That`s why I wrote "admittedly positioned at a larger distance from the OBs...". Just for the sake of this discussion and not as 'carved in stone', I´d say that 1 m would be the minimum acceptable distance (if you say 1.5 m that`s fine with me too) and 2 m is sufficient in most cases (gaining not much more when going farther). What I really was after - and what you ignored - was the highly nonsymmetric situation of your built-up, which cries for additional means of symmetry to get at least some imaging.
Old Opera houses had the kings loge. Placed in the best viewing position und a centered, but somewhat distant listening position. The loge was a room in itself with most of the front wall missing and instead opening to the larger auditorium. The king was sitting in the ambiance of a small room, which was flooded by the sound coming in from the big hall. Why did the king tolerate to sit in a room, which "destroy(ed) the original intention of the (performance)?"
Because his brain could very well discern between the acoustic effects of the small loge compared to the acoustics of the probably 100 or even 1000 times larger space of the opera hall.
Can we agree on forgetting about the word "destroy" and use "somewhat compromising" instead?
I'm going to be a very naughty boy now, but I can't resist. Your picture is a fine example of the prevailing situation of today: People fiddling with the reflections in their rooms before enabling their speakers to "fully reveal the recording".
I very much like how he says "not exactly accurate" instead of "destroyed" or "just a diffuse wash of sound anywhere across the middle".
The box we have to leave is not the technical contraption in the first place IMHO. It is predominantly our preformed notions of how and why things sound (or don't sound) the way they do.
Certainly you don't mean "rear wave" but "reflections from the rear (front) wall". I'm not sure that Linkwitz and others are against killing reflections in general. But they are against doing it in a wrong way: only locally at certain spots and/or in small bands instead of evenly across (most of) the frequency range. We have seen what you have invested to do it in the right way.
It would be interesting to compare your room as it is today with the original lacking any special means against reflections - in both cases with your current system of course. I wonder if the old room would be 5% (almost unlistenable), 50 % (nice, but still some miles to go) or 95 % (only missing the last tiny details)? I don't expect an answer to this - just wondering.
Rudolf
OK, Rudolf, you got me. I will agree to not say "destroy" anymore.
Yes, I have a huge hurdle to overcome as one speaker is against a wall and the other is about 4 feet away. That is a tough one, but I will make a large absorber of bonded logic for the near wall and "fix" any remaining, large scale discontinuities with some subtle EQ.
I think at this point we will have to agree to disagree on the "dead end" preference in a listening room. I have seen too many amateur level studios that place absorption literally everywhere in the control room. Bad, bad situation to MIX, but not the worst case to listen back to good mixes and recordings done properly. I would take too dead over too live any day for pure playback. I see so many audiofools who spend mega $ on gear and have hard walls, wood or tile floors and no curtains. Wow, what an expensive echo camber they must have. Too bad they never get to hear the true potential of their fancy pants systems.
Back to Tests regularly scheduled thread.....
Greg
PS. If the King really preferred his loge and didn't actually stay in there for other, more practical reasons, like being away from the common folk, then he was a fool too.
Yes, I have a huge hurdle to overcome as one speaker is against a wall and the other is about 4 feet away. That is a tough one, but I will make a large absorber of bonded logic for the near wall and "fix" any remaining, large scale discontinuities with some subtle EQ.
I think at this point we will have to agree to disagree on the "dead end" preference in a listening room. I have seen too many amateur level studios that place absorption literally everywhere in the control room. Bad, bad situation to MIX, but not the worst case to listen back to good mixes and recordings done properly. I would take too dead over too live any day for pure playback. I see so many audiofools who spend mega $ on gear and have hard walls, wood or tile floors and no curtains. Wow, what an expensive echo camber they must have. Too bad they never get to hear the true potential of their fancy pants systems.
Back to Tests regularly scheduled thread.....
Greg
PS. If the King really preferred his loge and didn't actually stay in there for other, more practical reasons, like being away from the common folk, then he was a fool too.
