Nebbiolo: Another 3-way experiment

[cjd]

Listening to just one side of a stereo track sucks.

That said, the midrange character is clear (and I'm not sure I like it), the tweeter remains incredibly smooth, and no mid at all sounds almost good enough! (-10dB through the middle...) I had a spade connection get in a weird way and not actually connect on the mid. Rock'n'roll I couldn't tell. Goes to show you... I may see what it takes to push the woofer crossover lower (it's rather high for a 3-way but well within the comfort zone of the Seas)

I'm not 100% satisfied, but I think I need a stereo pair up to do more in-depth analytical listening. It may just be the right channel only ndrom leaving me less than happy.

[Carl V]

Listening to just one side of a stereo track sucks.

Perhaps you have some nice MONO recordings.

I usually like to take the "stereo artifacts" off the table
and focus on Tone & timbre...and the sheer magic imho, ymmv

http://inkpot.com/classical/heifetzstg.html

http://www.amazon.co.uk/Beethoven-Mendelssohn-Violin-Concertos/dp/B001I1VFAE/

[cjd]

Perhaps you have some nice MONO recordings.

I usually like to take the "stereo artifacts" off the table
and focus on Tone & timbre...and the sheer magic imho, ymmv

http://inkpot.com/classical/heifetzstg.html

http://www.amazon.co.uk/Beethoven-Me...dp/B001I1VFAE/

No good mono tracks, I thought I did. I have an LP but that's not helpful, the TT needs to be rebuilt (caps are toast, possible other issues but those may go away once the power supply is fixed and things are run for a bit.) That said, equalizing SPL by fiddling with distance has gotten me a stereo "pair" that sounds close enough to where I want to go that I think diving in on the second one is the right way to go - things are likely very close, and even the drivers probably need more play time.

[JonMarsh]

Sounds like you're getting there... measurements?

[cjd]

No measurements today, and I'm going to wait till I'm confident the calibration is right on my rig - I suspect something may be off there. That should be possible later this week I expect. I spent a lot of my weekend engineering the crossovers... I didn't leave much room at all!

[Juhazi]

I have been investigating a similar system, but with 8" woofers. Have you considered the lower xo, better at around 500Hz perhaps? Benefits - bsc, power response/vertical dispersion.

Other issue with xo slopes, try to get good phase match with W and T, it is very important, don't you agree Jon? Then just fit the mid there with symmetric slopes and it kind of drops in! If phases don't match you will get strange undulation in spl-response and a weird sound!

[cjd]

I experimented with modeling between 300 and 1k, and while I may well continue tweaking this setup has a lot going for it. Measurements will tell a lot when I get to doing those in an assembled state, but the listening suggests a scary wide sweet spot with no clear issues. Woofer and T aren't quite as good as I'd like yet so still pushing things around. That said, it's as good as some very well received designs.. and the mid does basically just drop in and make that missing midrange magically materialize...

Anyhow, not sure there is anyBSC benefit of 500, power response and vertical are well within the comfort zone of the CA18RNX and overall layout. Using 8" drivers would clearly require a lower upper limit. Not a loot perhaps but... Anyhow, nothing new here and I'm usually a stickler for some of these details, I doubt that's going to change.

[---k---]

If these end up being gifts, you need to let me know when they are leaving your house so we can schedule some listening time before hand. I would love to hear them.

BTW, the finish looks great. I like the detail on port hole.

[cjd]

Got some data, but the Win8 system seems really unstable with the sound card - it ended in silence and no saved measurements, and I decided to give up for the day rather than snag my old laptop (win7 - it works but is starting to get unstable - it pauses every ~10 seconds when I'm typing now, for example.) Hopefully the M-Audio Fast Track Pro didn't get toasted, but it's a distinct possibility.

