Yarn mod question

0 Members and 1 Guest are viewing this topic. Read 1106 times.

josh358

  • Full Member
  • Posts: 1167
Yarn mod question
« on: 15 Jun 2019, 12:41 pm »
I just replaced the original midrange in my Tympani IVA's with 7 driver Neo 8 arrays, and while I tried putting strips of felt over the outer slots as per BG's suggestion. My measurements showed an improvement in dispersion without any apparent impact on output or LF performance -- the red traces are the array running full range without felt, blue with felt, gated 1 meter measurements at 0, 30, and 60 degrees, 2 dB/division:



Not only was the dispersion improved, but the 12 kHz on-axis peak, already significantly reduced by the array, as well.

The waterfall was improved as well. This is with without felt, 500 ms z axis:



And with:



Anyway, the reason I mention this is because I was wondering if anyone has compared BG's external felt to Danny's Neo 3 yarn mod. It seems to me that the yarn would slightly reduce internal reflections, but I'm not sure how the absorption compares and it would be a lot of work to thread yarn through for a test!

Something else I'm curious about is the effect of covering two slots on one side rather than one slot on either side, in order to bring the acoustic center slightly closer to the ribbon tweeter. I haven't tried it and I'm not sure what the effect would be, since the output of the Neos is presumably highest in the center owing to dishing of the diaphragm.

« Last Edit: 15 Jun 2019, 02:15 pm by josh358 »

poseidonsvoice

  • Full Member
  • Posts: 3637
  • Science is not a democracy - Earl Geddes
    • 2 channel/7 channel setup
Re: Yarn mod question
« Reply #1 on: 15 Jun 2019, 12:57 pm »
Josh,

Some advice if I may.

Your graphs have very low resolution. They are impossible to accurately assess.

The FR curves should have the vertical scaling changed, suggest making it about 50dB overall. It looks like you are wanting us to concentrate on the the >1khz area, so change the horizontal scale to reflect that. For example 1khz to 30 khz.

The waterfall time is at 1000msec and it should be about 500 msec. Similarly, please change the vertical scaling as well, or else it is impossible to interpret anything.

Best,
Anand.

josh358

  • Full Member
  • Posts: 1167
Re: Yarn mod question
« Reply #2 on: 15 Jun 2019, 01:37 pm »
Josh,

Some advice if I may.

Your graphs have very low resolution. They are impossible to accurately assess.

The FR curves should have the vertical scaling changed, suggest making it about 50dB overall. It looks like you are wanting us to concentrate on the the >1khz area, so change the horizontal scale to reflect that. For example 1khz to 30 khz.

The waterfall time is at 1000msec and it should be about 500 msec. Similarly, please change the vertical scaling as well, or else it is impossible to interpret anything.

Best,
Anand.

Thanks, and sorry about that -- REW coarsens the y axis scale when I export an image and I'm not sure how/if that can be changed. I've replaced the images in the original post above with high resolution screen captures and expanded the y and z axes on the waterfalls.

Danny Richie

  • Facilitator
  • Posts: 12705
    • http://www.gr-research.com
Re: Yarn mod question
« Reply #3 on: 15 Jun 2019, 02:37 pm »
This reminds me of the development of the LS-6 and LS-9 line arrays. There are many factors to consider when designing a line array. Lines of lower frequency drivers have to match the line of Neo's with output levels (sensitivity) and impedance.

So I had custom drivers made just for that application to achieve those objectives.

The Neo 8's were made with an 8 ohm diaphragm and used the pdr technology on the front side only. So they had felt on the outside rows. The felt improved dispersion and knocked down the 13kHz peak.

The LS-9 for instance used 3 groups of three to get back to an 8 ohm impedance.

7 drivers in a line though won't allow an even grouping.

The yarn used in our Neo 3's has the same measured effect as the flet used by BG.

And your acoustic centers shouldn't be an issue considering how low a line of these can play.

josh358

  • Full Member
  • Posts: 1167
Re: Yarn mod question
« Reply #4 on: 15 Jun 2019, 03:35 pm »
This reminds me of the development of the LS-6 and LS-9 line arrays. There are many factors to consider when designing a line array. Lines of lower frequency drivers have to match the line of Neo's with output levels (sensitivity) and impedance.

