Experimental fun - a bit different than the usual

[cjd]

Just a Neo8 and Neo3. Crossover as shown. Measured response at about 1.5m. With the signal cut at 200Hz, however it is ARTA does that. The Neo8 is open back here as well.



Clearly, some issues to deal with. A nasty problem on the Neo3 at just about 10kHz, and a sad dip between 1k and 2k on the Neo8. I'm pretty sure I know what's causing the 10k issue, but this is a test rig and you can't have everything.

Yes, that is JUST a Neo8, and yes, it really does go that low. My usual solo unaccompanied violin was pushing the excursion limits, but that reaches to ~200Hz (196Hz if you must know).

I think I might be able to cross this to a 12" driver though, no? Maybe just a 10" though.

Or maybe a 15"

C

[---k---]

I'm not sure we've seen finished pictures of your other projects. Until you get those done...

I think I might be able to cross this to a 12" driver though, no? Maybe just a 10" though.

Or maybe a 15"

C

Guess it depends on whether the woofer is going to be OB also? Might need multiple woofers.

[cjd]

I want to push as low as possible full dipole. If I need multiple woofers to do the job, I'll need to push to a 5-way I suspect, but if I can get one larger driver to bridge between subs and the Neo8 that would be nice. However, I suspect that's not going to happen.

On the other hand, I don't have room to actually place these anywhere in the house the way things are shaping up. I'm going to need to find myself a warehouse.

I wanted to build something yesterday so I worked on this. Lovely shade of Owens Corning pink. But that's why I worked on these.

C

[Saurav]

Why would it be hard to find a 12" or 15" driver to do 60-80Hz to 200-250Hz? The Eminence Alpha 15A is a popular choice for a somewhat higher Qts driver that won't need as much EQ. I'm using the AE OB12, which requires more shaping, especially on a smaller baffle. I've seen several other woofers used in various 3-way OB designs. You might be able to get away with a shallow U- or H-frame too, if you keep the cavity resonance up above 4-500Hz.

I haven't looked into matching sensitivity though, which you'll need since you're probably looking for an all-passive solution rather than a hybrid.

[cjd]

I want to push higher than 250Hz. While the Neo8 is playing that low, it's at its limits. That said, it should still be possible to find one that is comfortable up to 500Hz (maybe 750Hz) or so.

I do want that driver to be passively crossed.

Below that will have to be active I think.

C

[Saurav]

Maybe you need 2 Neo8's Yeah, subwoofer to 750Hz on an OB is asking a lot from a single driver. Many of the AE woofers have a response notch around 500Hz that makes them really usable up to about 300-350Hz. Maybe one of the M series midwoofers. I know there are some designs using the Alpha15A up to 1kHz or so, but I have no idea what the distortion performance is like on that driver.

[cjd]

No no, subwoofer is in addition. This will be a 4-way at the least. But I'd like to pull the sub in ~100Hz give or take. Lower is better.

Will have to be at least 24" wide, so there's room to play a little. Also will likely be ~24" deep...

[Saurav]

Yeah, I meant, "from a subwoofer hand-off on up". Do you really want to switch from a woofer to a planar at 750Hz? And you should post some pictures of your current baffle, I'm curious. Are you playing with your dipole horn ideas?

[cjd]

Ahh, I understand your "sub on up" now. Hah. Gotta love the ambiguity of English. And Americans.

You don't think a pair of Maelstrom 21's would fill in the bottom end? A pair, each...

Hmm. Maybe I need a bigger planar. I could go to a pair of Neo8's also. I think the amount of low end extension available here was better than expected, but I may also try to scale back a little to more of a waveguide than a horn.

I can take some pics I suppose, though I've been messing around with it already again.

[Saurav]

Someone on DIY Audio built something with the Neo10 and Neo3, but he admits he has low SPL goals.