Anyhow, with all the calibrations double-checked so I'm confident the mic is good, it looks to be about 4dB of BSC and pretty much dang flat all the way out. However, there's some reinforcement on the low end that could end up being too much bloom if I go for a full 6dB, I really need to get nearfield and do a little more detailed modeling to be sure what's going on. On the RTA scale it looks very nicely flat plus room issues. There is probably room to bring the mid down a little more, and I may still fiddle with more BSC and some alternative crossover points.

C

[JonMarsh]

Well, it sounds like you're making progress, but hardware issues are making the work twice as hard as it ought to be! It's funny to hear you talk about "old laptop Win7" system, as I'm still buying Win7 to run on macs (these days in Parallels 9); I still think of Windows 7 as being fairly new... probably says something about my age! Of course, I still have a 27" iMac at GF's with the original OS which she won't let me upgrade because it supports Rosetta, means it will run PowerPC based programs on an Intel processor, and she's using several now... that's OSX 10.5, and I'm running 10.9 on all but one of my Macs, which I just haven't gotten around to upgrading (I was keeping that one in 10.8 Mountain Lion because Epson hadn't gotten out a working printer driver for 10.9 until a couple of months ago).

Well, you sound like you work the same way I do- start off with what you think is a good idea or plan, then iterate, iterate, iterate, until you're happy with the results! :B

[---k---]

Windows 7 still seems rather recent to me too...

[cjd]

Crossover needs tweaking - a bit forward in the mids, and follow-up measurements confirm this. For a lot of music it's still incredibly listenable, if a bit dead sounding.

My finishing skills need work.

[Paulie]

They look awesome.

[JonMarsh]

Crossover needs tweaking - a bit forward in the mids, and follow-up measurements confirm this. For a lot of music it's still incredibly listenable, if a bit dead sounding.

My finishing skills need work.

Your finishing skills look pretty dang fine from where I'm sitting! :T :B

[JonMarsh]

I think this is where I'll start on a real crossover build and tweak from here. Roughly LR2, appears to be 950Hz and 3kHz where things seemed to settle in. I may try to push the mid-woofer lower before jumping in, we'll see. I'll definitely be ridiculed in some circles for the parts count Lots of impedance compensation in the mix.

On-axis measurements and response, with mid reverse null


15 degrees off-axis horizontally (modeled response on real off-axis measurements)


30 degrees off-axis horizontally (modeled response on real off-axis measurements)


Nulls from woofer or tweeter reversed alone


Impedance

So what is the parts count looking like, as well as the network? The curves here look a lot like what a typical Duelund approach with moderate aleph coefficients works to be - the mid range transfer function is a lot like an LR2 bandpass, and with typical alignments there is substantial mid and tweeter output with the mid padded down as you have. Works well IF the drivers are really clean for a wide band response- but often it requires a fair amount of shaping beyond the nominal network transfer function to get the drivers to behave (witness the original Ardent three way). SO, I doubt you'll get any criticism from me for component count... :W

[cjd]

So what is the parts count looking like, as well as the network? The curves here look a lot like what a typical Duelund approach with moderate aleph coefficients works to be - the mid range transfer function is a lot like an LR2 bandpass, and with typical alignments there is substantial mid and tweeter output with the mid padded down as you have. Works well IF the drivers are really clean for a wide band response- but often it requires a fair amount of shaping beyond the nominal network transfer function to get the drivers to behave (witness the original Ardent three way). SO, I doubt you'll get any criticism from me for component count... :W

I chose parts with the hope I might hit a duelund-ish approach, and I've toyed with those worksheets, so it's not really surprising it looks similar.

With the bracing the way I did (and no special crossover compartment) I had to get creative on the mid crossover. The mid inductors line up properly when back to back...



It's higher than most people consider acceptable, and will probably climb. It's been challenging, though interesting. The angle is perfect for the mid/tweet z-axis, and both drivers are very wideband (I probably could have crossed the Hiq to the CA18's directly), but it's phase that's a bugger. Really just a LC (3.5mH/.2uF) + zobel on the woofer, 2nd order electric everywhere else + impedance shaping. I may add a cap across the woofer out of band (really just to further deal with the rougher spot on the response ~2kHz) and will need a bit more twiddling to get the mid padded down more.