So I had custom drivers made just for that application to achieve those objectives.

The Neo 8's were made with an 8 ohm diaphragm and used the pdr technology on the front side only. So they had felt on the outside rows. The felt improved dispersion and knocked down the 13kHz peak.

The LS-9 for instance used 3 groups of three to get back to an 8 ohm impedance.

7 drivers in a line though won't allow an even grouping.

The yarn used in our Neo 3's has the same measured effect as the flet used by BG.

And your acoustic centers shouldn't be an issue considering how low a line of these can play.

Thanks, Danny. I'd originally bought 16 Neo 8's, but could only use seven per side drivers because that was all that would fit in the original baffle slot (which was however conveniently 3-1/2" wide), and the baffle isn't tall enough to enlarge the slot without splitting it in two. Then I added a 4 ohm resistor in place of the missing eighth Neo so that I could wire the array for 8 ohms. If I use two 4 ohm resistors to create a virtual 9 driver configuration, I can wire the array for 4 ohms (three paralleled groups of three drivers/resistors in series), but the Neos are so much more efficient than the other drivers that the 8 ohm impedance doesn't matter. Finally, I padded down the array to match the sensitivity to that of the ribbon while maintaining the 3 ohm impedance of the original midrange to keep the crossover happy.

I normally triamp using a digital XO, but two of the amps are dedicated to the two bass panels because I have to offset them physically to fit them in my listening room. So bass level adjustment isn't a problem. But I currently have just one amp on the midrange/tweeter panel. As much as I'd like to do it, buying another good DAC and amp to biamp the midrange and ribbon tweeter would be too expensive, so my current plan is to replace the old crossover with a single pole XO at a higher frequency, as in the newer x.7 Maggies. I hate throwing away the amplifier power and moving further out of Class A, though, so I'm considering getting another amp and using a PLLXO LPF on the array and a high level HPF on the ribbon tweeter to protect it.

I was thinking of the distance between the acoustic center between the Neos and the ribbon tweeter. They're too far apart as it is, with a 2-1/2" gap:



Someday, I hope to make new baffles that address that and allow me to install the eighth Neo, but realistically, that's pretty far away.

The midbass panel is actually separated from the mid/tweet panel because the imaging is better when I do that.


Danny Richie

  • Facilitator
  • Posts: 12705
    • http://www.gr-research.com
Re: Yarn mod question
« Reply #5 on: 15 Jun 2019, 04:36 pm »
Why add a tweeter to a tweeter? The Neo 8's will play to 40kHz.

Move them over into the same panel as your big planar panel and you're done. You'd also do away with a ton of surface reflections. so your imaging and sound stage layering will improve.

You could also drop down to six Neo 8's and match the length (acoustically) of the larger planars. Wire them in two groups of three and your impedance will be real good. You'd have 5 to 6 ohms DCR. And your sensitivity might better match the low frequency panels.

You can also experiment with a side wing down one side of the larger frame next to the bigger planars and extend their lower range a bit. Just like these: https://www.audiocircle.com/index.php?topic=159717.msg1746443#msg1746443

poseidonsvoice

  • Full Member
  • Posts: 3637
  • Science is not a democracy - Earl Geddes
    • 2 channel/7 channel setup
Re: Yarn mod question
« Reply #6 on: 15 Jun 2019, 05:08 pm »
Why add a tweeter to a tweeter? The Neo 8's will play to 40kHz.

Move them over into the same panel as your big planar panel and you're done. You'd also do away with a ton of surface reflections. so your imaging and sound stage layering will improve.
...

Not to mention the comb filtering that will occur.

In fact the measurements in the high end should improve if he sticks with the Neo8’s alone, especially if a line source is the goal.

That being said, the biggest pet peeve I have with line sources is that they completely lack image specificity at the expense of immense and exaggerated soundstaging. The polar responses particularly in the horizontal plane are poor compared to the vertical off axis response. Great for orchestral music though.

Best,
Anand.

josh358

  • Full Member
  • Posts: 1167
Re: Yarn mod question
« Reply #7 on: 15 Jun 2019, 06:24 pm »
Why add a tweeter to a tweeter? The Neo 8's will play to 40kHz.

