"NBAC" Update:
Busy long weekend.
The issues has mainly been about whether we can make a Waveguide that was designed for the Peerless HDS/Scan-Speak D2608 tweeter work with the SB Acoustics SB26CDC Tweeter.
The answer is no!
The fit in the throat is ultra-critical and despite trying to find a way, I am getting results that are not good and not consistent. The main but not only problem, is a severe and wide suck-out at 9KHz. The phase plate has to be cut away, to get the Waveguide throat in, but it is not possible to get a good fit. I have experimented with Blu-tack as a crude filler, to see if the gap there is causing the 9KHz suck-out. And I have confirmed that is the cause. Not a perfect fit and you get severe issues, like -9dB @ 9KHz, ouch!
It is clear that a Waveguide has to be designed from scratch to make it work. No surprise really.
But we are still going ahead with our regular HDS/Scan-Speak Tweeter.
The fact is that the Waveguide has got to be designed specifically for the Tweeter used.
So we are going ahead using the "NBAC" drivers and the HDS/S-S Tweeter, which will work perfectly together. The voltage sensitivity is much higher than the SB and it is nominally 8 Ohm and not 4 Ohm, so I have no doubt that this will work.
Important: Even if we could get a Waveguide for the SB Tweeter, there is no guarantee that it would work. It has other issues, like low sensitivity. But then again the "NBAC" is lower too, so it might still work, but just not guaranteed.
The existing Waveguide we are using now, is now around ten years old. It was designed to do a singular job and some may point to newer high-tech Waveguides now out. Fair enough. I have certainly looked at the SB TW29BNWG-8 Tweeter. Pre-pandemic visit to Denmark and Ulrik Schmidt of SB told me he was working on a Waveguide Tweeter, and he certainly has delivered. It looks great.
It is quite expensive and down here you have to get it ordered in because it is too expensive to keep in stock. As it has it own up-to-date WG built in, and yes it work work in the Elsinores. But I am more inclined to wait and see what Purifi comes up with.
But right now I have a pair of white boxes and I need to go ahead with that using the Waveguide that fits those cabinets..
The existing Waveguide is not to be sneezed at. On Sunday I got this pleasing result shown below:
A thing of beauty, if I may say so myself.
Our Waveguide was designed to be optimum at 15° off axis. That is how the above was measured at 2 metres outdoors on Sunday.
I am still quite proud of the work I did ten years ago with this Waveguide. It does exactly what I set it out to do. 😎 👍
CSD (Waterfall) plot:
Later this week, as I have all the measurements to proceed, the data will be processed and exported into SoundEasy where we can work on the Crossover.
Compared to earlier Crossovers, this design will still use our existing Crossover PCB, but values will be changed. There will be extra components, such as an LCR network that will go across the MidBass (the upper two main drivers) that will be tuned around 8KHz to control the main break-up mode that is typical of a hard cone (aluminium in this case). The correction will be as mild as I can get away with, but the good thing about this driver is that it is higher in frequency than I have seen before. As Troels Gravesen has said earlier, this aluminium driver does not sound like a hard cone driver. That is the main reason that drives me to do the "NBAC" version.
So that is where we are up to for now.
Cheers, Joe
Busy long weekend.
The issues has mainly been about whether we can make a Waveguide that was designed for the Peerless HDS/Scan-Speak D2608 tweeter work with the SB Acoustics SB26CDC Tweeter.
The answer is no!
The fit in the throat is ultra-critical and despite trying to find a way, I am getting results that are not good and not consistent. The main but not only problem, is a severe and wide suck-out at 9KHz. The phase plate has to be cut away, to get the Waveguide throat in, but it is not possible to get a good fit. I have experimented with Blu-tack as a crude filler, to see if the gap there is causing the 9KHz suck-out. And I have confirmed that is the cause. Not a perfect fit and you get severe issues, like -9dB @ 9KHz, ouch!
It is clear that a Waveguide has to be designed from scratch to make it work. No surprise really.
But we are still going ahead with our regular HDS/Scan-Speak Tweeter.
The fact is that the Waveguide has got to be designed specifically for the Tweeter used.