I left off a zobel on the tweeter - I think it's needed after all, but the impedance peak has been dealt with in the above graphs.

The mid needs some attention, and I'm still fiddling. It's very picky - each change requires adjustments all around. It feels like I'm still missing something. I did find that the impedance on the woofer is in range to muck with response shaping, which is unfortunate. The higher (frequency) peak takes a 15mH/500uF/?r filter to flatten, and I didn't try the other peak.

It will be interesting to try this with the Anarchy/Vifa NE95W-04/Vifa NE19VTT-04

[dar47]

I'm a sucker for cherry too. Finish looks great.:T

Love the narrow baffle and those sweet little OW1's. They look to fit great in most rooms and the next friend looking for free cab build is going to get these suggested. I would love to try and build these.

[---k---]

Looks fantastic from here. :T

[cjd]

Ahhhhhhh...

I had to tear in to make some very subtle changes in the crossover - miserable work. I took a drive in my latest frequency-related distraction (ran out of zip ties!) in the middle just for a bit of time away from the soldering work...

Not 100% sure it's there yet, but I think it's a lot closer... though I didn't want to tear things down because it was still sounding good before! Excepting the occasional bit of trouble...

Maybe close. Or there. or ... well, they'll be going to InDIYana in this trim, regardless.

C

-- in case you were wondering, suspension dynamics happen to look an awful lot like crossover work!

[JonP]

Your finishing skills look pretty dang fine from where I'm sitting! :T :B

Very nice, Chris! They have a balanced proportion to them... Turned out quite a handsome pair of speakers.

Speaking of where you're sitting, is my memory correct that you're the originator of the "sock shot", which is the defining indicator of a finished design? ;-)

[cjd]

So you can't see some less than perfect finish details - swirls, a couple drips that happened somewhere, and a couple bubbles that eventually showed up in the veneer - and I have no idea how to fix! In general they look great, especially if I'm sitting back listening to music with my eyes closed.

As far as I know, I'm the original sock-shot guy.

[dar47]

If those bubbles start to really bother you and it's happen to me months after finish, you can try something that feels evasive but has worked for me,

1. Slice through the finish and veneer with a flecto knife with the wood grain the whole length of the bubble to release air and tension.
2. Covered the area with thin cloth and heat with a hot iron or you can blast the bubble with a heat gun in a pinch. the finish may melt a little.
3. Apply at lot of pressure with a veneer scraper or a wood block to get it to lay down.
4. Do some spot repair of the finish after it's all hardened.

No rush maybe when you want the touch up those swirls and drips. I have done this with a natural finish on cherry and after the finish repair the slice disappears.

[JonP]

Something that could work for runs or drips, would be to use a card scraper to shave just the high spots level, something that they're rather good at. I was impressed with how well it worked to smooth out layers when I did an interminable Shellac finish. It even let me shave orange peel with great control, when I got too greedy on layer thickness. You might even be able to shave a swirl out, depending on thickness of finish.

[cjd]

Whee. So, I think the swirls, etc. are in the base shellac - not in the top layer Tung oil/poly...

And the bubbles are on the top bevel - so I'd be looking at refinishing. They had popped early, and I already tried the heat method. Pretty sure this is on a spot with heav(ier) bondo.

I may some day take the scraper to these, work back to the shellac and start fresh (proper lighting is obviously crucial - I thought it was adequate, but clearly not even close.)

Oh well. They're sounding pretty good now, and I can't see the problems from my chair.

[JonP]

That's the main thing!

[JonMarsh]

Whee. So, I think the swirls, etc. are in the base shellac - not in the top layer Tung oil/poly...

And the bubbles are on the top bevel - so I'd be looking at refinishing. They had popped early, and I already tried the heat method. Pretty sure this is on a spot with heav(ier) bondo.

I may some day take the scraper to these, work back to the shellac and start fresh (proper lighting is obviously crucial - I thought it was adequate, but clearly not even close.)