Move them over into the same panel as your big planar panel and you're done. You'd also do away with a ton of surface reflections. so your imaging and sound stage layering will improve.

You could also drop down to six Neo 8's and match the length (acoustically) of the larger planars. Wire them in two groups of three and your impedance will be real good. You'd have 5 to 6 ohms DCR. And your sensitivity might better match the low frequency panels.

You can also experiment with a side wing down one side of the larger frame next to the bigger planars and extend their lower range a bit. Just like these: https://www.audiocircle.com/index.php?topic=159717.msg1746443#msg1746443
I tried running the array full range, but despite the fact that my hearing no longer gets to 15 kHz, compared to the ribbon, they lost a lot of air. That said I only installed the arrays a week ago and still have some work to do to normalize the response, and I should make sure as well that I'm precisely on axis, so I plan to give it another try.

I have a friend who made a similar modification to his Tympani IV's, and he reports that the ribbons are more detailed:

"There is an obvious transition between the Neos and the ribbons in the 2-4 khz range - the ribbons are clearer and more detailed but have distortion at higher levels (all the time the way I play) but the Neos play so much better dynamically that the balance is in their favor. At 4-6 khz you have a more even balance between more detail from the ribbons vs. better dynamics from the Neos. Above 6 khz or maybe 8 khz you really want the ribbons to play pretty much on their own. For me, the issue is that dropping the low pass sounds so much better than having it there and then the very common occasions where I stretch the Ribbon's dynamic capacity - where you and others might not."

He ended up running his Neo 8 array full range and then bringing in the ribbon tweeter at a higher frequency with a single pole crossover.

I did consider mounting the ribbon (which is only 1" wide) and Neos right next to the midbass driver, which apart from the tweeter would I think be similar to what you're proposing. But since the midbass panel is essentially transparent to audio I'm concerned that I'd lose too much LF extensions from the midrange? My friend crosses his Neo 8 array over to the woofer at 240 Hz in the existing baffle, which is I think a bit marginal even for the array, but I haven't yet experimented with the crossover points -- I'm using Magnepan's high level crossover for now. I've also wondered about improving the imaging/reducing diffraction by using a narrow midrange/tweeter baffle with a wing to maintain the Neo array's low frequency response.

There are actually three panels, the mid/tweet, midbass, and low bass panel behind, which I have against the wall for bass reinforcement, with the timing of the three panels compensated digitally. With that arrangement, the 3 dB point is about 25 Hz. I'm losing a bit of low bass output by splitting the panels, but have no choice because of the width of my room and it's only a couple of dB at the bottom.

josh358

  • Full Member
  • Posts: 1167
Re: Yarn mod question
« Reply #8 on: 15 Jun 2019, 06:58 pm »
Not to mention the comb filtering that will occur.

In fact the measurements in the high end should improve if he sticks with the Neo8’s alone, especially if a line source is the goal.

That being said, the biggest pet peeve I have with line sources is that they completely lack image specificity at the expense of immense and exaggerated soundstaging. The polar responses particularly in the horizontal plane are poor compared to the vertical off axis response. Great for orchestral music though.

Best,
Anand.
Combing, yes, but from what I've heard it still sounds better with the Maggie ribbon, because it's a better driver, at lower SPL's, anyway. There is some discontinuity as you cross over to the "faster" ribbon, but that can be solved by using a single pole crossover higher up, and the comb filtering can be reduced by building a new baffle in which the array and ribbon are right next to one another. A single pole Butterworth would also allow linear phase and flat on-axis and power response, albeit there would still be lateral lobing.

I don't find that my current setup lacks image specificity. The smaller Maggies do, but that's because the acoustical centers of the drivers are widely separated in the two-way models. In the Tympani IVA, you have the midrange running from maybe 300 Hz to about 2800 Hz, and then the ribbon, which is 2-1/2" from the tweeter, from 2800 Hz on up, so you get pretty good image specificity. A lateral MTM design would be even better in that regard, but I don't know of any suitable drivers since a double midrange would start to beam.

