the three way nude swinging dipole thread - diyAudio
Go Back   Home > Forums > Loudspeakers > Multi-Way

Multi-Way Conventional loudspeakers with crossovers

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 21st October 2010, 02:50 AM   #1
diyAudio Member
 
Join Date: Jun 2009
Default the three way nude swinging dipole thread

Because, after all, it would be a shame to let one of the best unintended double entendres in audio go to waste. I was tempted to title this the ménage à trois thread, but that seemed a bit over the top.

I know of two and a half way swinging dipoles (mige0), a few minimally baffled three way dipoles (notably cuibono and keyser's builds), and StigErik's four way technically has got a three way swinging dipole embedded within but doesn't really count. I don't know of any three way swingers and so a new thread on them seems in order. All build threads broaden into a general discussion of various speaker designs sooner or later so I figured I'd just scope the thread to the category and kick it off with my current build.

The build in question is a BG Radia Neo3, BG Radia Neo10, Selenium 18SWS1100 crossed linear phase LR6 at 200Hz and 1.7kHz. Crossover and equalization are via a Focusrite Saffire 40 and Thuneau Allocator on my Dell Inspiron 1505 Core Solo laptop running Windows XP. The first couple figures are a shot of the right speaker to give you a sense of the build and how the swinging is done along with a detail of the lines on the left speaker. The next two are front and rear wave symmetry for SPL and phase, followed by front and rear directivity. All four measurement sets are in room, ungated, and 1/16 octave smoothed. The smoothing's mainly so it's easier to tell which curve is which and the chuckhole around 150Hz is the usual first reflection from floor, ceiling, and walls (mostly ceiling as I'm renting and hence haven't mounted a diffuser). The next three figures are SPL and THD of the individual drivers when equalized flat for nude dipole operation within their passbands. SPL on top, THD below, and the crossovers turned off. These are in room, ungated, and unsmoothed. SPL for the green 0dB curves is 5 to 10dB above peak SPL at my usual listening levels. I don't have an SPL meter but the measurements are around 80dB or so, give or take a few dB. The red +10dB curves are peak SPL at about the loudest level I ever listen at---it's somewhat uncomfortable to listen while the measurements are being taken---and the blue +20dB curves are loud enough I give up, put in my 35dB chainsaw earplugs, and wonder why I didn't put the earplugs in already for the +10dB curves. Granted, the +20dB setting's quieter than my chainsaw. But not by a whole lot.

DIYAudio only allows 10 attachments per post, so see the next post for the crossover, equalization, and a couple other interesting things. In the meantime, some discussion of the speakers seems in order.

In short, I like them. Quite a bit. Best setup I've heard yet. But then I'm 1) obviously biased since these things are sitting in my living room making dents in the carpet, 2) unable to afford a plane ticket to Norway to listen to StigErik's rig since I blew all my money on drivers, and 3) too polite to show up on cuibono's doorstep and invite myself in. There's also the minor logistical difficulty I don't actually know where cuibono's doorstep is, but that's beside the point. I believe the measurements mostly speak for themselves, but a few remarks seem in order. One is the front to rear SPL asymmetry above 10kHz due to the Neo3 shifting somewhat off vertical when I was turning the speaker around to measure the rear wave and I'm too lazy to redo the measurements. The asymmetry below 1kHz is due to slightly different mic and speaker placement and room interaction along with, possibly, some shadowing by the 18SWS1100 magnet at low frequencies where the diffraction radius is large. Another remark is the 0dB THD curves are higher than the +10dB THD curves because they're sitting at the ambient noise floor of my relatively quiet neighborhood on a quiet night---the floor would be higher if I'd left the fridge on. THD doesn't actually drop in the +10dB curves; it's that turning up the volume raises the SnR. The exception to this is the Neo10, where the +10dB curve becomes excursion limited below 400Hz. At +20dB all drivers are showing rising distortion from excursion. But they're also pushing 100dB SPL each and if all six of them that ran hard the system would be around 105dB. Respectable performance for not having a baffle.