So we are going ahead using the "NBAC" drivers and the HDS/S-S Tweeter, which will work perfectly together. The voltage sensitivity is much higher than the SB and it is nominally 8 Ohm and not 4 Ohm, so I have no doubt that this will work.
Important: Even if we could get a Waveguide for the SB Tweeter, there is no guarantee that it would work. It has other issues, like low sensitivity. But then again the "NBAC" is lower too, so it might still work, but just not guaranteed.
The existing Waveguide we are using now, is now around ten years old. It was designed to do a singular job and some may point to newer high-tech Waveguides now out. Fair enough. I have certainly looked at the SB TW29BNWG-8 Tweeter. Pre-pandemic visit to Denmark and Ulrik Schmidt of SB told me he was working on a Waveguide Tweeter, and he certainly has delivered. It looks great.
It is quite expensive and down here you have to get it ordered in because it is too expensive to keep in stock. As it has it own up-to-date WG built in, and yes it work work in the Elsinores. But I am more inclined to wait and see what Purifi comes up with.
But right now I have a pair of white boxes and I need to go ahead with that using the Waveguide that fits those cabinets..
The existing Waveguide is not to be sneezed at. On Sunday I got this pleasing result shown below:
A thing of beauty, if I may say so myself.
Our Waveguide was designed to be optimum at 15° off axis. That is how the above was measured at 2 metres outdoors on Sunday.
I am still quite proud of the work I did ten years ago with this Waveguide. It does exactly what I set it out to do. 😎 👍
CSD (Waterfall) plot:
Later this week, as I have all the measurements to proceed, the data will be processed and exported into SoundEasy where we can work on the Crossover.
Compared to earlier Crossovers, this design will still use our existing Crossover PCB, but values will be changed. There will be extra components, such as an LCR network that will go across the MidBass (the upper two main drivers) that will be tuned around 8KHz to control the main break-up mode that is typical of a hard cone (aluminium in this case). The correction will be as mild as I can get away with, but the good thing about this driver is that it is higher in frequency than I have seen before. As Troels Gravesen has said earlier, this aluminium driver does not sound like a hard cone driver. That is the main reason that drives me to do the "NBAC" version.
So that is where we are up to for now.
Cheers, Joe
Joe,
Just FYI, there are DIY waveguides for the SB26CDC tweeter that can be 3D printed. So an option if you want to try.
Just FYI, there are DIY waveguides for the SB26CDC tweeter that can be 3D printed. So an option if you want to try.
Yes, an opportunity to use an axisymmetrical waveguide.
Isn't there over 200mm of space between the mids? In fact, with 280mm of front baffle width, an overlap with the woofers could be printed into the waveguide.
Isn't there over 200mm of space between the mids? In fact, with 280mm of front baffle width, an overlap with the woofers could be printed into the waveguide.
That may be so, but it needs to be 148mm diameter and 18mm deep round cavity in the speakers.
We are not totally discounting using the SB26CDC yet, but since all the measurements have been done with the D2608 and that I have expensive Waveguides in stock, and that sensitivity is good (the SB26CDC somewhat challenged here) and 8 Ohm and not 4 Ohm... so you can understand my decision, right? There is no guarantee that the SB26CDC will work out, but it does not mean that we won't try again in the future.
Have you seen anybody who has put on the net drawings to fit SB26CDC. I would like to see more if available. I have seen website www.somasonus.net/sb-acoustics-sb26 - but I need a way to look at step files. Is there any software for free, it definitely looks worthwhile to look. I am interested/curious.
BOXES READY TO GO:
We are not totally discounting using the SB26CDC yet, but since all the measurements have been done with the D2608 and that I have expensive Waveguides in stock, and that sensitivity is good (the SB26CDC somewhat challenged here) and 8 Ohm and not 4 Ohm... so you can understand my decision, right? There is no guarantee that the SB26CDC will work out, but it does not mean that we won't try again in the future.
Have you seen anybody who has put on the net drawings to fit SB26CDC. I would like to see more if available. I have seen website www.somasonus.net/sb-acoustics-sb26 - but I need a way to look at step files. Is there any software for free, it definitely looks worthwhile to look. I am interested/curious.