In general, I've found that image specificity can be controlled by using absorption or diffusion at the first reflection point behind the dipoles. Absorption gives you a very precise, clinical image with precise placement, but doesn't sound very real. At the opposite extreme, diffusion gives you a much vaguer image that's more like what you hear in a concert hall. Either way, I think it's a compromise, since small ensembles are better served by precise imaging and larger ones by vaguer imaging and a deep soundstage. Singers can sound too large with dipole line sources, but an orchestra will sound too small with omni/cardioid point sources. So pick your poison!

Measurements of the high end are much better with the ribbons than with the Neos alone. The power response of the Neos starts to decline at about 7 kHz, while the ribbons have essentially perfect dispersion up to 20 kHz and extension so high that Magnepan can't measure it. But mostly they sound better, unless you push them too much and then they get a hard edge. While I haven't done a final comparison, based on what I've heard, I wouldn't even think about running the Neos full range, because there's just no contest.

By off axis horizontal polar response, are you referring to crossover lobing? That's certainly an issue, but with a single pole Butterworth both on-axis and overall power response are ideal, so if you listen on axis it shouldn't be a problem. Otherwise, if the driver width is small compared to wavelength, power response should be superb and you don't have irreconcilable problems like the baffle step.

Agree that orchestral music is the line source forté, but they're also mind bogglingly good on chamber music (as heard in a hall). They have a naturalism to them on acoustic music that I've never heard from a point source. In my experience, rock recordings are all over the place, depending on the recording. And then there's the low distortion of a full height line source, and the freedom from box resonances of a dipole. Not to mention their incredible bang for the buck.

I'd be hard pressed to think of a speaker I'd rather have than a line source dipole, but that's because I listen to a lot of acoustical music, much of it recorded in (or supposed to imitate) a concert hall -- if I just listened to rock and pop, I'd probably want something like a big Wilson instead.

Danny Richie

  • Facilitator
  • Posts: 12705
    • http://www.gr-research.com
Re: Yarn mod question
« Reply #9 on: 15 Jun 2019, 10:09 pm »
I've not heard the Neo's give up much to any ribbons even up top. It might have more to do with having an amplitude unbalance.

And crossing at or above 6kHz requires the acoustic centers to be close to 2 inches apart. And if a single pole or first order slope is used then they will be completely out of phase in either direction pretty easily. This really effects the overall room response. Even if the on axis response is close to accurate, the side wall reflections can be real hot in some areas and have no output in others. That also messes up the imaging and placement of things within the sound stage.

And I don't know what you are using for digital crossovers, but I have tried quite a few and I am not a big fan of most of them. In fact the lower cost models sound really bad. If you can move towards a passive filter and put your money in a good quality DAC you'll be way ahead in all areas.

satie

  • Jr. Member
  • Posts: 90
Re: Yarn mod question
« Reply #10 on: 15 Jun 2019, 11:36 pm »
Not to mention the comb filtering that will occur.

In fact the measurements in the high end should improve if he sticks with the Neo8’s alone, especially if a line source is the goal.

That being said, the biggest pet peeve I have with line sources is that they completely lack image specificity at the expense of immense and exaggerated soundstaging. The polar responses particularly in the horizontal plane are poor compared to the vertical off axis response. Great for orchestral music though.

Best,
Anand.

The tall thin planar line sources provide the best possible image specificity, given proper room treatment if dipole, And only high directivity horn speakers are competitive. That speaking from limited experience. The best image specificity I heard was from the Beveridge wave formed ESL.  The poor horizontal dispersion may be there in some models, but aside from the null area, which is not that wide, the dispersion is quite uniform, and the lack of reflections from the sidewalls in proximity to the speaker is very helpful to image specificity.
The problem that does arise in image specificity is created mainly by backwave reflections when, as is usually the case, the speakers are placed too close to the front wall, the front wall is insufficiently treated, or in the case of a Maggie, no effort was made to take out excessive connectors fuses and less than top choices in crossover components. 
Some control of the hot sidewall reflections where comb filtering or lobing  will create tonally unbalanced reflections, can be readily treated with absorption or diffusion. They can also be addressed in part with placement. I place the M/T section behind the bass panels so that they effectively don't have a primary sidewall reflection. That also allows time alignment with a steeper low pass (Butterworth 3rd order) to keep the bass panel thickness out of the midrange. This approach was also adopted for the new 30.7 model. (not the crossover) Another way to control the sidewall reflections is with toe in, such that the sidewall gets nearly no energy.

satie

  • Jr. Member
  • Posts: 90
Re: Yarn mod question
« Reply #11 on: 15 Jun 2019, 11:51 pm »
I've not heard the Neo's give up much to any ribbons even up top. It might have more to do with having an amplitude unbalance.