[cjd]

Yeah, I got the Neo8's before the Neo10 was around. *shrug*

Playing around with some 12" driver I had sitting around forever. It's interesting. The 12" is just propped up. No measured data on it either, so I have no idea at all how it's integrating (or not).

I have time.

[kevinp.]

this may sound silly, but what about the RSS-315hf-4? It's pretty flat out to 1k, but the breakup at 1800 would have to be addressed if crossing higher than 3-400 I'd think

[cjd]

Sensitivity is low for passive crossing without using multiples (or carefully set up amps, at which point you might as well be active), which means at least 4, and yeah - would need some pretty steep filtering as well.

I'm largely looking at pro drivers right now. 12" should make the crossover smooth work, but I'd need to fill in the gap between that and sub-woofage. I could push smaller as well (which would allow things overall to get smaller as far as the horn goes too, I think) but right now the experiment is to see just how crazy-ish this can get.

C

[cjd]

On the other hand, I may just need to step up and go Neo10 now that availability is less of an issue (seemingly)

[cjd]

Good response in one area brings with it a different set of problems, of course...

In this case, an overall cabinet design is heavily problematic for bass response because any type of horn loaded top end makes the default "obvious" choice for low frequency to be an H baffle...

But of course you get a nifty set of problems with that: bigger = lower frequency QW resonance and more issues.

So. Yeah. I think I have to see if the Neo10 can be similarly pushed low, and low enough that I can cross right to a dipole subwoofer driver or two.

Finding good high-Q drivers with decent sensitivity AND a good open frame is an interesting challenge (if I want to try to NOT go Neo10 in the mix).

[twest820]

What's the THD on the Neo8 look like at your target SPL with dipole EQ in place but without a crossover? That'll tell you where to cross and the woofer size falls out from that; if you want to run the Neos as low as feasible I would suggest crossing at least LR4, preferably steeper, and using a peak rather than a shelf for dipole EQ (a Linkwitz style shelf plus additional highpass works too but is more complex for, often, worse results); active or digital makes life a good bit easier here. The THD data I've seen so far for the Neo8 has been unimpressive; up 1% or so by 1kHz, so especially for low frequency extension I'd look more to the Neo8-S. Though for what Audax sells the 8-S for it's cheaper to get Neo10s from Parts Express or even Meniscus.

Which Neo3 do you have? Never seen a 10kHz chuckhole like that on the W or PDRW measurements. Baffle? No baffle? And what is the SPL target?

My nude Neo10 data's here. I think Saurav may have missed 1) the data includes 95-100dB SPL as well as more typical figures and 2) while I chose crossover points to match my listening levels higher SPLs, up to a point, just mean higher crossovers. Greg's posted baffled Neo10 THD in the drivers forum here and there's Zaph's infinite baffle data as well. Between the three data sets there's probably sufficient information for most any design.

Another thing to consider is the aspect ratio on the Neo8 is higher than the Neo10 so the 8's directivity is less consistent. How much that matters depends on the cross off the Neo3, which in turn depends on SPL targets. Neo8 directivity's pretty well characterized; IIRC the data's linked from the Neo8 measurements thread in DIY's planars and exotics forum. If your SPLs don't accomodate use of a Neo10 the best option's probably to switch to push/pull pairs of mids---see the push/pull thread at DIY. Push/pull directivity's tough to get as good as a single Neo10, though (and the Neo10's not all that great, really).

I've mentioned this elsewhere, but Selenium's 12PW5, 15PW5, 18WS600, and 18SWS1100 are hard to beat as pro drivers for dipole use. Qtses aren't particularly acommodating of passive crosses, though.

[Saurav]

and using a peak rather than a shelf for dipole EQ (a Linkwitz style shelf plus additional highpass works too but is more complex for, often, worse results)

Worse in what way? I vaguely remember you mentioning using a peak filter for dipole EQ, but never really visualized what you were saying until this post.