So I'm looking into a low cost IMD measurement setup to get the 18SWS1100s more fully characterized. Any recommendations? I'm fortunate enough to have access to test and measurement equipment at various levels up to an Audio Precision 2 so I don't need anything heavy duty. But a halfway decent in house capability would save a fair amount of hassle hauling drivers around and enable better use of the time on the serious gear. I'm aware of RMAA, SpectraPlus, TrueRTA, and a few others but it's more my style just to dump the input to ReaFIR and look at the FFT (RMAA unfortunately locks up my laptop---the driver for the Ricoh 1394 controller's not the greatest). However, it'd be nice to have a few measurement oriented features like cursors and I'm not spotting much for sources besides Multi Tone Generator.
It's overrated for home audio. At least if you want to listen to your stereo more than once. Interesting driver choice. Did you consider the SB15NRXC30-4?Not exactly 120dB though...
Will be interesting on what concusions you will arrive by IMD measurements of nude OB.
Keep aware that you are possibly not measuring the *driver* but rather the principle pf operation.
What I mean is what I've outlined here
Audio and Loudspeaker Design Guide Lines
Though its not exactly the same case as the amount of re-radiation through the diaphragm may be pretty close to nothing (just what is reflected by diffraction at the speaker's rim) - but you will have a large amount of overlay from the rear wave.
So - it might be good to theoretically clarify alongside your measurements what one can get "at best" by nude OB IMD measurements.
For measurements I used AudioTester, here an example form the "Back Diaphragm Mirror Distortion" link above:
Michael
Keep aware that you are possibly not measuring the *driver* but rather the principle pf operation.
What I mean is what I've outlined here
Audio and Loudspeaker Design Guide Lines
Though its not exactly the same case as the amount of re-radiation through the diaphragm may be pretty close to nothing (just what is reflected by diffraction at the speaker's rim) - but you will have a large amount of overlay from the rear wave.
So - it might be good to theoretically clarify alongside your measurements what one can get "at best" by nude OB IMD measurements.
For measurements I used AudioTester, here an example form the "Back Diaphragm Mirror Distortion" link above:
An externally hosted image should be here but it was not working when we last tested it.
Michael
Last edited:
Yeah sure, always watch the right turn on your volume knob !
thanks for the link.
Those SB guys make a loooot of interesting drivers as it seems (equally high Qms, only slightly less power handling and sensitivity, plus even a non-rubber surround - WHOW !)
Michael
Last edited:
Lowest Rms I've seen, cheaper than most alternatives (at least in the US), less smoothing than usual in the datasheets, usually pretty low inductance, good xmax, and distortion's generally a wash with other low distortion drivers. For eight inch and smaller cone drivers SB's ended up as my default, Selenium for anything bigger. One option I was considering was push pull SB15NRXC30-4s since, if you mill flats onto the flanges, it's just possible to cram a pair into about the same height as a Neo10. Considerably more work to build what with cutting down the flanges and the radiating surface is larger than the Neo10 so the directvity probably wouldn't be as good, but the pair's a fair amount cheaper than a single Neo10, has good xmax (more than the SS parts, IIRC), and the sensitivity's decent by hi-fi driver standards.
Close mic in the deep bass gets a pretty clean front or rear field. I would be kinda surprised if mirror distortion is visible in the far field over the room interactions in normal operation but we'll see what happens at 20Hz. Thanks for the audioTester suggestion; I did a quick hello world with the demo version and it should do just fine. Power amp parts arrived yesterday so I'm busy for a while with getting the prototype channel up and running though.
Pre-ringing has to be looked at from the total system point of view.
As beautifully outlined in ACOURATE manuals (scroll down to page 13, 14) preringing *in theory* would be canceled out by the two branches of XO *if* done properly.
I don't think this happens perfectly in real world but seems to work "good enough" - at least for me too.
http://www.acourate.com/TutorialsDoc/CrossoverGeneration.doc
Michael
Oh duh. Thanks for the remider.
it's interesting to note most of the preringing is slower than the rest of the impulse and is roughly a 1.7kHz wave shape. There's a 30 degree phase movement around the Neo3 to Neo10 crossover; that could be the cause of incomplete cancellation. If so, results should be better in the far field where there's less path length difference between the drivers. I'll have to check that and see if I can measure anything beyond room effects.
Thank you, first measurements are promising. Electrical measurements are below the distortion floor on the Saffire---gonna have to upgrade its op amps---but the impact on the driver from getting the amp GBP up and operating at unity gain to take some of the stress off the Saffire DACs is interesting. That THD hump on the Neo3 at 1.4kHz appears to be a pendulum resonance. So far I've only figured out ways to make it worse (my record so far is boosting it to -20dB by attaching an oscilloscope probe to the drivergood luck with the amps
), but there should be ways to make it better.Attachments
Last edited:
With measuring at a swing, there is only *a single* pendulum resonance determined by the length of the pendulum (way lower) ???
Michael
floating resonances
There may be only one theoretical pendulum frequency but the cord suspending the driver isn't rigid and is not very massive so the drivers can also show some vibrational behavior that is similar to being free floating in the air with no gravity. Some mass added to the frame of the Neos, such as stick on wheel weights, would make a huge difference. It would certainly lower the frequency of the resonance to get further away from the transition band and should also knock down the amplitudes of any modes except those near the subsonic pendulum frequency which will never be a problem. The wheel weights could also be easily shaped into a small, 30-45 degree wave guide so as not to cause problems with the propagating wave front and also to slightly round over the edge transition without adding any width.
There may be only one theoretical pendulum frequency but the cord suspending the driver isn't rigid and is not very massive so the drivers can also show some vibrational behavior that is similar to being free floating in the air with no gravity. Some mass added to the frame of the Neos, such as stick on wheel weights, would make a huge difference. It would certainly lower the frequency of the resonance to get further away from the transition band and should also knock down the amplitudes of any modes except those near the subsonic pendulum frequency which will never be a problem. The wheel weights could also be easily shaped into a small, 30-45 degree wave guide so as not to cause problems with the propagating wave front and also to slightly round over the edge transition without adding any width.
See posts 1 and 20 in this thread. Any Neo10 front/rear asymmetry from the Neo3 cable is below measurement noise so I expect equally good matching once I get around to vertical polars. My avatar should give you some ideas about knocking the back cup off the Neo3W, though I believe some folks take them off with knives or chisels. You could probably also get some without back cup from Madisound since they're sold that way in the Nao Note driver set.
Sorry, I was using pendulum in a catch all sense rather in a specific one. For a single pendulum there's a single swinging resonance if, as sendler noted, the mass of the of the line is small compared to the mass of the driver and the pendulum swings freely. However, the hang structure I'm using for the Neos is a pantagraph with three flexible sides, all under tension. So there are nominally three pendulum modes as well as the usual resonances of tensioned strings. I'm not sure the 3mm aux cord holding up the AWG10 zip to the drivers satisfies the free swinging constraint and I'm quite sure the zip itself doesn't. Since the zip attaches away from the Neo3's center of mass it could be a zip cord swinging mode---think pendulum, but horizontal. However, quarter wave at 1.4kHz is about 6cm and there are several things about that size here so I can't rule out some sort of acoustic coupling. And then there are string modes in the 2mm aux cord or, less likely, the 3mm and zip.
Maybe; with the few things I've tried add mass seems to increase the Q of the resonance and leave its frequency unchanged. I have some ideas which I think are more likely to decrease the Q and, ideally, lower the frequency. But since this is all below -50dB and I'm busy with power amps it's not a priority at the moment. I suspect it's more a theoretical interest; supposing it cleans up nicely and THD drops to -60dB I would be somewhat surprised if I can hear a statistically significant difference. However, I'd like to verify that. I'll get to this eventually, but it's likely to be a bit of a wait.
I've remarked on this in a few other threads, but diffraction control in dipoles is not something which has seen much investigation. A plate flush with the Neo3 face but with a roundover on the sides has been something I've had in mind for some time, though adding some width on the sides is probably acceptable as it should make horizontal and vertical directivity more consistent.
Geddes suggested toroidal baffles with OS throats some years ago and I suspect the Neos are close enough to plane wave emission for the OS solution to be admissable. The only data I have in this direction is saurav and villastrangiato expirimented with sections of 10-20mm pipe on the sides of a Neo3 and arrived at the unsurprising result the pipe doesn't affect the 6.8kHz dipole bloom. However, those experiments did not attempt to control the waveguide throat. I suspect the difference between an OS throat and just slitting some ABS pipe or Delrin rod and press fitting it on to the sizes of the Neo3 is probably small for structures of waveguide size. For horns I would be more concerned.
The other idea which looks interesting is a Le'Cleach profile. jmmlc calculated a dipole horn profile for mige0 and he did a Neo3 build with it which targeted horizontal pattern control but neglected vertical. My understanding of Le'Cleach theory is rudimentary, but I think it should be calculate a Le'Cleach dipole waveguide. Michael, any thoughts here?