And crossing at or above 6kHz requires the acoustic centers to be close to 2 inches apart. And if a single pole or first order slope is used then they will be completely out of phase in either direction pretty easily. This really effects the overall room response. Even if the on axis response is close to accurate, the side wall reflections can be real hot in some areas and have no output in others. That also messes up the imaging and placement of things within the sound stage.

And I don't know what you are using for digital crossovers, but I have tried quite a few and I am not a big fan of most of them. In fact the lower cost models sound really bad. If you can move towards a passive filter and put your money in a good quality DAC you'll be way ahead in all areas.

I am not sure that outside of the intermediate placements of mid and ribbon between very close (less than a couple of inches) and "far"  apart (say beyond one foot) there is that much significance to where the tweeter is relative to the midrange so long as they are aligned correctly in time and aimed on axis and both are at locations 30-40 degrees or more from the listening axis.

Danny Richie

  • Facilitator
  • Posts: 12705
    • http://www.gr-research.com
Re: Yarn mod question
« Reply #12 on: 16 Jun 2019, 01:08 am »
I am not sure that outside of the intermediate placements of mid and ribbon between very close (less than a couple of inches) and "far"  apart (say beyond one foot) there is that much significance to where the tweeter is relative to the midrange so long as they are aligned correctly in time and aimed on axis and both are at locations 30-40 degrees or more from the listening axis.

In short, not quite.

josh358

  • Full Member
  • Posts: 1167
Re: Yarn mod question
« Reply #13 on: 16 Jun 2019, 01:50 am »
The tall thin planar line sources provide the best possible image specificity, given proper room treatment if dipole, And only high directivity horn speakers are competitive. That speaking from limited experience. The best image specificity I heard was from the Beveridge wave formed ESL.  The poor horizontal dispersion may be there in some models, but aside from the null area, which is not that wide, the dispersion is quite uniform, and the lack of reflections from the sidewalls in proximity to the speaker is very helpful to image specificity.
The problem that does arise in image specificity is created mainly by backwave reflections when, as is usually the case, the speakers are placed too close to the front wall, the front wall is insufficiently treated, or in the case of a Maggie, no effort was made to take out excessive connectors fuses and less than top choices in crossover components. 
Some control of the hot sidewall reflections where comb filtering or lobing  will create tonally unbalanced reflections, can be readily treated with absorption or diffusion. They can also be addressed in part with placement. I place the M/T section behind the bass panels so that they effectively don't have a primary sidewall reflection. That also allows time alignment with a steeper low pass (Butterworth 3rd order) to keep the bass panel thickness out of the midrange. This approach was also adopted for the new 30.7 model. (not the crossover) Another way to control the sidewall reflections is with toe in, such that the sidewall gets nearly no energy.
I have some 703 on the sidewalls at the first reflection points. This I think helps with the reflections of the mid/tweet backwave off the baffles on the low bass baffles behind them, although I still have some stubborn peaks of about -15 dB in the first 20ms impulse response -- tracking them down and getting everything below -20 is among my long-range plans.

For some reason, when I toe in the midbass panels, the imaging goes to hell. And conversely, it's better when the midbass panels aren't right next to the mid/tweet panels, not sure whether that's due to diffraction or to the baffle blocking the reflected backwave? It took me a long time to find the optimal panel configuration for this undersized room, but since I did, and added diffusion to the corner bass traps, the imaging has been phenomenal. Not I think that it couldn't be even better!

I'm intrigued by the small baffle idea that Danny discussed, which is similar to one of the ideas I've bounced around. I can imagine combining a single three-way panel with the GR OB woofers -- a more compact system that would be more sensible in this space. Though really if I could do anything at this point it would be to add a midbass coupler that doesn't color the Neos/require a high order crossover. But I don't know of a suitable planar driver, I don't think the Neo 10 goes deep enough. In any case, I don't want to jump to conclusions until I've experimented with XO points.