I think Saurav may have missed 1) the data includes 95-100dB SPL as well as more typical figures

My bad

[cjd]

It may not be clear if you've not followed my various random ramblings that happen all over the place, but the Neo3/8 pair are in a horn for the data I pulled. The null is a direct result of that, I think, as the throat width at the driver is equal to the space between the driver and the top of the horn (which is straight) ... Not absolutely sure on that.

Neos are both non-PDR plain jane versions. Cup still on the 3, 8 with nothing behind at the moment (so only horn loaded on half, and I know that will also change things - I'll get to measuring that).

And while I am not opposed to active crossovers, I'll need MUCH improved noise (or lack) through the midrange (i.e. I need to improve my PS design, and possibly more) to want to go that route. So right now I am looking to cross the top 3-way all passive.

And none of this resolves the inherent problem with overall dimensions this way or that and how a baffle would naturally try to find its way together given the whole package at the moment.

C

[Saurav]

the Neo3/8 pair are in a horn

Pictures!

[cjd]

Neo10's on the way.

[CraigJ]

but the Neo3 are in a horn for the data I pulled. C

I'll show you mine if you show me yours. Pictures?

[twest820]

Pictures!

Hear, hear! :T Any thoughts on Geddes's suggestion of OS throat toroidal type baffles as dipole horns/waveguides? Or Le Cleac'h profiles?

And while I am not opposed to active crossovers, I'll need MUCH improved noise (or lack) through the midrange (i.e. I need to improve my PS design, and possibly more) to want to go that route.

An LM7x15 pair and some LME49740s would give you 150+dB PSRR, ~4.5nV/sqrt(Hz), and ~2.5pA/sqrt(Hz). What am I missing? MiniDSP's parts aren't on quite the same level, but the boards measure out well, are cheap, and getting the filter into digital more than compensates even if you're using state variable filters for an analog cross. Still easier to do linear phase with a PC crossover, though.

Worse in what way?

Nothing major. What I generally end up with when using shelving filters is a second order lowpass with Q>1 to continue the slope of the shelf up to the cutoff frequency. Produces a decent roll off, but is a fourth order solution and requires component matching between the shelf and lowpass to get things properly lined up. And in my experience the slope on the shelf hasn't been the greatest match to the driver's actual roll off. A peak/dip parametric EQ accomplishes the same thing with fewer resources and opportunity for error---for the same complexity it's my experience a second peak/dip EQ is more useful. It's also easier to correct in Arbitrator if you're going that route for linear phase.

That said, I have a few other ideas stacked up to try. Mostly amenable to digital, though.

My bad

I wrote a small book.

[cjd]

I'll get pictures later today I hope. Maybe another data pull also, though that may just have to wait, but I need to get moving on another project (holidays are approaching fast).

Profile is a rough exponential right now, but it's assembled from straight sections, which adds some roughness to things (clear to me as I work with trying crossover topologies, but maybe less so from the raw data).

And LM whatever's still need a good quiet power supply, and as I said...

I really really do not want to go digital>analog>digital>analog and I can't see whether the MiniDSP's take digital input at all - nor can I tell if I need to keep them connected to a computer. For prototyping, tons of value to be found there, ignoring those potential issues.

C

[Saurav]

And LM whatever's still need a good quiet power supply, and as I said...

PSRR = Power Supply Rejection Ratio, I think he's talking about chips that aren't much affected by power supply noise. Still, I agree with you, rolling your own analog active XO isn't trivial.

I can't see whether the MiniDSP's take digital input at all

There are modules for USB and S/PDIF input, I think.

nor can I tell if I need to keep them connected to a computer

Don't know. I suspect not, you should only need a PC hooked up to program the thing.

I haven't really taken a close look at miniDSP. Functionally, it doesn't seem to do much that the DCX doesn't. It's cheaper, but you also have to build it and add wire, connectors, enclosure, etc. Probably lower noise, but I've mostly solved that with an analog volume control after the DCX. Better sounding? I've seen so many opinions / measurements / etc. on how good the DCX sounds (or not) that I have no idea what to believe. I haven't taken any measurements of my own. So... for me, not sure what I'd gain by swapping the DCX for a miniDSP.