BOXES READY TO GO:
Joe,
You can view STL and STEP files on this site under instant quote feature:
https://craftcloud3d.com/
You will be asked to upload the file to generate the quote, and once you upload you will be able to see the 3D model.
You can view STL and STEP files on this site under instant quote feature:
https://craftcloud3d.com/
You will be asked to upload the file to generate the quote, and once you upload you will be able to see the 3D model.
Perhaps your current CAD software already can (but clearly it's not a file format we normally construct with).
My copy of TurboCAD is too old. I gather newer versions will.
I am actually looking for dimensions. Particularly the throat.
But 3D printing option might be the way to work, great for what would be speculative work. I think I know somebody with the resources who could do it.
But in the meantime, I am doing the "NBAC" version with what I know 100% will work (D2608) and the other is more doubtful - and I the boxes ready to do.
The SBC26CDC has lower voltage sensitivity and lower 4 Ohm. But comparing the gap in the S-S D2608 the VC in gap is noticeably tighter and the SB26CDC gap is wider. What that tells me is that the D2608 is built to tighter tolerances. If the gap in the SB26CDC was tighter/narrower, the sensitivity would have been higher.
Joe,
With the site I shared earlier you can check the dimensions of a step file when you upload it for instant quote.
No need to install any software.
With the site I shared earlier you can check the dimensions of a step file when you upload it for instant quote.
No need to install any software.
The throat on the SB26 waveguides are 33.02mm with a 0.5mm thick spacer on the back.
To make something with the same type of performance to fit the Elinsore cutout needs something like this.
The yellow highlighted sketch point co-ordinates in the image will allow a circular radius of 100.808mm to be drawn between the throat and mouth edge.
To make something with the same type of performance to fit the Elinsore cutout needs something like this.
The yellow highlighted sketch point co-ordinates in the image will allow a circular radius of 100.808mm to be drawn between the throat and mouth edge.
Thanks. So the face plate of the tweeter removed? That would explain the 33.02mm, that diameter sits right outside the Tweeter's surround.
Yes. Something could perhaps be done to match the existing faceplate, but it is usually better to start the waveguide profile as close to the surround as possible.
Must be really close. It is critical.
With the SBC26CDC it might be possible to do with the original face plate.
But not unless you snip the little phase plate cover off. It prevents any waveguide getting right in there. It is possible that the current Elsinore Waveguide, which has the 32mm diameter at the throat and that is bang-on where it needs to be. If we can modify the back of the waveguide enough and get it right inside so that there is no gap, that could work very well. But the phase thingy has to go and I don't mind that too much, but it does leave it more unprotected against accidental physical damage. But the depth of the waveguide does give it some protection. Keep children away, hmmmmm...
Interesting, something I have not really taken note of this before, Lars Risbo said their new super-tweeter (😉) will have VC underhung. I suspect they will go, based on past history, with a tweeter with more than the usual excursion/Xmax. But then I note, because I have them open in front of me, that both the D2608 and SB26CDC are underhung as well - so it must be normal. Funny how that can be missed that, as most drivers we usually think of are overhung drivers and we obsess about that at LF excursions and all that. But Purifi's greater excursion/Xmax would indicate very low distortion where most tweeters increase distortion, at the bottom of the passband. So I am interested in that. It also means that those into doing 1st order Crossovers might be interested in it. Just some musings on my part...?
"NBAC" Update:
I showed the Tweeter, and I thought I should show you some results of the far-field/near-field splicing and FR of the driver in situ, in the box.
This is the top two drivers, these will of course be part of an MTM array with the Tweeter.
The slight bulge around 70-80Hz is a product of the nearfield measurement, it is somewhat flatter than that, by around 2 or 3dB.
Below is the bottom two driver, note the suckout is caused by the fact that, unlike the two top drivers are co-distant, the bottom two drivers are not. The second from the bottom driver is closer and the bottom driver is slightly further away. Keep in mind that the microphone is at the height of the Tweeter and at 2 metres:
The two driver will be rolled off from 400-500Hz up first order. By the time we get to the midrange above 1-2KHz, we get a 3rd order effect and this is a good thing for the midrange. All Elsinores benefit from this trick.