HAL

  • Industry Contributor
  • Posts: 4344
Re: Yarn mod question
« Reply #14 on: 16 Jun 2019, 02:16 am »
Yes, wide front baffles loose the imaging capability, but not all line sources have that problem.

These do very well for imaging based on Danny's angled winged side baffle.




satie

  • Jr. Member
  • Posts: 90
Re: Yarn mod question
« Reply #15 on: 16 Jun 2019, 03:07 am »
What are the tweeters?

10" bass units in the towers?

satie

  • Jr. Member
  • Posts: 90
Re: Yarn mod question
« Reply #16 on: 16 Jun 2019, 03:19 am »
I have some 703 on the sidewalls at the first reflection points. This I think helps with the reflections of the mid/tweet backwave off the baffles on the low bass baffles behind them, although I still have some stubborn peaks of about -15 dB in the first 20ms impulse response -- tracking them down and getting everything below -20 is among my long-range plans.

For some reason, when I toe in the midbass panels, the imaging goes to hell. And conversely, it's better when the midbass panels aren't right next to the mid/tweet panels, not sure whether that's due to diffraction or to the baffle blocking the reflected backwave? It took me a long time to find the optimal panel configuration for this undersized room, but since I did, and added diffusion to the corner bass traps, the imaging has been phenomenal. Not I think that it couldn't be even better!

I'm intrigued by the small baffle idea that Danny discussed, which is similar to one of the ideas I've bounced around. I can imagine combining a single three-way panel with the GR OB woofers -- a more compact system that would be more sensible in this space. Though really if I could do anything at this point it would be to add a midbass coupler that doesn't color the Neos/require a high order crossover. But I don't know of a suitable planar driver, I don't think the Neo 10 goes deep enough. In any case, I don't want to jump to conclusions until I've experimented with XO points.

The narrow baffle idea is on all our to do lists with the ribbon and mids nearly touching.

My guess as to the midbass panel toe in is indeed diffration. Whenever I extend the baffle for the mids I cover it in a fuzzy blanket/throw to avoid diffraction. A;so in the equidistant arc setup a bit of absorption at the baffle by the mids is useful. For the same reason I don't run the tweeters nor the mids without the cloth to cover the baffle.

HAL

  • Industry Contributor
  • Posts: 4344
Re: Yarn mod question
« Reply #17 on: 16 Jun 2019, 03:20 am »
Four Custom GT Audioworks ribbons and 12" GR-Research servo subs in the towers.

josh358

  • Full Member
  • Posts: 1167
Re: Yarn mod question
« Reply #18 on: 16 Jun 2019, 10:18 am »
Yes, wide front baffles loose the imaging capability, but not all line sources have that problem.

These do very well for imaging based on Danny's angled winged side baffle.


Impressive. So the front facet is what, about 10" wide? How much LF response are you getting out of the Neo 10's with the wing?

josh358

  • Full Member
  • Posts: 1167
Re: Yarn mod question
« Reply #19 on: 16 Jun 2019, 10:32 am »
The narrow baffle idea is on all our to do lists with the ribbon and mids nearly touching.

My guess as to the midbass panel toe in is indeed diffration. Whenever I extend the baffle for the mids I cover it in a fuzzy blanket/throw to avoid diffraction. A;so in the equidistant arc setup a bit of absorption at the baffle by the mids is useful. For the same reason I don't run the tweeters nor the mids without the cloth to cover the baffle.
I've considered felting the baffles. The earlier production of the IVA did have felt on the mid/tweet panels -- I'm not sure why stopped. I have to ask Wendell at some point.

It seems to me that the styles should be replaced with something round and flush, too, to minimize diffraction at the edges.

That said, I get very good, even spectacular imaging with the 18" mid/tweet panels separated from the bass panels. I assume that a narrower panel would do even better. The 30.7 uses a narrower mid tweet panel, and Wendell is quoted somewhere as saying that the only reason it wa

isn't narrower was that a narrower panel doesn't look as good.