[cjd]

PSRR = Power Supply Rejection Ratio, I think he's talking about chips that aren't much affected by power supply noise. Still, I agree with you, rolling your own analog active XO isn't trivial.

Er. I missed seeing that, but still think power supply is key.

Then again, I may have a nice quiet power supply, and the other circuits I was using are where I have problems. I'll get there.

MiniDSP talks about I2R headers and it sure seems you keep it digital till you're done, but the diagrams don't help. Maybe they're not technical enough for me. Or maybe the assumption is that most folks will not be trying that, and will just want to use analog inputs and not care about DADA. It does appear you can source into the MiniDIGI and keep it digital till the final output stages.

[cjd]

Image not available

When I took the measurements, the little "ramps" in the horn mouth were not there.

[twest820]

Hmm, what's the target profile of the horn? Why not, say, a Le Cleac'h 360 profile?

If you're trying to build an active analog crossover but aren't familiar with audio op amps (currently dominated by National's LME497x0 line and related series) or linear regulators (LM7815, LM7915...) some homework's probably in order; there are worse places to start than doing Sallen-Keys with the parts I mentioned. Happy to take a look at your schematics.

For moderate cost digital without a PC MiniDSP's site and their forum over at DIY cover the basic questions raised above (personally I'd do SPDIF in, balanced out). Motu's interfaces are worth a look as well. Plenty of PC crossover threads over at DIY too---post number 3 in the Neo10 measurements thread I linked above has my setup.

[Dennis H]

If you're trying to build an active analog crossover but aren't familiar with audio op amps

I've been away from the forums for a while but I got a chuckle out of that one. Trying to 'educate' cjd about crossovers is like trying to teach your grandma how to bake cookies. Hi, cjd!

[Dennis H]

About high-Q drivers for dipoles, that's sort of a straw man. They most often achieve high Q by having a weak motor (small magnet). Everything else being equal, you can use a series resistor with a big-magnet driver and get the same Q, SPL curve, sensitivity, etc. as the small-magnet version.

[cjd]

Hey Dennis Yeah, I'm not really that fussed with high-Q, but series impedance is its own mess. I'd much prefer active filtering to clean things up, all things considered, so I'm starting to lean that way, but I'd still like to go lower on passive filters, so... the Neo10 purchase.

On me and op-amps... I have done kit chip-amps, self-designed chip-amps, a bit of P2P, and some pre-amp and DAC work as well (some kit, some partial kit). I've struggled with noise almost every step of the way in one form or another. I'm going to see about rigging up a plain circuit on one of my self-designed chip-amps, as I never did that - only tried something rather more complex.

Horn profile was where I started getting sensible data out of sims. So I figured I'd run a real-life test to see just where things were going and what needed work. LeCléac'h is also rather difficult space-wise for dipole use. Nice results if you're just looking to horn load and go monopole.

[Dennis H]

I've still got this notion of a passive dipole from 80Hz on up floating around in the back of my head. The AE TD15M seems like a good candidate. It's been described by John Janowitz and Nick McKinney as the best of the Lambda drivers. It's big enough for good sensitivity and it's well behaved well above where you'd normally cross it.

[cjd]

That's one of the 15's I'm looking at.

Also have my eye on one of the B&C Neo magnet drivers (15"). Superb assembly on the back for dipole. Or even 18" like the B&C 18PS76 - if I read this right, 16.5 ohms in front of it still nets 92dB and an F3 of about 40Hz. Baffle size depending. And as I will be almost guaranteed mounting into an H baffle of some sort... will have to measure to be sure though. Not really an option unless the Neo10 fills the gap adequately, and I'm beginning to think it just might.

[cjd]

Whoops. Missed a setting that was applying some active EQ to bring up the bottom end. So much for that. Still, promising.

[Dennis H]

And as I will be almost guaranteed mounting into an H baffle of some sort... will have to measure to be sure though. Not really an option unless the Neo10 fills the gap adequately, and I'm beginning to think it just might.

You can always cut some holes in the sides of your H-frame if the cavity resonances are a problem. A naked dipole 15" should be able to get down to 80Hz so you don't really need the H-frame for bass extension.

[twest820]

I've struggled with noise almost every step of the way in one form or another.

If Dennis is indicating help is unwelcome I'm happy to move on. If not, I'd need specifics to say anything useful but chronic noise problems seem a bit surprising. For example, the chipamp board I just did is below the -110dBV floor of my measurement gear and it's just a basic layout with no unusual attention paid to noise management. (As an aside, I have to admit my grandmother's cookies were rather pedestrian and I've also improved upon her pie crust, may she rest in peace.)

LeCléac'h is also rather difficult space-wise for dipole use.

Hmm, the dipole profile Jean Michel calculated is fairly compact by horn standards. Can you put this in the context of a desired cut off frequency?

[gbegland]

I've still got this notion of a passive dipole from 80Hz on up floating around in the back of my head. The AE TD15M seems like a good candidate. It's been described by John Janowitz and Nick McKinney as the best of the Lambda drivers. It's big enough for good sensitivity and it's well behaved well above where you'd normally cross it.

Well, I'm getting down to 35Hz just fine with the Dipole 15 from Lambda on a baffle not too much larger than the driver. Only using minimal EQ too. When I'm really pushing things, levels at the couch are peaking at around 90-95dB SPL which makes about 105-110 at the drivers. Been using a couple of Maxwell CDs called "umbra" and his latest "A Black Summers Night". Pretty much audiophile quality R&B. Lots of really players, not too compressed and gobs of beautiful low end. Lambdas are cruising along with a low cut at 35Hz on Maxwell and need no low cut on most other discs.

Greg

[cjd]

Re: help and noise

Help is welcome. I am just omitting lots of bits that would probably help you understand better where I'm coming from, but folks that have been around probably have picked up some of this over time.

Noise: It may be that my hearing is fairly acute randomly up to and above 20kHz, or I'm otherwise sensitive to certain types of noise. Some things I am sure better power supply isolation would solve. Some things are probably fat-fingered soldering technique. I'll get there, I am sure, or maybe I won't. I have no background in electronics, so where I dove into self-designed (or extrapolating beyond any published work) I probably did a lot of fairly simple stupid things.

So, when I get to putting together active circuits, I'll be posting more about that - if I go get PCB's made I'll post that too, prior.

Can you perhaps point me more specifically at what you have in mind as far as horn profile? I may just be looking at the wrong thing, and I'm still spending way too much time reading up on various tidbits. Just starting to make some vague sense out of things, and at that, not really. Yet. My goals are to both narrow the dispersion profile to be more in-line with what large low-frequency drivers end up with (the classic 8, or not, as it ends up) and to extend low end to allow fewer drivers with better characteristics. If the Neo10 can comfortably cross down into territory where a more plain dipole "sub" driver is playing cleanly I'll be pretty happy with that result. Otherwise, a 15" (or possibly 18" with Neo10) would be lovely, and subs additional below that.

I've not found an easier pie-crust than my great-grandmother's but it lacks something in richness and flavor that I've moved beyond. It got me understanding HOW to make pie crust, though. The family hot-cross-buns (i.e. easter buns) which she translated from "about this much" into actual measurable amounts, however, is just about un-touchable.

[twest820]

Can you perhaps point me more specifically at what you have in mind as far as horn profile?

I spent a while last night trying to find mige0's build photo before my previous post without any luck. Tossed him a PM but it might be a bit as I'm likely offline for the next couple days.

I'm still spending way too much time reading up on various tidbits.

Wait, you mean horns are supposed to do something besides consume infinite amounts of project time? I'm also interested in using dipole horns/waveguides to reduce room interactions by narrowing the pattern and extend the operating bandwidth of drivers. Probably more in the waveguide space as I want to control both vertical and horizontal directivity and it gets hard to do that with full length horns. Their many other virtues aside, the Neos are attractive for this as their rectangularity builds in a certain increase in vertical directivity. I wouldn't mind some horn gain, either, but mostly I'm convinced H and U baffles represent a design point optimized for ease of construction and not for directivity or transient response.

I'm rather fussy about both those, hence my choice of a point source Neo3/Neo10/18 nude build. Dipole peak on nude 15s is around 630Hz and about 525 for nude 18s and for most SPLs the Neo10 wants a cross in the vicinity of 350 (assuming moderate order like LR4). So there's really no difficulty crossing the Neo10 to a sub acting as the bottom of a three way. And plenty of scope to lower the sub's dipole peak for improved efficiency. The nude rig sounds pretty good and posts reasonable THD at reasonable volumes. Main problem I hit is the time of flight across the 18s is long enough the first reflections off my ceiling aren't point source and the imaging suffers. I can also hear the vertical directivity mismatch between the Neo3 and Neo10, though it's minor with my 1.7kHz cross. Increased vertical directivity would be helpful for both issues, as would being able to use a driver smaller than the 18. I don't have a sense of how much horn loading might reduce the Neo10's 150Hz Fs but, unless you like to listen really loud or have funky room modes which need filling, I would not expect a single 15 in a horn to require supporting subs.

-110dBV noise into drivers like your horn loaded Neos is around -20dB SPL 1W/m. If your ears can hear that, I will be impressed. I don't look at HTGuide all that regularly unless there's a thread I'm actively involved in so feel free to PM me.

The family hot-cross-buns (i.e. easter buns) which she translated from "about this much" into actual measurable amounts, however, is just about un-touchable.

Excellent! My great-grandmother was such a terrible cook my grandmother took over from her at the age of 12.

[cjd]

-110dBV noise into drivers like your horn loaded Neos is around -20dB SPL 1W/m. If your ears can hear that, I will be impressed. I don't look at HTGuide all that regularly unless there's a thread I'm actively involved in so feel free to PM me.

-20dB is quite audible - not necessarily directly, but the impact on the resulting sound, very much so. When a recording has a 30-40dB dynamic range, even if that -20dB is around the center point of the range, it's as loud as the quietest parts of a recording. I usually find it's slightly louder, so I hear it, and not silence, on quiet passages. Room cues (live and on recordings) are often -50dB or more (less?) and I pick up on those. I blame being classically trained, not to mention a few other things. Some kinds of music it's not really audible. Lots of DA transitions also (seem to) become audible - again, not directly, but in the tendancy to wear the listener out - sort-of like a study I read recently suggesting that low bit-rate MP3 induces sleepiness.

My PM box fills up way too fast, but I'm watching your thread over on DIYA. This is likely to be a project over quite some time, as are so many of my projects.

I've been pondering ways to avoid H baffle construction and still work. Among them, wrapping things in AT material so it looks solid, but isn't. But I wonder, at what point does an H baffle turn into a very very small waveguide? Can it? This is just starting to get interesting.

[twest820]

I wonder, at what point does an H baffle turn into a very very small waveguide? Can it?

I inadvertently ignited some controversy over at DIYA when I pointed out if you simulate even fairly shallow H baffles in Hornresp the wavefront launch is obviously less clean than more complex wall shapes. Hornresp is assuming a plane wave source which isn't particularly accurate but there's enough there I think some investigation with AxiDriver or similar tools is in order. I'm not likely to get to that any time soon---there's a lot still to do with my current build---but there's no reason someone else can't take a look. My suspicion so far is the reason H and U baffles work is they're nearly always used on subs and the audibility of things below 200Hz or so is not so good and measurement not so easy. So I've been thinking along the lines of putting a Neo3 in different baffle and waveguide structures as the scale's smaller, easier to work with, and getting decent data at 1+kHz doesn't take much fuss.

AT material?

My PM situation's not too bad, so no worries there.

-20dB is quite audible

I agree -50dB in relative terms is a good rule of thumb for the limits of audibility but I think there's some confusion over units. 0dB SPL W/m is, for most frequencies, lower than the quietest audible sound. While I'm sure there's listener to listener variation with some folks doing better I would be surprised if there are many people who can hear -20dB SPL W/m, even around 3kHz. Horn efficiencies and moderate playback volumes still provide around 70dB SnR over a -110dBV noise floor in the electronics.

As an aside, I live in a quiet neighborhood and even with the furnace off, the fridge unplugged, and late night or early morning times with little wind my acoustic noise floor's something like 40dB above the electronic floor at high frequencies, rising to 0dB around 20Hz. So I'd bet in most situations at moderate listening levels the audibility of details -40 to -60dB below signal is due to the ears' ability to discriminate sources based on direction. My background's more bluegrass, blues, and classic rock (albeit with a certain amount of speed metal), but I find as someone who plays an instrument and is hence constantly monitoring fretwork and tone I'm a more critical listener than folks who've invested less effort in developing their musicianship. Like so many things, what you get out of music tends to be proportional to what you put into it.

[cjd]

Whoops. Yeah, that -20dB, if right, would be harder to hear.

I hear noise in the quiet passages - how's that for enough data to go on? That suggests to me that all is not as it should be, thus my statement that I've had issues with noise.

Not that the circuits, done right, would necessarily have issues with noise.

Unless a cylindrical waveguide were used on a large driver in place of an H baffle, there are still launch irregularities those graphs end up getting wrong. But, interesting none the less.

On the other hand, a large gear-head lathe with the ability to swing a 24" square could be a fun addition to the workshop I don't have. Probably an expensive addition, too. But then I could spin, and that's a ton of fun. 1/8" T6 aluminum horn, anyone

[twest820]

Unless a cylindrical waveguide were used on a large driver in place of an H baffle, there are still launch irregularities those graphs end up getting wrong.

Yep; part of the reason why I looked at a couple different H baffle widths. mige0 suggests asking Jean Michel directly about the profile on his thread but this should give you some ideas on dipole horns.

I hear noise in the quiet passages - how's that for enough data to go on?

A step in the right direction. Have you taken a look with a spec an or RTA to characterize the SnR and noise structure? Your mention of ADC/DAC fatigue makes me curious about quantization issues---the parts may be speced for -100dB THD or better at full swing but it's easy to end up operating in the -50dB range with typical single ended power amp gains of 25-30dB even when trying to maximize the bit depth. Moving the amp to 0dB gain helps with that and 0dB gain, differential input still provides 20W RMS into 8 ohms per channel with excellent linearity in the op amps. Enough power to make a decent ruckus in a triamp. The increase in ground noise rejection from 0dB to 40-60dB doesn't hurt either.

a large gear-head lathe with the ability to swing a 24" square could be a fun addition to the workshop I don't have

[Saurav]

I have a friend with a CNC. Now you've got me thinking about a translam construction for that dipole ribbon horn...

[cjd]

I've done no measurements to try to figure out where issues I hear might exist in the gear I've put together. Right now I don't listen in my living room as much as I'd like because the power supply hum (toroids) are too loud. They were too loud when they were right next to the speakers, and at the moment they're right next to me (no balanced pre at the moment so long speaker level runs).

I have a friend with a CNC. Now you've got me thinking about a translam construction for that dipole ribbon horn...

For that kind of horn, you just need a jigsaw and a belt-sander.

Yes, I know. It's that much more accessible to you now!

On the other hand, having someone else CNC the stuff sure is a lot less work for you.

[CraigJ]

Twest,

Are you talking about this horn:



cjd, FWIW, I have a new Neo3-pdrw in a shallow waveguide (like Nao Note) that has a similar dip at 10khz.

Cj