I guess you should dig deeper into swings
1. a swing is *meant* to suspend the driver like in outer space - meaning - this would be the ideal case to aim for.
2. there *must* be what you call "vibrational behavior" - this is part and parcel of 100% impulse cancellation with swings - these movements are *not* resonance - its a perfectly time aligned counter movement with no phase shift ringing or decay
3. there is no such thing as a resonance except the pendulum resonance
4. there is no such thing as a frequency down shift with added mass for pendulums
5. adding mass to the chassis affects the impulse cancellation movement of the chassis with respect to the diaphragm, yes, but this mass ratio usually is already way good enough - even for lightweight chassis as the NEO's
Ok, got your point...
Michael
Last edited:
Apropos of the couping issues being discussed in DQ828's build thread I thought I share some of this weekend's crosstalk data from my amp project. There are various advantages of putting amps onboard speakers---tidiness probably being the main one, particularly if cable sleeves or a snake are used for the inputs---and I happen to be working with a compact triamp which represents something of a best case for aggressor signal rejection. The amp channels are 2cm by 5cm with balanced inputs whose shield stops 5mm from the board and the passives are 0805 and 1206 resistors and caps. In the attached figure blue is the noise floor of my measurement setup, red and teal are the amp output with the board literally sitting directly on the 18SWS1100 magnet in a couple different orientations right over the voice coil, and green is the DS/Rnonlin multitone IMD signal I like to use in such tests. Specifically, red is with the board horizontal and teal is with the board vertical and parallel to the Neo10. Horizontal should be minimum field coupling to the 18SWS11000 and vertical maximum. Pickup from the Neo10 is roughly equal in both orientations and would reduce with the board vertical but perpendicular to the Neo10.
audioTester seems to have some issues resolving peaks---the tones should be even at -55dBFS and a look with the oscope confirms that's what's happening on the wire (I have to get around to trying ARTA). But, possible audioTester dodginess aside, I have yet to find any placement of the amp board where I can get any crosstalk from the drivers to show up on the amp output; there's no sign of an IMD floor lifting above the noise floor. There might be a few peaks coupled in with the board vertical but this data is averaged across several samples to try to go fishing below the noise floor and even then there's a couple dB of variation in the peaks around the averaged noise level. So I don't see the bobble at, say, 520Hz as significant.
I'll see about doing the same check on the other drivers.
audioTester seems to have some issues resolving peaks---the tones should be even at -55dBFS and a look with the oscope confirms that's what's happening on the wire (I have to get around to trying ARTA). But, possible audioTester dodginess aside, I have yet to find any placement of the amp board where I can get any crosstalk from the drivers to show up on the amp output; there's no sign of an IMD floor lifting above the noise floor. There might be a few peaks coupled in with the board vertical but this data is averaged across several samples to try to go fishing below the noise floor and even then there's a couple dB of variation in the peaks around the averaged noise level. So I don't see the bobble at, say, 520Hz as significant.
Yea, verily. I've not seen an Mms estimate for the Neo10 but for the Neo3 I recall it being around 0.03g, a 100000:1 chassis:membrane mass ratio. Cone drivers are usually more like 1000:1 and I would expect the Neo10's probably comfortably above 10000:1.
Let me know if you come up with any ideas.
I'll see about doing the same check on the other drivers.Attachments
Keep it simple:
First off, think of diffraction simply as "bending* (bending of the wavefront that is).
You immediately will realize that whatever you do with dipoles - you always end up with bending completely around the corner - meaning - diffraction will always be the same (=180deg).
This is the principle of OB dipoles.
From that you further easily will realize that "treating" diffraction with OB principle is kinda nonsense.
*If* you'd like to play around with round overs (diffraction alignment), you pretty quick get into horn territory - with all the fuss of pro and con and the never ending discussions connected to it.
All in all i'd say: try nude and go for equalizing as far as it gets - as some have reported good results this way.
Anything else will be pretty time consuming and trade offs never end anyway.
Michael
Last edited:
Had no such issues - have you tried to zoom in to see if its a display problem?
Maybe difference between sound card and osci measurement is related to the huge difference in input impedance?
Michael
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.