Finally, since the top two drivers will supply the midrange, then below is the 15° off axis at 2 metres, this is the CSD (waterfall) plot:
Around 7-8KHz there will likely be some kind of LCR correction. That means our component count will be higher than "MFC" (14) and "ULD" (15) and possibly as high as 18.
From what I can see so far, the final system voltage sensitivity of the completed design will be around 89dB/1m.
I showed the Tweeter, and I thought I should show you some results of the far-field/near-field splicing and FR of the driver in situ, in the box.
This is the top two drivers, these will of course be part of an MTM array with the Tweeter.
The slight bulge around 70-80Hz is a product of the nearfield measurement, it is somewhat flatter than that, by around 2 or 3dB.
Below is the bottom two driver, note the suckout is caused by the fact that, unlike the two top drivers are co-distant, the bottom two drivers are not. The second from the bottom driver is closer and the bottom driver is slightly further away. Keep in mind that the microphone is at the height of the Tweeter and at 2 metres:
The two driver will be rolled off from 400-500Hz up first order. By the time we get to the midrange above 1-2KHz, we get a 3rd order effect and this is a good thing for the midrange. All Elsinores benefit from this trick.
Finally, since the top two drivers will supply the midrange, then below is the 15° off axis at 2 metres, this is the CSD (waterfall) plot:
Around 7-8KHz there will likely be some kind of LCR correction. That means our component count will be higher than "MFC" (14) and "ULD" (15) and possibly as high as 18.
From what I can see so far, the final system voltage sensitivity of the completed design will be around 89dB/1m.
I take it this is common between the measurements. What are the three measurements per plot, and do they each represent playing two drivers at a time?
https://www.diyaudio.com/community/threads/the-elsinore-project-thread.97043/post-7256593
A little blue tack stuff and the SB26ADC-C000-4 goes right on the Visaton waveguide with the phase plug thing remaining intact.
Thanks DT
A little blue tack stuff and the SB26ADC-C000-4 goes right on the Visaton waveguide with the phase plug thing remaining intact.
Thanks DT
Yes, I noted that. It proves it will work with a Waveguide. We are looking at modding the Elsinore waveguide right now.
"NBAC" Update:
I have now been busy for a couple of days dealing with the data current available. At this stage we don't have anything reliable to go in with regards to the SB26CDC tweeter and there is no way of saying with any certainty that it will work satisfactorily. Again, be have a lover sensitivity and lower 4 Ohm impedance compared to our well-know Peerless HDS/SS-D2608 8 Ohm tweeter.
So for now I went with the HDS/D2608 tweeter and familiarity already told be that it should not be an issue at all.
Results below.
Please note that the results below were modeled in SoundEasy V28 and not actual final measured results, but I have high confidence in the results as this is familiar territory.
As per previous Elsinores, the optimum frequency response will be 15° off axis:
Comment: Very flat response indeed. A single LCR is tuned around 7KHz on the upper two drivers that forms an MTM array with the tweeter. It was tried also on the bottom two drivers (they fill in the diffraction loss of the bass) and deemed not required, they are well rolled off by 7KHz. I am in the school of not over-correcting everything I see, and sometimes don't see (something about not throwing out the baby with the bathwater). The voltage sensitivity will be around 88dB/1m. The rise at LF are the result of the near-field measurement of the port, it can largely be ignored (besides, it faces to the rear and the rise is not in the near-field of the drivers itself). The port is also relatively free off issues (minor at 400Hz) , as both the measurement above shows as well as the impedance plot. Again, the minor 400Hz is face rear and not front towards the listener. All in all, I am very pleased.