Oh well. They're sounding pretty good now, and I can't see the problems from my chair.

My girl friend's former husband used to say, "Anything you can't see from horseback is not worth worrying about..."

Sometimes we just have to keep that in mind, as opposed to going all OCD about little details. Kick back and enjoy the music!

One thing I still appreciate in retrospect is how well the decision with the Isiris went to build them entirely in LBL, not have any veneer, and the finishing was pretty dang fast and easy, though not quite as classy looking as the Ardents. I'm seriously starting to think about stripping them down and shooting them in lacquer, even though the wipe on poly looks pretty good- I suspect the lacquer will look and feel even better.

Yours look great in pics- if they sound like you describe, I'd expect you should be pretty happy at this point!

[cjd]

Thanks for all the compliments, guys.

For a first time veneering, I sure learned a lot!

One of those things was "it sure is nice to not veneer and have a finish you're happy with!" - LBL would definitely do that.

We'll see how they show at InDIYana this weekend.

[---k---]

I think they look great. Good luck in Indiana.

[cjd]

The crowd was pretty thin by the time these went up - only one design after, which had been played Friday evening already (mine hadn't)

The bad: not as many people got to hear them. The good: Folks brought up some of their own demo music, so they got some extended playtime.

Feedback has been great, and getting these out in the room really let them play.

They also got quite a run (4-5 hours?) in the after-event where it was finally identified that one of the mids is bad! Picks up a nifty buzz at ~100-120Hz. It can go for a long time before something tickles it just right... then it's gone. I'd heard it before and I didn't pay enough attention or try to track it down. But now I know the value of running sweeps on all my drivers (and probably the finished product.)

They still sound great. Swapped out the Ansonica and took a nap listening to a random mix - ran across some tracks with really precise placement of sounds, and of course I have no idea which tracks they are!

Gotta swap out that mid and publish the final crossover.

[JonMarsh]

They still sound great. Swapped out the Ansonica and took a nap listening to a random mix - ran across some tracks with really precise placement of sounds, and of course I have no idea which tracks they are!

Gotta swap out that mid and publish the final crossover.

Sounds like a good plan! Getting a pair of speakers ready AND perfect for a DIY meet always seems to be a challenge... now that the Northern CA meet is put off to the fall or later, I may have a chance... :W

[dlneubec]

The crowd was pretty thin by the time these went up - only one design after, which had been played Friday evening already (mine hadn't)

The bad: not as many people got to hear them. The good: Folks brought up some of their own demo music, so they got some extended playtime.

Feedback has been great, and getting these out in the room really let them play.

They also got quite a run (4-5 hours?) in the after-event where it was finally identified that one of the mids is bad! Picks up a nifty buzz at ~100-120Hz. It can go for a long time before something tickles it just right... then it's gone. I'd heard it before and I didn't pay enough attention or try to track it down. But now I know the value of running sweeps on all my drivers (and probably the finished product.)

They still sound great. Swapped out the Ansonica and took a nap listening to a random mix - ran across some tracks with really precise placement of sounds, and of course I have no idea which tracks they are!

Gotta swap out that mid and publish the final crossover.

I heard them, though I missed a few cuts as I was trying to get my stuff out and loaded in the truck for the drive home. They are excellent. Anyone would be very happy to have them in their home. Lots of sound in a very small package. I like that aspect as well. I hope you are able to work those little bubbles out of the veneer. I know they were bothering you, though I doubt anyone would have noticed without them being pointed out.

Excellent work, IMO, FWIW. Too bad they did not get up sooner, when a few more were present.

[cjd]

I heard them, though I missed a few cuts as I was trying to get my stuff out and loaded in the truck for the drive home. They are excellent. Anyone would be very happy to have them in their home. Lots of sound in a very small package. I like that aspect as well. I hope you are able to work those little bubbles out of the veneer. I know they were bothering you, though I doubt anyone would have noticed without them being pointed out.

Excellent work, IMO, FWIW. Too bad they did not get up sooner, when a few more were present.

Thanks! I think among the 3-ways, yours, Dave P's, and mine stood out for not standing out. I didn't expect the Nebbiolo to sound as good as it did - and they keep surprising me. I think they like a slightly live room.

The place really emptied out after the door prizes, which surprised me a little. It certainly got noisy!

I'll work on the bubbles, leaving that at a "maybe someday" - I think I'll set up a practice piece to try some things out and make a decision whether to mess with it after I see results (or not!)

Congrats on those RAAL tweets! I'm jealous, though I have plans for the Dayton tweets I won so it'll save me a few bucks in the end, where the RAAL would simply have been another un-planned project (and I have enough planned!)

[JonMarsh]

One methods that works pretty well with bubbles is to use a needle glue applicator to pierce the bubble and release the air, then follow up in the same small hole with some adhesive (superglue is fast and works well...) if you can find a dark element in your wood grain, that may help hide the puncture mark.

[5th element]

One methods that works pretty well with bubbles is to use a needle glue applicator to pierce the bubble and release the air, then follow up in the same small hole with some adhesive (superglue is fast and works well...) if you can find a dark element in your wood grain, that may help hide the puncture mark.

I've done this and it works rather well, certainly not perfect, but if you can get the bubble down it's far less visible, in the right light, than the bubble would be.

[cjd]

Finally posted up the final response/impedance and the schematic for these.

I had already forgotten about the veneer - it's in the "dark" corner of the room and doesn't get light to cause it to show...

I should also say, these are now my go-to for sitting back and listening. Simple gainclone, passive attenuator, and a Squeezebox II.

The bass response is unexpectedly awesome - they dig down without giving anything up. And the OW1 ... I think that's why I like these so very much. Unbelievably excellent tweeter. What finally brought it back was the zobel, which allowed me to use a smaller cap (er, value... I went for one of the big red Jantzen caps, and it's huge!)

And, while the soundstage is always very very good, its best is not always there (I think the position of my head relative to the head-rest on my chair matters) I was scared the other day listening to something or over when there was a consistent noise coming from my fireplace - which is pretty much dircetly to my left. I realized it was in the recording when I sat up a bit more to look and see what was making the noise, and it popped to a spot further up...

Whee. If I exclude padding resistors, half the crossover is impedance compensation.





Also, note that it never dips below 4 ohms. Something is wrong with me, for sure.

[5th element]

I do find this design interesting as it uses such a high xover point to the mid range and pretty low xover (comparatively speaking) to the tweeter. There are lots of people who'd simply cross the OW1 straight to the SEAS mid/bass units at that frequency and think they'd hit audio nirvana. Of course using the 6.5"s up that high is a recipe for disaster. Of course this design, if listened to, would hopefully highlight why. The off axis response of these should be pretty dang excellent.

Looking at the cabinets again I think they look excellent. I especially like the way you've done the port.

I think it would be interesting to try different drivers in this type of configuration, but perhaps trying for something even more compact and trying to squeeze more out of the design concept, just to see if it makes any appreciable difference. Something like the SEAS L16 on the bass, a Vifa OT19 as the tweet and whatever floats your boat as the mid. A scan 10F springs to mind.

What'd be interesting with that design would be what sounds best RE the xover points. You could go for a 350Hz/5k xover type design, maximising the bandwidth that the mid covers, or go for an 800Hz, 2.5kHz design, where you'd be more using the mid as a 'filler'.

2 ways with 5-7" mid bass drivers and 1" tweeters are plentiful, yet it surprises me that there aren't more designs using the same concept, but plopping a small mid in between them.

[cjd]

I do plan to attempt an all metal variant - I have some Anarchy in hand, and a couple 3" full-range (Vifa with inverted surround, Fountek) and one of the new 3/4" Vifa domes to try - we'll see if I stick with that set of drivers or look around more. The box is unlikely to be like this, though I'll be trying for time alignment still. The secondary goal this time is to make it easy for anyone to build - plain rectangle boxes - and still see most of the benefits. This may hold some potential to build a horizontal center, which is one of the main reasons I agree whole-heartedly with you on the driver setup - useful centers with reasonable off-axis and 2-ways are a tough mix. As for compact - I think one of the most subtly damaging things you can do is undersize your box.

The thing with the crossover points - I certainly started lower, and ran into a few issues (obvious in hindsight.) First, to maintain 2nd order LR slopes, I started losing net sensitivity (or so it seemed). To make this worse, both the mid and the woofers ran into more obvious troubles with the Fs impedance bump (though trouble may be relative, I would have had to quash it on the mid I think.) I also never got the crossover to work as well. I didn't want to go much higher on the tweeter to keep C2C in line - plus, why waste such a superb tweeter?

Every single 3-way I've done makes me wonder why people bother with two-ways. Then I go to a DIY event and once again understand. 2-ways, none are bad any more. There is still a clear range - I heard a TM and an MTM with the same drivers, and the difference was significant in sound - especially spatial representation. But the 3-ways are less consistently good. They're harder to accurately sim (no real measured data) and a lot of people still work that way exclusively. It also more than doubles cost in most cases, not to mention what you need to pay attention to in the crossover.

I've also got a spin-off from the Ocho series planned - house-brand tweeter (I won a pair of the silk domes at InDIYana) and probably pushing a 3-way option in addition to some two-ways. Budget will remain key. That may be where you see me focus on compact a bit more - maximize SAF in addition to the budget. The inconsistencies we've observed with the tweeter used by the original Ocho leaves me unhappy. This one will again use existing available PE boxes, or be optimized for flat-pack production (if I can get the necessary info...)

We'll ignore the Marzemino in the wings, slowly taking shape. And the other two designs with all the parts on hand - one with an AMT+Neo8 WWWMTW and one TMWW with a waveguide (or MTW, we'll see)... I'm gonna have a fuuuun move once we find a place, but I will gain a permanent workshop out of the deal (not to mention a much shorter commute - down an hour each way, so 2hr/day) Also, I think we be likely to gain a few rooms so will NEED more speakers. I'll also have to build a new HT (still on the fence whether the current HT will help the house sell.) I have to work on something - need to use up the plywood I have!

So much to do, so little time.

[JonMarsh]

These comments are all very true, and mirror my experience over the years- doing a good three way takes much more effort to balance and implement with all the potential issues, and this shows with commercial speakers, too. The only "easy" three way is a two way with a sub!

[5th element]

If you're going for an all metal three way, providing you don't need the mid to be that sensitive (like in this case with the bandpass gain element) there's the SB65WBAC25-4 wideband driver that I recently measured over in the drivers wanted forums. It seems to hold up very well and apart from some rising 2nd harmonic at higher output levels, all other distortion products are very low.

[cjd]

The SB65WBAC25-4 came up in my research, but I haven't run across a source here advertising it. I could drop Madisound a note, but I want the drivers to be reasonably easy to source for folks that might want to build the result. Just in case I build something worth building again...

[5th element]

I think availability is purely down to the drivers being quite new. You can buy them from intertechnic in Europe, which is where I placed the order for mine from, they should have the other three I want in 16 days. Solen in Canada has stock, but not much.

Ever since SB released info on them I thought they'd be perfect for a couple of projects so I've been keeping tabs on them and literally nowhere was stocking them except Solen and intertechnic. Both had them listed quite early on, but didn't have any stock, then slowly intertechnic got a few and then Solen did. I think there's possibly a stock shortage somewhere.

If you do contact Madisound it will be interesting to see what they have to say as they might have more knowledge on the issue.

[---k---]

We need to get this in the Completed Speakers thread.

Fill this out, and upload a picture to the htguide server and I'll get it done.
Blank Template
Design:
Price:
Size:
Tweeter:
Mid:
Woofer:
Designer:
Other Design Notes:
Related Designs:
Comments: