Remember, the midrange is component rich, and there's a LR4 acoustical high pass in there, too!

How's it look?




Pretty dang good, I think, as regards the overall pass band, and the suppression of break up modes.

Of course, the obligatory inverted phase response:



Nice deep symmetrical null, and I believe that's exactly what you'll measure, as there won't be lobing issues from non-coincident drivers. And being coincident drivers, the summing in the crossover region with the normal connection will be as complete and ideal as possible, with the best overall forward power response in the crossover region.

I can't emphasize enough that I think the time aligned coincident configuration will allow this crossover configuration to perform and sound much better than with conventionally mounted drivers. And I believe the extra attenuation is really necessary to give the C18EN001 it's best chance to bloom.

Of course, there may be other voicing issues to address; with as uniform a forward power response as I suspect we'll wind up with, some down tilt from the crossover region may be desirable, and adjusting the input resistor to the tweeter network has that effect. Obviously, a lot more work to do, but for one day, I think this is some real progress, definitely getting a handle on things.

[dar47]

Man they look old school, where are these going again?

Have to say though when you get time off you sure can push out woodworking and networks fast, Great job.:T

[JonMarsh]

The configuration shown above is just for prototyping. I'm planning a 12" wide by about 26-27" tall by about 16" deep bamboo cabinet for everything, with a three layer front panel, with a tapered bevel top to bottom on the sides, and bevel on top. Gross volume about 48L, with a tapered midrange enclosure front to rear, plenty of room for crossovers internally. Might have a partially removable back to facilitate that (behind the midrange; woofer/PR chamber completely sealed separately.)

These have two planned functions: monitors for music recording and mixing, and also possibly demo monitors or evaluation speakers for the Class D SMPS project, which seems to be a full go ahead, based on the response to my proposal. Demo room reserved at CES. I have NO idea what the class D module guys use as reference equipment, but based on a sort of 3rd party eval of their current pride and joy, it could be better, perhaps. Anyway, I'm leading the local team, developing two scalable designs, one in the "low power" range, (225W) and another in the high power range (1kW max for now). And I think I have a cool plan and concept, but shouldn't talk about it online until we have running hardware. Maybe not even until January.

A lot of this is like a contingency plan - it's a situation I never expected to find myself in, where I'm doing audio related work at work (actually did some a couple of times before, but not with this exposure or impact). Leasing a team on this, working with the El Segundo TM manager.

So that's why a more portable digital source system is being put together (the Auralic Aires should ship shortly, to go with the Auralic Vega and Mutec that I have; I'm going to build a LLC resonant SMPS supply and Rubidium oscillator into a small Par Metal chassis soon). I have no idea what the IR guys use... I'm going to be prepared- if I have to go to CES and demo it, and also for subjective evaluations (I'll build a version compatible with nCore modules, too, and I have my M22 as a reference. On the module side, the big deal is the expansion of GaN based versions, for higher switching frequencies, and moving the output filter load dependence well above the audible range.

I am jazzed about yesterday's work- the crossover actually started out in Modula/NatP territory, and quickly evolved into something a bit different as I wrestled with the whole midrange thing. I also quickly concluded that LR4 just didn't provide enough suppression of the break up modes, or protect the tweeter on the low end, but LR8 was just a bit twitchy for component tolerance, though at one point I had a nice version of the tweeter high pass working nicely that way. LR6 elliptic filter just ended up "feeling right". (that's what happens when a musician and psychologist tries to play an engineer on TV...)

Of course, it's the listening that will be critical- I'm debating building a sort of low cost parts version crossover just for measurement basis and initial listening before buying all the expensive caps... and of course, wondering how far to go with the caps this time... :W

Now, after we get the GF's big activity of the day out of the way (Danville 4th of July parade, she is one of the participants) then I'll do some "normal" measurements of the SW223BD02+PR box, and see how close we can get to a finished crossover.

I don't want to sprain my arm patting myself on the back, but I'm starting to think these are going to rock, to use an older turn of phrase. In an old school sort of way, hence the thread title. But when I started the thread, it was just a hope. Now I think I can taste it a bit, after the work finished yesterday.

And they'll confuse the average person, too, because from the front, there will seem to be just three woofers of different sizes: one seven inch, one almost nine inch, and one ten inch. I love it... :B

[dar47]

At 12"x27"x16" they will go in the element nicely but I bet they are going to be very heavy and require a sturdy short stand.:T

This should be fun seeing where you end up with parts quality and I'm very curious to see what you do to power the ncore module as I still debating a ncore build or NAD M22.

[JonMarsh]

The nominal target is an IR AMP6 modules, but for myself, it would be nCore. I'll PM you some info about it later this weekend or early next week.

[DS-21]

Are those IR AMP6 modules load-invariant like the Hypex/Icepower/Anaview?

All I've been working on is the midrange to tweeter network. I figure a plain old LR4 will do nicely for the mid to woofer; for now I'm thinking around 300-400Hz, to keep the component cost in check- yes, we could take this down lower, but then that would also lead to cost issues with the woofer crossover.



So, here we have "Exhibit A".



Believe it or not, this IS an effort to keep the component count down. Remember, the midrange is component rich, and there's a LR4 acoustical high pass in there, too!

Interesting coincidence: in terms of parts count you're right where Andrew Jones was on the S-3EX concentric: 10 on the mid, and 7 on the tweeter.

[TEK]

Any rough sketches for how you are thinking the cabinet will be?
Having a bit of a hard time picturing it from the text.

[JonMarsh]

Are those IR AMP6 modules load-invariant like the Hypex/Icepower/Anaview?



Interesting coincidence: in terms of parts count you're right where Andrew Jones was on the S-3EX concentric: 10 on the mid, and 7 on the tweeter.

They're supposed to be. I haven't had a chance to listen to them, though another colleague has extensively; there seem to be some issues with tonal balance, but it could be the power supply; could be a few different things. I'm promised some from the next sample batch. Also, we're doing one for the smaller modules; that run from single ended supplies. Same conceptual design for both, though the bigger ones will have synchronous rectification on the output. Think in terms of an LLC converter that never operates in boost up mode below Fo, but works more like an IBC (Intermediate Bus Converter) with moving out of the unity gain resonance point to effect protection or soft start. There's a few other ideas that may make it into the bigger ones... more like stretch goals. Using PFC, with 3rd harmonic injection shaping of the current profile to get lower ripple voltage on the bulk bus, OR smaller bulk caps for the same ripple voltage, or some trade off of the two, while still getting a PF (power factor) of 0.9. Conventional rectifier-capacitor setups have a PF of about 0.55... not so great for getting power out of the AC line.

There were quite a few more until I started peeling back some zobel networks- I have a deservedly bad rep for complex crossovers. And weird ones- see the Modula MT and NatalieP.

I presume from your statement that the S-3EX crossover is posted somewhere on the web? I'd be curious to see it...

[JonMarsh]

Any rough sketches for how you are thinking the cabinet will be?
Having a bit of a hard time picturing it from the text.

Just a normal rectangle, with some stuff done to the baffle; though I did think about complicating it with a lean back... it's a lot of trouble as we all know to do that. Something a little like a Joseph Audio Perspective but shorter...

[JonMarsh]

I've got a couple more sets of woofer measurements, trying to get cleaner ones while still indoors (it's 96 here...) and almost through integrating that with the rest. And I'm trying moving the lower mid crossover point down a little further...

[JonMarsh]

Well, it's only a sim, and unfortunately I had to up the component count after lowering the midrange to woofer crossover frequency, and doing my best to optimize the phase in the mid roll off as well as all the other things going on- the Fs driver impedance was an issue.



Took out one cap, added in a conjugate network for impedance compensation, that I just decided I had to have to get the midrange roll off to really behave properly.


The SPL magnitude and phase behavior looks pretty good, could have been nicer looking if I had outdoor measurements- too hot!



But this will be a good test for load invariance in amplifiers, as with a nominal 4 ohm woofer the impedance is sort of punishing in the low end, being around 3 ohms from 50 Hz to 250 Hz. So, while the speaker should have a nicely extended bottom end, if the amplifier or power supply doesn't, it will likely be obvious...

Baseline will be any of my Aragon's (8008 X3, or Palladium Monoblocks) or the NAD M22. The Palladiums are kind of like the "anti-Class D" amplifier, being about 100W in class A, and not clipping with 8 ohm loads until close to 600W. Pretty brutal power supply. Of course, they weigh about 72 lb. a piece, too. If this ends up being able to compete with that at all, we just may have something. Thats a big "if" though.


I'll probably look in the parts bins and see how close I can come to roughing this out and testing it; I can always unwind some inductors to get values I need. But we'll have to see if I have time to do that during the next week- have a lot of logistics chores and other stuff on the to-do list.

[DS-21]

They're supposed to be.

Everything past this sentence was too much for me, whether it's talking about "tonal balance" in amplifiers (IME, the only amplifiers with tonal issues are ones that aren't load-invariant, i.e. ones that mess up the FR) or the technical details. (I have no mind for electronics, and no incentive to learn because the performance differences in electronics throughout my exposure to this hobby - i.e. since the mid-1990s - have been minimal to non-existent unless something really sucked.)

I presume from your statement that the S-3EX crossover is posted somewhere on the web? I'd be curious to see it...

No, it isn't. At least, to my knowledge it wasn't until this post.

From the service manual:

Tweeter:




Midrange:

[Steve Manning]

Baseline will be any of my Aragon's (8008 X3, or Palladium Monoblocks) or the NAD M22. The Palladiums are kind of like the "anti-Class D" amplifier, being about 100W in class A, and not clipping with 8 ohm loads until close to 600W. Pretty brutal power supply. Of course, they weigh about 72 lb. a piece, too. If this ends up being able to compete with that at all, we just may have something. Thats a big "if" though.

I thought you said the weather was hot out your way ....... a 100W class A ought to get the room good and toasty

[JonMarsh]

Everything past this sentence was too much for me, whether it's talking about "tonal balance" in amplifiers (IME, the only amplifiers with tonal issues are ones that aren't load-invariant, i.e. ones that mess up the FR) or the technical details. (I have no mind for electronics, and no incentive to learn because the performance differences in electronics throughout my exposure to this hobby - i.e. since the mid-1990s - have been minimal to non-existent unless something really sucked.)

the performance differences in electronics throughout my exposure to this hobby - i.e. since the mid-1990s - have been minimal to non-existent unless something really sucked.

This doesn't mirror my experience at all, but maybe I haven't been far enough up the food chain. It's only in the last 5 years that I'm starting to become OK with the performance of available DACs (sonically), for example. And power amps and preamps have long been a vexing issue.

I think that a better term would be load and level invariant. A host of amplifiers may sound pretty similar at a few watts out, but things come apart at the seams when you're pushing peaks to 100W and beyond, and especially depending on the crest ratio (peak to average power- high average power usually makes things worse, but sometimes it's peak reproduction on which issues become most apparent).

There are amplifiers that are clearly limited by the power supply (in my experience and opinion), and if with sufficiently low output impedance (that load invariant part) will sound quite similar at lower output levels assuming equivalent frequency response and distortion characteristics that are in the ball park. I was a co-owner in a high end audio shop in Boulder Colorado in the mid 70's, and it was interesting how different some of the power amps sounded when you pushed them beyond the 5-10 watt area. The STAX class A DA300 did better than most, because you can't really cut corners on a power supply for class A amplifiers, (compared with the Luman M4000 or M6000)



But it did not have that quality I would call "effortless" when an amplifier has no impact on it's large signal behavior due to either the internal circuitry or the power supply. The smaller DA80 introduced in '75 didn't have it either. A heavily modified Dynaco Stereo 400 did have that quality, and we used that as our comparative reference when reps would bring around something new for us to consider selling. Besides several classical war horses the title cut from Stanley Clarke's "School Days", released in '76, was a great test- to keep from wearing out copies of it we taped it on an Otari MX-5050 half track.


Then there are the many amplifiers that are quite load dependent, even as shown by what I would call small signal tests, such as John Atkinson does for Sterephile with his simulated loudspeaker load (I have built one of those myself, copying his circuit and scaling it to be able to handle fairly high power without significant heating effects- you don't need an AP to do that tests, Fuzzmeasure or any other measurement program that does sine chirps works nicely).

These are typically the vacuum tube amplifiers, and the non-negative feedback amplifiers that Theta and Ayre and others build. Unless you go to heroic measures in the output stage, the unfortunate thing about NFB amps is the often trade one problem for another. Some loud speaker manufacturers deliberately cater to this issue, by spending extra money in impedance conjugate networks in order to make the impedance curve as flat as possible. Early Avalon models did this- later ones do not, and can prove difficult for many amplifiers.



But back to Class D.... like linear amps, they are usually expected to have a fair amount of PSRR- Power Supply Rejection Ratio, at least at low frequencies- but they accomplish that by negative feedback, as much as possible- because the output of a rail to rail PWM modulator is intrinsically dependent on the power supply rail voltage, UNLESS you do something like I did in the 1990 AES paper I wrote (published in the 1991 AES journal) describing a clocked class D amplifier design which used power supply voltage feedforward to adjust the modulator triangle wave height and the modulation index as a function of PS voltage. We were developing high speed low QRR 250V MOSFETs at that time, using a platinum doping process, but they closed down the fab they would have been produced in (in Munich) and relocated the power MOSFET manufacturing to a former memory fab in Villach, Austria. So that never reached fruition. BTW, my name is Jon Mark Hancock, not Jon Marsh; the nom de plume is for forum purposes, and derived from my mother's maiden name.

Another bugaboo for Class D amp power supplies is rail pumping- when driving low frequency reactive loads (woofers) there is current transferred from the load back into the negative rail, (assuming conventional split rail power supplies) which tends to pump up that rail voltage. One way Class D guys deal with that without spending money is to wire the two channels out of phase, so that they are pumping opposite rails.

The brute force methods few use because it costs more money in parts is to use large output filter caps.

Then, let's consider a possibly typical if specific case, the IRAUDPS3 reference power supply design, a split rail LLC with +/- 30V rails. In order to take the feedback for voltage regulation, they decided to sum the voltage across both rails- it's a little tricky to regulate two outputs with a single feedback point and a single primary converter, unless you can tightly couple the secondary sides, say be using synchronous rectification. You may see this coming- the problem is that if the rail not being loaded is being pumped up by the reactive current flow through the Class D stage, the feedback loop has no reason not to allow that, but will decrease the voltage on the loaded side, so that the net total voltage across the +/- rails is constant. Not really a good thing at low frequencies. The output filter caps are 4700 uF, enough to supply and absorb power in the midrange and upper bass, perhaps- not enough for lower frequencies at more substantial drive levels.

I haven't heard this combination myself, but one of our managers in Munich has, and shared his impression and his wife's in their system, with which I'm quite familiar, having spent a lot of time there listening to and evaluating gear over the years. The sample he had used GaN transistors and higher switching frequencies, which improved the top end, as it pushed the LC filter load dependence out of the audio band, but the LF performance was wanting. IR uses a 2nd order self oscillating Class D configuration, whereas the nCore Hypex uses a 4th order, which includes the output filter in the feedback loop.

Another problem for low frequency reproduction, which can include how large signal transients are reproduced that can be described as having a low frequency power envelope, is the difference between RMS power and DC power- I had to explain this twice to our marketing guys- being used to telecom and server applications, the requirements of audio took a while to sink in.

IF the AC load current is at higher frequencies, for a regulated power supply, it can come out of the caps (assuming sufficiently low ESR), and 100W of DC power can supply 100W RMS more or less (class D) to the load (typically about 95W); but if it is at low frequencies that the output filter caps can't supply, and the regulator must track directly the required output current. Then in principal one must supply up to 1.414 X the DC power for the peak of the since wave, which would make 100W RMS more like 150W DC power.

An SMPS supply can have one striking advantage, if done right: energy storage. Energy stored in a capacitor is defined by (C X V^2)/2, so upping the storage voltage gets you a lot more than upping the capacitance the same amount. Storing energy at 325V or 450V is much more effective than at 30 or 50 or 75V, as long as you have a way to deliver it when needed.


Well, those Pioneer schematics are very interesting! A remarkable coincidence, or just engineering inevitability? The first time I used an elliptic filter in a crossover network was in an 8" two way project I designed in 2003, and which was published in 2004 in AudioXpress - same component topology in the tweeter network, because if you're doing a modified 4th order electrical converting one element to an elliptic filter, there just aren't many ways to do that.

Bandpass crossovers see a lot more variation- especially depending on what the driver performance is, if you're working near a natural roll off point of the driver, or if you have to go full boat electrically. The most effective way to do a complex high order bandpass is to split the series elements to opposite sides of the signal path - again, converting the last element to an elliptic filter. There aren't a lot of different good options for how to do that, if the basic electrical intent is similar...

Have you built or modeled the Pioneer crossover? (since you have those drivers) How does it look? I did come across the crossover for the in wall using that driver, but it's quite a bit different...

[AJINFLA]

Nice project Jon.:T
That's a rather hefty "bookshelf". Of course Earls Summas would make them look like mini-speakers.
Remember those Madisound Woodstyle cabs well. Too bad it seems the company went kaput.

cheers,

AJ

[DS-21]

This doesn't mirror my experience at all, but maybe I haven't been far enough up the food chain.

You go way higher up the food chain than I do...I drive my primary system with an AVR used as pre-pro and use an AppleTV as my source most of the time.

It's only in the last 5 years that I'm starting to become OK with the performance of available DACs (sonically), for example. And power amps and preamps have long been a vexing issue.

Perhaps I am fortunate that every time I've tried to hear such differences (i.e. with some sort of voltage-matched, and blind if possible, protocol) I've been stymied. Most recently, I compared a miniDSP PWR-ICE125 plate amp (Icepower 125ASX2 module, unknown-to-me DAC chip, AND a AD-DA loop because we used the analog input) with processing zeroed out to a Benchmark DAC feeding my reference ATI amp. Speakers were KEF RDM Twos. We matched voltage at 120Hz at the binding posts with a cheap DMM, but listened sighted. I couldn't tell them apart for the life of me. Now, there was a big difference between the two that disqualified the Benchmark as a serious product to me: it had a turn-on thump. But that sort of unacceptable usability error is not a sonic difference.

The exceptions, electronics that have sounded different to me are amplifiers that suck, i.e. are load dependent (e.g. most tubes, most Class D) or have high noise floors.

I will leave this discussion (I consider discussions of audio electronics thoroughly uninteresting) by noting that it's certainly possible that you, because of electronics expertise and experience that vastly outclasses mine and indeed most of the rest of the online audio world's combined, have trained yourself to hear artifacts that go unnoticed by most passionate music lovers.

a modified 4th order electrical converting one element to an elliptic filter,

Now I get it. I was puzzling over exactly what was going on there.

Have you built or modeled the Pioneer crossover? (since you have those drivers) How does it look?

No, but it's on the to-do list. I actually use them with the stock in-wall crossovers (dual-woofer model) and for the left/right speakers use Anthem ARC to get the bass balance right. Other mod is two F-13 felt strips between concentric and woofers to reduce diffraction. It's admittedly a total kludge that should not have lasted as long as it has, but it works pretty well and I find myself wanting to listen instead of tinker when I have free time. I may try to find time

In the future, I expect I'll be using some variation of that circuit (all three standalone models with this driver use a similar circuit design with different component values based on desired voltage sensitivity, etc.) for the concentric, except with the midrange highpass reduced to one pole (that just to protect the driver from DC) and just go active from midbass to concentric. Since I already own a miniDSP and enough amp channels, the cost of going active is much less than buying big coils to build and tweak a midbass passive crossover. Also that way I can use whatever midbasses I want, be they 94dB/W/m B&C's or 85dB/W/m Revelators, without worrying about matching things up.

But enough about my future vaporware project. I'm eager to see what you come up with next for this pairing.

[---k---]

Some day, I would love to read the Jon Marsh biography.

[JonMarsh]

Some day, I would love to read the Jon Marsh biography.

Sadly, I doubt that will ever be published! :W :B

Besides, like a lot of people, I have a somewhat misspent youth I don't want advertised!

[JonMarsh]

Nice project Jon.:T
That's a rather hefty "bookshelf". Of course Earls Summas would make them look like mini-speakers.
Remember those Madisound Woodstyle cabs well. Too bad it seems the company went kaput.

cheers,

AJ

I think this is going to be a fair amount of fun, once I get them running...

There are multiple motivations, including wanting to try out a nice concentric driver, and also intending "deliberately, and with malice aforethought", to make a test system this is something of a b*ll-buster as regards impedance characteristics and testing the old concept of load invariant amplifiers. If the LspCAD modeling is correct, it will be in the 2.5-3 ohm range from about 60 Hz up to over 200 Hz, but as you'd expect, going much higher in portions of the midrange and high frequencies, due to the waveguide operation and resulting crossover. Not at all like the Wavecor Ardents or the Isiris; for example, the Ardent is one of the most benign to drive systems I've ever designed, varying between about 5 and 9 ohms, due to the behavior of that funny 3rd order crossover alignment I like:



But that's a big part of the reasoning, to have a challenging test speaker for some of the power supply and Class D work I'm going to have to do over the next six months.

[JonMarsh]

Well, discussions of amplifiers and source gear and what one person hears versus another can be a very big time sync for CPU cycles. And not always a very productive affair...

A friend of mine in Munich has a wife who's something of a litmus test... she's a pediatric doctor, a fairly sharp cookie, and likes a lot of different kinds of music, though she's partial to classical and native acoustic music from her homeland in India. She's sometimes not all that vocal about what she things about a new component or system modification, but when all she does is getup and walk out of the room after a few minutes of the music playing, there's not much doubt where she stands! :W

I was a classically trained musician before I went down the primrose path of rock and roll, and even then, was involved in projects in the 70's like recording acoustic jazz and classical performance live; the most challenging one was a performance of Saint-Saen's Symphony Cminor Organ Symphony, by the Boulder Philharmonic. OTOH, the tape did sound pretty good played back on the Levinson HQD system we had at the store... I used to have a copy of that, but it was stolen in a home break in, along with my precious vinyl rig in the early 80's. (Denon platter, Rabo tone arm, Denon and Dynavector cartridges, DIY custom build Phono preamplifier patterned after Mark Levinson in physical design, but using my own design moving coil transformer, and passive interstage EQ.

Not long after that I got married, never did replace that setup, and in fact went with CD's, mostly because of the convenience- because they sure didn't image or sound as natural at the vinyl, though they were quieter. Sort of like a good mediocre tape machine, but much more convenient. (a familiar tale for digital).

It wasn't until getting a Berkeley Audio Alpha DAC (where else, in Berkeley from the dealer near them!) that I finally believed digital had a real sonic future at a vaguely affordable price (not talking about things like a dCS stack). It was/is far, far better than my Benchmark DAC, and still serves me in my bedroom system.

A friend at work couldn't believe I'd spend that much on a DAC, or that it would make that much difference, especially with the frequency response and distortion characteristics of speakers, but it's been my opinion for a long time that electronics and speakers have rather different sorts of error signatures... anyway, we did a little experiment that I proposed, comparing the Berkeley Alpha DAC and his CD player (his choice, not mine!!) on a cheap pair of computer speakers (Audioengine P4's) that didn't use internal amplification (so as to screw things up with DSP or whatever), just on a small Sony HT power amp I have had for ages- he thought I couldn't possibly hear any difference if everything was level matched (by scope and meter monitoring the voltage on speaker terminals using a Sheffield test CD), but not only could I hear the difference and identify blind which was which, well, he could too, as it turned out.

Benthe8track has recently build a DIY Hypex stereo power amp to replace his AVR for driving the Ardents; this made a nice difference, but not as much as getting a new DAC, the NAD M51 (my favorite recommendation in the $2k class).

I don't hear as much natural live sound as I'd like, but I do usually manage to get to 2 or 3 classical performances per year- I loved being over in Prague a couple of years ago, because we probably went to half a dozen concerts while we were there. It's a good way to set expectations, if you like the music.

[JonMarsh]

I forgot to mention, I did go ahead and order parts for building one test crossover. It wasn't cheap, even though I tried to keep the pricey stuff down to a minimum- well over $400 for one channel worth. Should be here by Friday... have to see what else is going on if I'll make some progress, depends on how things go with GF's daughter's surgery this week.

[dar47]

Hope tings tun out well for her.

Well Jon much to my annoyance, I'm starting to agree that the amp my be down the list for me too but I want a new one.8O

Went to my local b&m to see if he had brought in an NAD m22 yet (he didn't) but notice there was a nice sale on the Moon Mind 180 streamer. Store wasn't busy so I got a manger to do a little test for me, some 44.1 and some 192 jazz, note I hadn't heard these tracks before. We did a little compared,

1. stream from his notebook usb through the M51, Moon 740p pre, 760a amp and them to some PSB monitors (not sure of the model number).
2. stream form his NAS through a Bluesound Node (usb) and the same.
3. stream form his NAS through a Mind 180 (AES/EBU) and the same.

I got to say 2 distinct walk ups just feeding the dac. I had planed just to do a HTPC to handle both music and movies but now I want the little Mind it just had a blacker background, better lows, mids and sweeter highs. The Bluesound Node was better then the notebook, cleaner but not as much as the Mind. I think I my do NAS and Mind first, then maybe HTPC and then amp.:T

One more thing after we were done I notice a guy waiting on him who had what looked like a little old sewing machine box in his hand. He opened up the box and pulled out a set of cans, I guess he wanted to check amps out for them. I almost lost my sh!!!

[JonMarsh]

I think you've got the right idea for that upgrade path- and a high grade network streamer is a good path to go down, if it has the right technology. I just finished the "basic" load on to the QNAP HS-210, which is going to be my "mobile" NAS (i.e., back to Livermore, in to work, possibly to El Segundo, etc.), and now I FINALLY have fired up the LaCie BIG NAS and am starting to load it up as I write this- the Acoustic Alchemy folder is copying over now with all the albums in it. This will stay here, in Danville, connected to the main network and with a separate leg for the Audio Subnet running off it's own WiFi router, a recent model Airport Extreme.

I have no trouble accessing these NAS via my Mac (only test so far); later this week I'll try out the NAD M50 and the new AURALiC AIRES that arrived earlier this week. The AIRES has dual "femto clocks", supposedly very stable and low phase noise, one for the USB setup, one for the S/PDIF and AES/EBU, though that doesn't really make sense to me, as what WOULD make sense is having two master oscillators, one to support 48 kHz based frequencies (96, 192), and the other to support 44.1 based frequencies (88.2, 176.4). Maybe their marketing people got it wrong, or maybe both clocks are dual oscillators. Who knows? As long as it sounds good, I'm not going to worry...

In a pinch I'll use a Mac, but my personal experience is that a well designed dedicated ARM CPU based device is a better sounding route to go.

This review of the MIND 180 is pretty typical.

http://www.soundstagehifi.com/index.php/equipment-reviews/779-simaudio-moon-180-mind-music-streamer

Nice looking UI.




So far, I think the AIRES is going to do what I expect fairly well; the only down side I've noticed so far is that the plastic top and bottom lips make it really hard to unplug connectors in the back. Build quality is first rate, especially for something with that much plastic. Has WiFi built in, you connect to it via your iOS device for control. You can have multiple streamers in one setup, running from one server, and you select which to control by changing which one you're logged on to.

[JonMarsh]

I thought you said the weather was hot out your way ....... a 100W class A ought to get the room good and toasty

VERY much so; which is part of why there is now an NAD M22 at GF's, as the family room doesn't really have any air conditioning. Even the Cambridge Audio 840XD power amp idles at 180 watts; a pair of Palladium mono blocks is about triple that. Too much for this time of year! :nonod:

But you know what, I'll probably use the Palladiums to do the voicing, because they will drive anything- they're about as load invariant as anything I've ever owned.

[AJINFLA]

I forgot to mention, I did go ahead and order parts for building one test crossover. It wasn't cheap, even though I tried to keep the pricey stuff down to a minimum- well over $400 for one channel worth.

One of the myriad of reasons why I personally prefer active bass.
Not to mention smaller enclosure, mid sensitivity/impedance sets baseline, etc, etc.
You are a Luddite! :P Btw, I do agree with much of what you said about some thrifty Class D designs and current limiting supplies. DS-21 actually owns a prime example.

cheers,

AJ

[JonMarsh]

You are a Luddite! :P

cheers,

AJ

One of the prerogatives of having reached my level of maturity is being a luddite is just considered age appropriate behavior... no one expects any better! So if I surprise them every now then, it has even more impact... :B

In this case, though, a MAJOR reason for building this speaker is making a single amplifier design that separates the wheat from the chaff when it comes to amplifier and power supply behavior, so let's just say I've picked this primrose path through the weeds pretty carefully... at least, more carefully than it may seem to the casual observer... :W

[DS-21]

"Fundamentals" for a new stand mounted speaker project -what goes around come...

Btw, I do agree with much of what you said about some thrifty Class D designs and current limiting supplies. DS-21 actually owns a prime example.

I do? (My SAM1's have the BASH amp) The icepower module? FWIW JA measured at least one product using the Ice 125 for Stereophile: http://www.stereophile.com/content/b...mgfucgVcQkv.97

[AJINFLA]

I do? (My SAM1's have the BASH amp)

No, implemented specifically to avoid dynamic limitations.

IThe icepower module? FWIW JA measured at least one product using the Ice 125 for Stereophile: http://www.stereophile.com/content/b...mgfucgVcQkv.97

Yep, is the BC using the stock ASX2 version?

cheers,

AJ

[AJINFLA]

In this case, though, a MAJOR reason for building this speaker is making a single amplifier design that separates the wheat from the chaff when it comes to amplifier and power supply behavior, so let's just say I've picked this primrose path through the weeds pretty carefully... at least, more carefully than it may seem to the casual observer... :W

Gotcha, that should do it.

cheers,

AJ

[ergo]

I tried to order a pair of C18 too to kick of my own build.... but local reseller got this reply from Seas
" About the date for possible production date we do not know at this stage, but I believe it will probably be closer to 6 months than 1-2 months."

That might be the reason why they have 'hidden' this model from web for now too. So I guess I need to find a source in EU who would still have stock.