Let us look at the impedance plot:
Comment: Why don't we see this in any commercial speakers? The lowest Z is 100Hz and just above 5 Ohm. Often, when finalised and a real measurement made, it will be even more flat at LF than the modeling, but we shall wait and see. Above 5KHz it looks like it will be even flatter than "MFC" or "ULD" versions. The reason is that the HDS/D2608 tweeter is padded down and looks even more resistive. You can see an undulation centered around 7KHz, this is where the LCR referred to earlier, shows up. Avoiding the second LCR pays off here, the undulation would otherwise be greater. Now the only question remains, is it possible that this improves the sound, to get a result flat like this? I believe it does and is part of what Elsinores listens to and likes.
Next, the electrical phase:
Comment: Pretty much flat above circa 100Hz. Does it make a difference? I believe it does and I am working on showing why and come up with results that should go along way to satisfy the critics.
Finally, the family of frequency responses:
Comment:
Grey is ON axis.
Red is the same as before, 15° OFF axis.
Green is 30° OFF axis.
This is fairly similar to "MFC" and "ULD" versions. If a bit dull in a damped room, toe in more, if a bit bright, toe it out to suit.
I like what I see.
Everything I see says to go down this route with the HDS/D2608 and the waveguide we already have. It might be a ten year old designed waveguide, but it still does exactly what I want from it.
Is there still a hankering for the SB17CDC tweeter? Even if there is, I can't make any promises.
Any comments?
Cheers, Joe
I have now been busy for a couple of days dealing with the data current available. At this stage we don't have anything reliable to go in with regards to the SB26CDC tweeter and there is no way of saying with any certainty that it will work satisfactorily. Again, be have a lover sensitivity and lower 4 Ohm impedance compared to our well-know Peerless HDS/SS-D2608 8 Ohm tweeter.
So for now I went with the HDS/D2608 tweeter and familiarity already told be that it should not be an issue at all.
Results below.
Please note that the results below were modeled in SoundEasy V28 and not actual final measured results, but I have high confidence in the results as this is familiar territory.
As per previous Elsinores, the optimum frequency response will be 15° off axis:
Comment: Very flat response indeed. A single LCR is tuned around 7KHz on the upper two drivers that forms an MTM array with the tweeter. It was tried also on the bottom two drivers (they fill in the diffraction loss of the bass) and deemed not required, they are well rolled off by 7KHz. I am in the school of not over-correcting everything I see, and sometimes don't see (something about not throwing out the baby with the bathwater). The voltage sensitivity will be around 88dB/1m. The rise at LF are the result of the near-field measurement of the port, it can largely be ignored (besides, it faces to the rear and the rise is not in the near-field of the drivers itself). The port is also relatively free off issues (minor at 400Hz) , as both the measurement above shows as well as the impedance plot. Again, the minor 400Hz is face rear and not front towards the listener. All in all, I am very pleased.
Let us look at the impedance plot:
Comment: Why don't we see this in any commercial speakers? The lowest Z is 100Hz and just above 5 Ohm. Often, when finalised and a real measurement made, it will be even more flat at LF than the modeling, but we shall wait and see. Above 5KHz it looks like it will be even flatter than "MFC" or "ULD" versions. The reason is that the HDS/D2608 tweeter is padded down and looks even more resistive. You can see an undulation centered around 7KHz, this is where the LCR referred to earlier, shows up. Avoiding the second LCR pays off here, the undulation would otherwise be greater. Now the only question remains, is it possible that this improves the sound, to get a result flat like this? I believe it does and is part of what Elsinores listens to and likes.
Next, the electrical phase:
Comment: Pretty much flat above circa 100Hz. Does it make a difference? I believe it does and I am working on showing why and come up with results that should go along way to satisfy the critics.
Finally, the family of frequency responses:
Comment:
Grey is ON axis.
Red is the same as before, 15° OFF axis.
Green is 30° OFF axis.
This is fairly similar to "MFC" and "ULD" versions. If a bit dull in a damped room, toe in more, if a bit bright, toe it out to suit.
I like what I see.
Everything I see says to go down this route with the HDS/D2608 and the waveguide we already have. It might be a ten year old designed waveguide, but it still does exactly what I want from it.
Is there still a hankering for the SB17CDC tweeter? Even if there is, I can't make any promises.
Any comments?
Cheers, Joe
Last edited: