Member
Joined 2003
Ah well. Sometimes I wonder why bother will all that painful optimization. I just did a pretty gross experiment. I switched off the mid and went directly from the 15" to the tweeter, X-O at ca. 1.7 kHz.
Lo and behold it actually sounded good - not smeared from any IMD that might (must) have been added, not "closed" from energy storage of the much higher mass 15" as compared to the usual 6.5" mid. Imaging was there, image height was correct, image depth slightly reduced.
Well, it did sound a bit grainy and also a bit tinny. But had I heard this in a store as-is I would have found it rather good. And now consider the monstrosity that this experiment actually was:
- the woofer sits at 16" or so from the ground. The tweeter at 42", for a separation of several wavelengths at X-O. But the lobing it must have produced was not really noticeable. And the mids going into the carpet didn't bother a bit - all directional cues were dominated by the tweeter, as close as 1.5m from the speakers.
- woofer fires its dipole rear radiation into wall and corner below knee height and yet there was no closed-in feeling or mud
- the 15" has a 5dB+ breakup exactly at 1700 Hz which was not eq'd. That made for the slight graininess, but nothing ear splitting
- the 15" also has a significant dB drop from 800 to 1300 Hz (when the breakup picks it up) that also went without eq. That made for the tinniness (not enough upper mids). But nothing beyond say, a change to a brighter room for instance, in terms of coloration.
- dispersion of the mid also drops around 800 to 900 Hz, but it holds acceptable pattern almost to 2k in dipole configuration, so I don't think this was a factor.
Basically the ear fills in a lot of deficiencies wonderfully well, as long as some basics are covered, and it seems to me from this experiment that past the modal range, FR irregularities, even gross, are far less bothersome than one might expect. Also, it makes me wonder how important energy storage in the driver itself really is. Maybe again it is only bothersome in some frequency range and within some specific time constant ranges.
In any case I see that taking a 15" beyond 1k is perfectly rational. A bit of eq and it would be indistinguishable, possibly better, than an additional mid cum x-o. And Paul, you should be wonderfully fine with your driver choice.
As for me I can already see where this logically leads to, budget permitting - a dome can't quite reach down to 1-1.3k which should be the ideal x-o range for the 15ND930, but a mid does not seem to be required until there. So that leaves a 1" or 1.4" compression driver, horn calculations, and trying to do a good job on the CNC.
Too bad my pockets aren't deep enough for this kind of fun at the moment...
Lo and behold it actually sounded good - not smeared from any IMD that might (must) have been added, not "closed" from energy storage of the much higher mass 15" as compared to the usual 6.5" mid. Imaging was there, image height was correct, image depth slightly reduced.
Well, it did sound a bit grainy and also a bit tinny. But had I heard this in a store as-is I would have found it rather good. And now consider the monstrosity that this experiment actually was:
- the woofer sits at 16" or so from the ground. The tweeter at 42", for a separation of several wavelengths at X-O. But the lobing it must have produced was not really noticeable. And the mids going into the carpet didn't bother a bit - all directional cues were dominated by the tweeter, as close as 1.5m from the speakers.
- woofer fires its dipole rear radiation into wall and corner below knee height and yet there was no closed-in feeling or mud
- the 15" has a 5dB+ breakup exactly at 1700 Hz which was not eq'd. That made for the slight graininess, but nothing ear splitting
- the 15" also has a significant dB drop from 800 to 1300 Hz (when the breakup picks it up) that also went without eq. That made for the tinniness (not enough upper mids). But nothing beyond say, a change to a brighter room for instance, in terms of coloration.
- dispersion of the mid also drops around 800 to 900 Hz, but it holds acceptable pattern almost to 2k in dipole configuration, so I don't think this was a factor.
Basically the ear fills in a lot of deficiencies wonderfully well, as long as some basics are covered, and it seems to me from this experiment that past the modal range, FR irregularities, even gross, are far less bothersome than one might expect. Also, it makes me wonder how important energy storage in the driver itself really is. Maybe again it is only bothersome in some frequency range and within some specific time constant ranges.
In any case I see that taking a 15" beyond 1k is perfectly rational. A bit of eq and it would be indistinguishable, possibly better, than an additional mid cum x-o. And Paul, you should be wonderfully fine with your driver choice.
As for me I can already see where this logically leads to, budget permitting - a dome can't quite reach down to 1-1.3k which should be the ideal x-o range for the 15ND930, but a mid does not seem to be required until there. So that leaves a 1" or 1.4" compression driver, horn calculations, and trying to do a good job on the CNC.
Too bad my pockets aren't deep enough for this kind of fun at the moment...
MBK - very interesting. Simple can be good sometimes. 
I looked back thru this thread for photos for a description of what you have, couldn't seem to find it. Can you repost or point me to it?
So after your experiences, let me chime in and sing the praises of the Scan Speak 8200 tweeter, which will do exactly what you want to do, but better.
The SS D3806/8200 is big 38mm tweeter that can play down to 750Hz, if you're careful and use a steep x-over. But in normal use, like an octave higher, it is easy to use and sounds great- really right. What I like about it is its harmonic structure. It just seems to get instruments right. Works wonders on piano, very, very good.
The 8200 mates well with just about any 12" driver out there for simple and great sounding 2 -way. That's what I use at work, big sealed boxes with cheap Chinese 12s and the 8200 crossed at about 1200Hz. Great sound, very loud with little power. Fills a huge room.
My buddy John in California turned me on to these tweeters. He has built many 12 & 8200 combos. And has even used them with an 18" as a 2 way system. It worked, and worked well!
The reason I mention this driver is because it sounds so good and it's a great choice to do what you are trying.
Only downside to the 8200, it ain't cheap. Goes for about $150 these days.
I looked back thru this thread for photos for a description of what you have, couldn't seem to find it. Can you repost or point me to it?
So after your experiences, let me chime in and sing the praises of the Scan Speak 8200 tweeter, which will do exactly what you want to do, but better.
The SS D3806/8200 is big 38mm tweeter that can play down to 750Hz, if you're careful and use a steep x-over. But in normal use, like an octave higher, it is easy to use and sounds great- really right. What I like about it is its harmonic structure. It just seems to get instruments right. Works wonders on piano, very, very good.
The 8200 mates well with just about any 12" driver out there for simple and great sounding 2 -way. That's what I use at work, big sealed boxes with cheap Chinese 12s and the 8200 crossed at about 1200Hz. Great sound, very loud with little power. Fills a huge room.
My buddy John in California turned me on to these tweeters. He has built many 12 & 8200 combos. And has even used them with an 18" as a 2 way system. It worked, and worked well!
The reason I mention this driver is because it sounds so good and it's a great choice to do what you are trying.
Only downside to the 8200, it ain't cheap. Goes for about $150 these days.
Paul W said:
A lot of glue, staples and patience. 4 hands make it go faster.
Basically the whole skin was stapled in place along the edges just to keep it from moving. Then a hole was cut about 8" in diameter in the middle of the woofer cut-out.
From that 16 radial cuts were made back toward the woofer-hole edge. See photo below. The flaps were pulled back around and staples in place - 1 staple per flap.
Now that the flaps were in place, we unstapled the skin edges and began gluing it to the front of the baffle. Just used ordinary carpenters glue. We spread a layer over the part we wanted to glue, and then smoothed the vinyl skin onto the baffle by hand. It was very easy to get it flat and tight.
We worked outward from the hole, section by section, until the entire face was done. Then we flipped the baffle face down and tightened the flaps.
In that step, working flap by flap, I pulled out the staple, put glue on the back of the flap then pulled it tight. 2 new staples were applied to keep it tight. Then repeat on the next flap and the next and the next. 16 times. This gave a very tight, smooth skin around the hole. See the detail in the next post. BTW, the hole had been rounded over with a router bit from both sides. You could also do that by hand with sandpaper.
They are for the most part built. Now just got to get them running. Crossovers, here I come. I will start a new thread with more details as soon as these get monsters running.
Attachments
Hi Pano,
here are previous pictures
here
and
here
The Bottom (purple cloth) has now been replaced by a very open plastic mesh, with these woofers mounted:
15ND930
I still haven't gotten around to make really "clean" photos of the entire setup.
Tweeter: Thanks for the infos. I have had thoughts about the 3806/8200 at some point. There is also the Dayton RS52, with the bonus of a grid (kid protection, important to me, that's why my speakers look like tanks) which measures very well according to Mr. Krutke aka Zaph.
Trouble is, now that I have heard the bottom/midrange end of a good pro driver, the sheer effortlessness of the presentation, the low distortion at high SPL, and the sturdiness of the construction, I am more tempted by making the top end a pro driver too. Right now I can clearly hear that it is my tweeter that gives up first on peaks.
Ideally I'd like a combo where pretty much all system components have the same limit, and I tend to assume a higher limit now, to make dynamic peaks at say 110-115 dB reachable at least in theory. The SS are also hard to get in Singapore (US ordering mandatory which adds at least USD50 to the job) while I have an 18Sound and RCF dealer that is eager to hand deliver to my house for free (go figure). Even though I just bought 1 pair of woofers from him. But all this is not really imminent, I am sure I'll spend a few months thinking over options and models...
Very thorough job on the skin. It looks like it took some patience. But it is all worth it in the end.
here are previous pictures
here
and
here
The Bottom (purple cloth) has now been replaced by a very open plastic mesh, with these woofers mounted:
15ND930
I still haven't gotten around to make really "clean" photos of the entire setup.
Tweeter: Thanks for the infos. I have had thoughts about the 3806/8200 at some point. There is also the Dayton RS52, with the bonus of a grid (kid protection, important to me, that's why my speakers look like tanks) which measures very well according to Mr. Krutke aka Zaph.
Trouble is, now that I have heard the bottom/midrange end of a good pro driver, the sheer effortlessness of the presentation, the low distortion at high SPL, and the sturdiness of the construction, I am more tempted by making the top end a pro driver too. Right now I can clearly hear that it is my tweeter that gives up first on peaks.
Ideally I'd like a combo where pretty much all system components have the same limit, and I tend to assume a higher limit now, to make dynamic peaks at say 110-115 dB reachable at least in theory. The SS are also hard to get in Singapore (US ordering mandatory which adds at least USD50 to the job) while I have an 18Sound and RCF dealer that is eager to hand deliver to my house for free (go figure). Even though I just bought 1 pair of woofers from him. But all this is not really imminent, I am sure I'll spend a few months thinking over options and models...
Very thorough job on the skin. It looks like it took some patience. But it is all worth it in the end.
MBK - Nice wood work! Very pretty. Puts my lizard skin monsters to shame.
I know what you mean about pro drivers, I love them. A lot of people are afraid of them for many reasons, size, cost and the fact that they have heard pro drivers abused at concerts where they don't sound Hi-Fi at all.
But used well, they give the most musical, real life sound of all.
I mentioned the SS 8200 because it is such a good driver, but not used very often. Maybe it looks like not a "real" tweeter, more like a mid/tweet. But it sounds so good that it makes a great top to a killer 2 way. Very true to life. Goes louder than you might think, too.
Hope you find your joy for the top of those big boys you've got. RCF and 18Sound are good choices.
I know what you mean about pro drivers, I love them. A lot of people are afraid of them for many reasons, size, cost and the fact that they have heard pro drivers abused at concerts where they don't sound Hi-Fi at all.
But used well, they give the most musical, real life sound of all.
I mentioned the SS 8200 because it is such a good driver, but not used very often. Maybe it looks like not a "real" tweeter, more like a mid/tweet. But it sounds so good that it makes a great top to a killer 2 way. Very true to life. Goes louder than you might think, too.
Hope you find your joy for the top of those big boys you've got. RCF and 18Sound are good choices.
Not at all. I think your idea is great and the finish will look professional. With the woodworking, even after 10+ coats of clearcoat it's still not at a level where I'd dare to, say, sell them as professional cabinetry. There's just always some small defect here and there unless you have unlimited tools, time, or both.
Well, I hope I don't bore you guys but I think it fits into the "large midrange for open baffle" theme: I did some more measurements to answer the question, is it rationally defensible to use a 15" pro driver for a mid rather than a 6.5" high quality hifi driver? Subjective likability is very positive, but how about data?
So I checked for distortion at equal output. Simple measurements, SPL calibration is only approximate, and of course at same SPL the 15" has much less work to do so it may seem unfair - but the question to me is on final result at equal output, not on comparing technical ingenuity.
It looks like there really is no replacement for displacement. The 15ND930 mostly wins against the SS8543, already a very low distortion driver and surprisingly, it holds up very well into treble territory.
Attached is a comparison sweep 150-2000 Hz, 15ND930 red, SS8543 blue, at an SPL equivalent to 96 dB/1m but measured at ca. 5". I am at a loss wrt the large peak in the 600 Hz range for the SS8543, I suspect a mounting hardware buzz or possibly the metal grille that I mounted in front of it, in other words an artefact. The 15ND930 is remarkable in its constant ca. 0.5% distortion figure over such a wide bandwidth - those demodulation rings seem to work rather well. In absolute terms distortion is very low for such a high level anyway, for both drivers. Level tests show that for both drivers, THD rises monotonically for all frequencies equally, when you increase levels. While at 96dB both drivers hover at the 0.5-0.6% level, at 90 dB they drop to ca. 0.2-0.3% (not shown).
Attachments
And I also wanted to see the distortion spectrum, especially since I was puzzled that THD should be so relatively constant over the bandwidth. But the actual spectrum was consistent with the sweep. For example, here is the 15ND930 at 800 Hz/96dB/1m (ignore the IMD text at the bottom, I did not set the parameters necessary to make the IMD part meaningful)
Attachments
Interestingly, or sadly, even at reasonable SPLs the tweeter (Seas 27TBFCG) is no match THD wise to either woofers. I am somehow confused by this, but the numbers correspond roughly to Zaph's results for this tweeter at same SPL, 0.5% or about -46 dB at 90dB/1m. Any higher in level and you go to 1 or 2% real quick. Attached the SS8543 sweep and the overlapping 27TBFCG sweep.
Energy storage and thus, transparency, is another can or worms of course. But subjectively it doesn't seem to be an issue, quite the contrary. Dispersion aside, which is extended by dipole application, there seems to be no penalty in running a good 15" well into kHz range, as compared to a typical sized mid.
Energy storage and thus, transparency, is another can or worms of course. But subjectively it doesn't seem to be an issue, quite the contrary. Dispersion aside, which is extended by dipole application, there seems to be no penalty in running a good 15" well into kHz range, as compared to a typical sized mid.
Attachments
Member
Joined 2003
That's for sure! At a Steely Dan concert a couple of years ago, sometimes we couldn't even understand which song they were playing! That concert was definitely top of mind when I ordered pro drivers!
15" mid...dipole version of Summa
. HD for the 15ND930 looks good! What about CSD and more complex IM measurements? If it is good up to 1k, a 1" compression driver becomes a reasonable match. If not a 1", a 1.4 or 1.5" should do for everything but bats.Finding or building the right waveguide might be the hardest part. The DDS 90 doesn't hold pattern control low enough for a 1k/15" XO and (I understand) the 18 Sound waveguides use a diffraction slit in the throats.
Yes... I don't know if I should be pleased to come to similar conclusions as some more highly qualified people than me, or feel silly about it. OTOH - it's not quite the same. Dipole would have a greatly improved power response vs. f wrt the Summa, and I am also not so sure about Geddes' rationale for eliminating side reflections by designin for off axis listening. The x-0 xould be as high as 1.2 or 1.3k. And yes, the waveguide is the challenge. Have since found a spreadsheet for oblate spheroid contour, and an article describing Peavey quadratic. A quadratic equation produces a parabola, and we all know that these are used in reflectors, in other words, this should be a theoretically ideal profile to make sure any waves are always reflected to a path parallel to the axis (hence no production of the HOM that Geddes is concerned about).
I do wonder about the loading though - a "perfect" match to a dipole would require a 120 degree coverage (-6 dB at 60 degrees, like a dipole). This is very shallow, and for power reasons it might not be enough loading for a 1" compression driver. Now, I suspect any difference between Peavey quadratic and OS might lie in which part of reality is deemed "negligible", but the quadratic seems to be a tad deeper, so maybe this should be tried first.
The distortion measurements are primitive for now, to get a first guess. Audiotester can not generate, or import for that matter, multi tones or tone bursts, so no SL style HD and linear decay measurements. It can do dual tones for IMD however, I'll start playing with that at some point, and regular CSD of course, with all its windowing artefacts. What is the software you are using, Paul?
Member
Joined 2003
I use Soundeasy for measurements. Because of the user documentation, that is the only thing I use Soundeasy for.
LspCAD Pro for crossover work, the Edge for dipole baffles, BDS for monopole baffles, several other FRD Consortium tools for arrays etc.
The software referenced in this thread looks very interesting but haven't tried it yet.
software link
OK, quite a handful ;-) LSPCad would be nice for accurate XO estimation. I work with trial and error...
Well I've tried many packages as demos but in the end I only got over myself to pay for Audiotester. It has its quirks and limitations but does measurements as well as others and has the best price/feature ratio for my needs. Praxis seems to be the most versatile, they do have the simultaneous FR and HD method implemented that you mention in the thread, and you can do SL style bursts. But they force you to buy a mic and preamp to get all functions, in demo it is very basic, and the price is not even on the website.
So, have you progressed on the integration of the waveguide to the mids? Are you sticking with the simpler one, that holds the pattern lower, or are you moving the x-o up?
Well I've tried many packages as demos but in the end I only got over myself to pay for Audiotester. It has its quirks and limitations but does measurements as well as others and has the best price/feature ratio for my needs. Praxis seems to be the most versatile, they do have the simultaneous FR and HD method implemented that you mention in the thread, and you can do SL style bursts. But they force you to buy a mic and preamp to get all functions, in demo it is very basic, and the price is not even on the website.
So, have you progressed on the integration of the waveguide to the mids? Are you sticking with the simpler one, that holds the pattern lower, or are you moving the x-o up?
Member
Joined 2003
Yup, wish I had Praxis...unfortunately it is significantly more expensive so couldn't justify it.
Though the inexpensive WG looks promising, the decision is not settled yet. Spare time last weekend was spent with the DDS trying to identify HOMs (without much success). One anomaly did show up, but only in narrow frequency bands. (upper trace is the input signal, lower trace is microphone output)
The first shot is a very high amplitude pulse with the mic in line with the 45 degree wall of the WG. Notice the apparent compression on one half of the waveform.
The second shot is an identical pulse, identical mic location, but with the WG filled with open cell foam.
Any thoughts about this comparison?
For the moment, the DDS WGs are in service with a 1.5k XO. The other set of baffles have been modified so they can accept the 18sound mids for quick A-B comparisons of the waveguides. (Before, Seas mids were paired with the 12" WGs, and 18sound mids with DDS WGs.) I hope to make at least a preliminary WG decision after more tests this coming weekend.
Though the inexpensive WG looks promising, the decision is not settled yet. Spare time last weekend was spent with the DDS trying to identify HOMs (without much success). One anomaly did show up, but only in narrow frequency bands. (upper trace is the input signal, lower trace is microphone output)
The first shot is a very high amplitude pulse with the mic in line with the 45 degree wall of the WG. Notice the apparent compression on one half of the waveform.
An externally hosted image should be here but it was not working when we last tested it.
The second shot is an identical pulse, identical mic location, but with the WG filled with open cell foam.
An externally hosted image should be here but it was not working when we last tested it.
Any thoughts about this comparison?
For the moment, the DDS WGs are in service with a 1.5k XO. The other set of baffles have been modified so they can accept the 18sound mids for quick A-B comparisons of the waveguides. (Before, Seas mids were paired with the 12" WGs, and 18sound mids with DDS WGs.) I hope to make at least a preliminary WG decision after more tests this coming weekend.
Very clean looking signals in any case. The first one seems to have some cancellation, plus some ringing. THis could be internal reflections (HOM) or simple 1/4 mouth reflections, both would be attenuated by the foam. Maybe a combination of both. But I'd guess 1/4 wave, because only one half of the signal is attenuated, it doesn't seem to be messed up at random. Funy how even the treated signal is a full double as long as the input signal.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.