The Neo10s make this system. I originally started out designing a four way with five drivers---the fifth being a second sub---which eventually shrunk to four drivers as my preference for keeping speakers vertically compact asserted itself and I realized a single big driver was shorter than two smaller ones. I eventually chose the 18SWS1100s for their good bang for the buck, generous xmax, and low (for an 18) Mms and Rms. Neo3s were a given for the tweeter as AMTs are expensive, Zaph's shown the Neo3 handily beats most ribbons, and Rudolf's amply demonstrated getting good directivity out of back to back dome tweeters is hard. But no matter how I looked at it I could not find a mid for a three way which didn't involve more sacrifices than I was willing to accept. Until BG Radia finally made Neo10s available for DIY and Meniscus got them back in stock. Greater vertical dispersion would benefit the Neo10s the Neo3s cross low enough to mitigate the problem. Otherwise the Neo10s are to die for. Excellent front to rear wave symmetry, low distortion, detail as good as you'd expect from their low Mms, and no cone breakup. Spendy. Though cheaper and more compact than two mids and the extra pair of power amp channels needed by an actively equalized four way.

I'm also pleased with the hanging. The mechanical benefits of swinging drivers are covered at length in StigErik's Open baffle with Beyma TPL150 build thread (more accurately titled the no baffle without TPL150) and mige0's Beautiful Swinging Speaker thread. I've little to add here but agreement and it doesn't hurt swinging saved me a bunch of MDF dust and spending two hundred bucks on a router. Good deal. mige0 used a cable and bolt system for hanging and StigErik progressed from bolts and chain to bolts and rubber straps. I'm not a fan of either approach, but my previous solution of picture wire was subject to creep and I had to keep readjusting the drivers. So I picked up 2mm, 3mm, and 6mm climber's auxiliary cord and set to with bolins and rolling hitches. Way better. The frames are obviously prototypes and I've not trimmed the lines yet but the reason I included a detail of the hanging is so you can see how the Neo3 is supported with a pair of lines, one for the driver itself and one to support the cable so that it doesn't pull the Neo3 around. The Neo10 is light enough to be subject to same problem and the combination of using parallel lines to get a pantograph type of motion in the swing works well for it too.
Attached Images
File Type: jpg LokiRight.jpg (115.1 KB, 4187 views)
File Type: jpg LokiHang.jpg (66.6 KB, 3832 views)
File Type: png LokiFrontRearSymmetrySPL.png (25.3 KB, 3738 views)
File Type: png LokiFrontRearSymmetryPhase.png (27.9 KB, 3608 views)
File Type: png LokiFrontDirectivity.png (26.6 KB, 3536 views)
File Type: png LokiRearDirectivity.png (26.5 KB, 555 views)
File Type: png Neo3-THD.png (30.1 KB, 582 views)
File Type: png Neo10-THD.png (32.0 KB, 666 views)
File Type: png 18SWS1100-THD.png (33.0 KB, 656 views)

Last edited by twest820; 21st October 2010 at 03:11 AM.
  Reply With Quote
Old 21st October 2010, 02:57 AM   #2
CLS is offline CLS  Taiwan
diyAudio Member
 
Join Date: Jun 2005
Location: Taiwan
Where're the attachments? I mean the photots.
  Reply With Quote
Old 21st October 2010, 03:44 AM   #3
diyAudio Member
 
Join Date: Jun 2009
The next batch of figures is for the crossover and equalization. The first is the overall VST setup in Bidule followed by a screenshot from Allocator which shows the equalization for all three drivers, the crossovers, and the arbitration to get the crossovers to linear phase. As has been my usual experience, using warp phased equalization for the minor driver wobbles produces flatter phase than using linear phase equalization. However (as has been discussed over in the Arbitrator forum at Thuneau) dipole rolloff does not induce a phase shift and wants linear phase phase equalization. Additionally, drivers under a phase shift around Fs which wants correction. My current solution is to use a peak/dip Q = 0.5 biquad centered at Fs and then correct for both the biquad and Fs phase shifts with a fourth order Arbitrator dedicated to that particular driver. Hence the additional three Arbitrator VSTs for the Neo3s, Neo10s, and 18SWS1100s. They operate at 575, 150, and 39Hz, respectively, though as the yellow line in the Allocator screen shot shows I've used a soft dipole equalization on the Neo3s to avoid excessive excursion. The difference in levels between low, mid, and high is minimal as the 18SWS1100 and Neo10 are a good bit more efficient than the Neo3 in dipole operation. However, there's still a goodly amount going on to get the output SPL flat. I've also EQed the room interaction at 150Hz so it measures flat but so far it's sounding better to me without EQ. That's about what I'd expect as the 18SWS1100 is pretty flat between 100 and 200Hz and reflections within the first 3ms are minimal.

Preringing is an issue which arises with all linear phase methods---time reversed IIR (what Allocator uses), FFT, and FIR---so I've included the impulse response as well. I'm at a loss as to why HOLMImpulse shows more preringing 60 degrees off axis, but I find I can't hear the preringing shown here. Seems unsurprsing to me as it's 30+dB below the causal impulse response and short enough to fall well within the 2-3ms window within which events are percieved as simultaneous.

I hope others find this data useful. While there was pain in selecting a midrange most of the unknowns were in how much sub was needed to get reasonable SPL at reasonable THD and hence some guesswork was involved in selecting the 18SWS1100s. Personally I tend to listen somewhat softly so I might be able to get away with Selenium 15PW5s or 18WS600s but I wouldn't want much less driver than the 18SWS1100s for nude bass. At the risk of stating the obvious, a baffle to pull the dipole peak on the low channel down to 200 or 250Hz would be more efficient and the resulting decrease in excursion would reduce distortion. However, I was more interested in what it'd take to get nude bass on a scale more modest than StigErik's 117dB at 16Hz and the 18SWS1100 seems sufficient for most purposes. I've personally chosen to EQ them flat to 35Hz for no other reasons than 1) it's low enough for me, 2) happens to be what falls out naturally from the biquad at Fs method described above, and 3) I've not hit problems with hearing harmonics instead of fundamentals despite the boost low frequency distortion gets from dipole EQ. With digital it's trivial to extend the equalization lower in frequency and so my last figure is of the 18SWS1100 running nude and flat down to 20Hz. Curves from top to bottom are SPL, phase, and THD. Distortion rises significantly so I suggest multitone IMD testing to see how badly deep bass harmonics bleed upwards but, depending on your criteria, it may be feasible to extend nude operation below 35Hz.

Full disclosure: Thuneau supported my investigations into this VST topology with a donation of Allocator Lite and Arbitrator so I've chosen to combine the full Allocator instance I purchased with the donated Arbitrator instances for clarity and as a roundabout way of saying thank you. One could accomplish the same thing with additional full Arbitrator instances or by combining Arbitrator with VSTs like Cockos' ReaEQ (possibly better known as Reaper ReaEQ).
Attached Images
File Type: png LokiBidule.png (30.4 KB, 900 views)
File Type: png LokiAllocator.png (176.6 KB, 822 views)
File Type: png LokiImpulseResponse.png (17.9 KB, 653 views)
File Type: png 18SWS1100-20Hz.png (30.1 KB, 448 views)
  Reply With Quote
Old 21st October 2010, 03:58 AM   #4
SunRa is offline SunRa  Romania
diyAudio Member
 
Join Date: Aug 2004
Location: Romania
Excellent work. Truly nice stuff.

Congratulations!
  Reply With Quote
Old 21st October 2010, 05:00 AM   #5
CLS is offline CLS  Taiwan
diyAudio Member
 
Join Date: Jun 2005
Location: Taiwan
Hey I like the tree

Well done, indeed. Congratulations.

Say, did you try bringing the woofer lower, to the floor? That can provide some bass support and should be very little negative effect on L-M xover as you cross them under 200.

And, if possible, add another woofer or two filling the very bottom would be significantly nicer. 20-some and 30-some are very different and you'll like the former once you have it, believe me.

I've tried on my own system with/without center sub. I EQ'ed both conditions to the same measured response as possible for the comparison, and I like 3-channel bass a lot more. I didn't push them hard on that test, so 2-ch condition was far under the limit of the woofer, they were not stressed at all. Still, 3-ch condition sounds far more effortlessly and fills the room more evenly and smoothly. I'm not sure the improvement is from the double cone area or the distribution, or both. Anyway it's worth trying. Please make it your next project, or at least bring it up in your to-do list.
  Reply With Quote
Old 21st October 2010, 07:33 AM   #6
diyAudio Member
 
StigErik's Avatar
 
Join Date: Feb 2008
Very nice work. Well executed and documented! And I love to see that you actually have an Oscilloscope...

Very interesting thoughts you have there around EQ, phase and impulse response, and combining warped phase and linear phase EQ. I will try swithing to warped phase EQ myself for some of the EQ points in my own setup.
__________________
dipoles dipoles dipoles dipoles dipoles dipoles dipoles dipoles and dipoles
  Reply With Quote
Old 21st October 2010, 08:24 AM   #7
diyAudio Member
 
StigErik's Avatar
 
Join Date: Feb 2008
I agree with CLS that bringing the woofer closer to the floor is a good idea. It might cure the suckout you have around 150 Hz as well. I would also try a slightly higher XO between the woofer and Neo10 (like 250 or 300 Hz), to lower the strain on the Neo10 a little bit.
__________________
dipoles dipoles dipoles dipoles dipoles dipoles dipoles dipoles and dipoles
  Reply With Quote
Old 21st October 2010, 05:59 PM   #8
Helmuth is offline Helmuth  Netherlands
diyAudio Member
 
Helmuth's Avatar
 
Join Date: Oct 2008
Location: Winterswijk
Nice drivers choices twest820.

The response has 10dB non linearity are your trying to improve this or are you finished for the time being.

Keep on the good work!

Regards Helmuth
__________________
( (( KUGELWELLE )) )
recent projects :OB-mk1 /fatboy / monitor-xl / Horn-AM / dappolito / td124-mk1-rb301 / Hybrid-pse / Vfet
  Reply With Quote
Old 22nd October 2010, 01:39 AM   #9
diyAudio Member
 
Join Date: Jun 2004
Location: Orlando, FLA
I too would try the woofer to mid xover a little higher. I'm using 2 per side and still prefer over 300Hz, but maybe my max SPL requirements are higher than yours.

How much Eq are you needing to bring up the low end of the Neo to meet the woofer flat? With 2 working together on the low end of mine and having some baffle, I'm not adding ANY boost at all and getting great results.

see here:

Raal 140, Neo10, AE Dipole 15 in the works

Greg
  Reply With Quote
Old 22nd October 2010, 01:57 AM   #10
diyAudio Member
 
Join Date: Jun 2009
Thanks for the kind words and praise; glad that small book I composed last night is of some value.

We get into some of the true beauty of DIY and swinging speakers here. Bass on the floor just takes letting out a bit of line and lowering the subs. I've tried it and also done a few experiments with tippping the subs back so the front wave is more direct to the listening position. Got extension down to 30Hz with the sub magnets sitting on the floor but couldn't find a radiation pattern I cared for. That's typical of my hearing; it rolls off around 40Hz but is quite particular about consistent directivity. There's no way I'd go back to two way but my previous build had a compact presentation I quite enjoyed which the three way can't match. I've listened to the 18SWS1100s EQed flat down to 20Hz but can't really hear any difference. Not too surprising as when I'm playing stand up bass I tend to keep track of the E1 by resting the instrument against my knee and listening more via bone conduction than with my ears. On some tracks with the 20Hz EQ I think I hear deep bass harmonics clouding the higher frequencies but I'd have to do IMD testing to tell. StigErik would probably know exactly what's up with a quick listen.

I considered a floor woofer arrangement much like what CLS has suggested but rejected it as a three way on directivity grounds. As a three and a half way or a four way it seems like a good option. One of the cool things about the Neo10s is they enable a dipole four way to use two subs rather than two mids. I think an approach with something like Selenium 12PW5s to 75Hz and the second sub channel below that could be quite nice---my 18SWS1100s are 13 inches above the floor so one could insert a 12 and lower the 18s to floor level without much fuss. However, experimentation's shown I personally wouldn't be satisfied with the directivity of such a setup. Unless I can find a way around that my future speaker projects lie in the direction of building smaller three ways which still have good performance to the mid 30s.

I've tried crosses up to 500Hz between the subs and the Neo10 but prefer crossing at 200Hz. I rarely listen loud enough for it to be excursion limited at 200 and the lower the Neo10's run the more compact the speaker sounds. The drivers are reasonably broken in but I found that over several months the low end THD rise on the Neo3 softened up. I'm hoping the same will happen with the Neo10s so I can lower the cross to 175 or 150Hz---the LR6 slopes I'm using are helpful here. Those who aren't so fussy about directivity and want to listen louder or use flatter slopes should get fine results raising the cross to 300 or 400Hz. I'd also point out digital makes configuration changes trivial, so if one's feeling like listening loud it only takes a few seconds to load a different config with the crossover points moved up.

Quote:
Originally Posted by studiotech View Post
How much Eq are you needing to bring up the low end of the Neo to meet the woofer flat?
Someplace between 3dB and 10dB or so depending on choice of reference point. The Allocator screenshot in my second post is probably your best answer, though as a bit of context I find the Neo10 pretty efficient in dipole operation. BG Radia may rate it at 92dB SPL 2.83V/m but I'm running mine 2dB lower than the 96dB efficient Selenium 8W4Ps I was previously using as mids.

Quote:
Originally Posted by Helmuth View Post
The response has 10dB non linearity. Are you trying to improve this or are you finished for the time being?
Oh, this gets complicated. I assume you're referring to the 150Hz chuckhole which, as I remarked above, is mainly ceiling bounce that's not attenuated because I'm renting and hence haven't mounted a diffuser. As you can see from the close mic data in this post there's no raggeness in the wavefront coming off 18SWS1100s themselves, so the direct radiation from the drivers is just fine (curves from top to bottom are SPL, phase, and distortion as usual---EQ for flat far field is in place; SPL slopes due to near field bass boost). One school of thought---probably mostly people with hardwood floors and significant early reflection which is merged with the direct sound perceptually---favors EQ so the mic says flat. In my case carpet minimizes the floor bounce so I'm left with mostly late reflections that are heard separately from the direct sound and mostly ignored perceptually. So there's an alternate line of argument that even though the mic shows a dip there's no need for EQ because the listener perceptions are dominated by direct sound.

I've tried both approaches and found the latter sounds best to me (mentioned this in passing above). I'm not entirely convinced I prefer zero EQ but accurate discrimination becomes difficult for me when listening to relatively narrowband adjustments of one or two dB. I'm high confidence I find 0dB EQ preferable to the 6-10dB suggested by EQ to the mic (the amount needed depends on where in the room you measure). But what about 3dB? Not sure yet.

Quote:
Originally Posted by StigErik View Post
I will try switching to warped phase EQ myself for some of the EQ points in my own setup.
I look forward to your findings. We're pursuing rather different things with similar approaches and the resulting differences have always proved informative so far.

Quote:
Originally Posted by CLS View Post
Hey I like the tree
The ficus is indeed popular. It's a rescue tree but it's been doing well in that corner. I've a somewhat orphaned dracaena deremesis in another corner that's recovered nicely. They both make good corner dampers, too.

Quote:
Originally Posted by StigErik View Post
I love to see that you actually have an Oscilloscope...
It's quite handy for seeing how much power you actually use while listening and designing power amps accordingly. Plus you learn a few things about music if you sit down and watch some songs instead of just listening to them. I have to confess the scope's not actually mine though, just borrowed from tomchr.
Attached Images
File Type: png 18SWS1100-CloseMic.png (27.4 KB, 437 views)
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
swinging question .... east electronics Solid State 3 10th May 2010 10:02 AM
Nude Denon 103? tubesguy Analogue Source 19 28th April 2010 01:09 PM
Swinging choke? jazz Tubes / Valves 34 14th February 2009 04:47 PM
Looking for a site/thread with various HT dipole surround designs... jm_kzo Multi-Way 2 23rd May 2008 07:38 AM
Esoteric D-1 DAC nude photos hatasa Digital Source 27 31st August 2006 04:58 PM


New To Site? Need Help?

All times are GMT. The time now is 05:36 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2