I don't understand why people don't just upload images in their posts. It prevents all the bs with image hosting sites going down in the future and your posts losing their images. I find so many forum posts (not just here) where that has happened, and what could have been a useful source of information is instead completely worthless. If you put your images in the post you create, the only way the content you create here will disappear is if diyaudio disappears, but in that case everything is gone, not just the image.
Member
Joined 2003
Sorry about the firestorm!
I actually didn't intend to show any pic...website is just in my signature. Hadn't looked at the site in a long time and didn't realize photobucket was screwing it up (now planning to get rid of them).
Here is the pic that apparently started the conversation. FR of my LCRs from the MLP, 10Hz to 24kHz vs target. Except just below 100 and 200 Hz, +/-1db from 10Hz to the target 20k rolloff.
I actually didn't intend to show any pic...website is just in my signature. Hadn't looked at the site in a long time and didn't realize photobucket was screwing it up (now planning to get rid of them).
Here is the pic that apparently started the conversation. FR of my LCRs from the MLP, 10Hz to 24kHz vs target. Except just below 100 and 200 Hz, +/-1db from 10Hz to the target 20k rolloff.
Attachments
Member
Joined 2003
For your entertainment I will share my almost-finished project here. I received much help from many kind members here on diyaudio and in Ottawa, and feel the least I can do is give a little back to this community.
In 2013 my brother Daniel and I decided to upgrade our speakers to "the next level." He owned Totem Mani-2, and I a pair of DIY MTM towers designed by André Perreault (E-Speakers) around the PHL 1340 mid-woofers and Aurus Cantum G2 ribbons. We both enjoy spinning vinyl, playing the piano, and attending jazz, rock and classical concert. In may ways we have similar taste. So I took on the design and building role while he contributed a critical ear, shared the cost and provided a challenge function that many times led us back to the drawing board.
We were not sure at first what type of speaker we wanted. The qualities we were aiming for are good subjective tonal balance in our medium sized living rooms; extended dynamic range; well-behaved load for tube amplification; size and weight fit for domestic use and acceptability by significant others.
We went to an audio show, spoke with fellow DIYers, tried different drivers and eventually settled on the TAD 1601B and the Radian 745NeoBPB-16 Beryllium drivers. We spent 90% of our time iterating the passive crossover, measuring and listening. I thought the project would be over in a year, and working at a casual pace it took us four.
Daniel’s speakers are finished, were auditioned at the Ottawa DIY Event in April, The Greater Ottawa and Environs DIY Audio 2017 Get together. and now reside in his living room in Montreal. Mine are nearly complete. I have most of the parts and just need some time to finish them.
In 2013 my brother Daniel and I decided to upgrade our speakers to "the next level." He owned Totem Mani-2, and I a pair of DIY MTM towers designed by André Perreault (E-Speakers) around the PHL 1340 mid-woofers and Aurus Cantum G2 ribbons. We both enjoy spinning vinyl, playing the piano, and attending jazz, rock and classical concert. In may ways we have similar taste. So I took on the design and building role while he contributed a critical ear, shared the cost and provided a challenge function that many times led us back to the drawing board.
We were not sure at first what type of speaker we wanted. The qualities we were aiming for are good subjective tonal balance in our medium sized living rooms; extended dynamic range; well-behaved load for tube amplification; size and weight fit for domestic use and acceptability by significant others.
We went to an audio show, spoke with fellow DIYers, tried different drivers and eventually settled on the TAD 1601B and the Radian 745NeoBPB-16 Beryllium drivers. We spent 90% of our time iterating the passive crossover, measuring and listening. I thought the project would be over in a year, and working at a casual pace it took us four.
Daniel’s speakers are finished, were auditioned at the Ottawa DIY Event in April, The Greater Ottawa and Environs DIY Audio 2017 Get together. and now reside in his living room in Montreal. Mine are nearly complete. I have most of the parts and just need some time to finish them.
High pass filter
Our first build was a three way prototype completed in early 2014. The filter was a 6th order electrical L-R with Zobel for the TAD - similar to the TAD TN-2 filter, and a 4th order electrical L-R on the Yamaha JA-6681B midrange-tweeter. This Yamaha uses a unique high compliance, low fatigue beryllium/copper suspension (patented) and has a 35 mm (1.38 inch) throat. For super tweeter duty I relied on RAALs high pass filtered above 10 kHz. On the bright side, the extended dynamic range conveyed a natural, unconstrained delivery on well recorded music. Somehow live recordings were more enjoyable than studio recordings.
Version 1 of the crossover had many flaws though.
1. A lack of low end.
2. Poor top end spatial integration (3 way) or dull top end (super tweeter removed)
3. Overemphasized presence/ intelligibility.
4. Blurry imaging.
5. Cupped hands coloration.
Over two years we came up with the following solutions.
1. Added baffle step compensation.
2. Chose a 2 way over a 3 way configuration, and added a resistor and a cap to equalize the response of a Radian Beryllium driver.
3. Attenuated the 3 to 6 kHz bump in the response of the Radian Be driver. The Radian is efficient enough to be equalized passively. Since it was not selected for this reason, we got lucky.
4. Stiffer cabinets, with dampening to reduce the midrange back wave. The final cabinets have a thick bamboo baffle and curved plywood sides.
5. Lowered the Q of the high pass filter and adjusted the low pass filter to equalize the response of the TAD.
The most recent high pass filter for the Radian fitted to the Azura AH425 horn is shown below.
For newcomers to this thread, the Azura Horn AH425 is a medium-size horn that is the result of a collaboration started between Lynn Olson, Jean-Michel Le Cleac'h, Bjorn Kolbrek (who performed detailed BEM simulations) and Martin Seddon. Among its virtues, opening the horn angle reduces the reflection of waves from mouth to throat. The angle of the AH425 throat matches that of the Altec 288 and similar drivers, and its size and cutoff frequency allow integration with a large diameter midwoofer. Martin built a first pair for me in the summer 2013. His horns are superb in every way.
To equalize the response of the Radian driver, instead of adding a resistive L-pad, I chose to use the taps of the autotransformer already in the high-pass circuit. Simulations provided a starting point in the right ballpark for bypass capacitor and resistor values, but empirical tests were essential here. Choosing different autotransformer taps does not only adjust the level of the high frequencies, but also the shape of the response of the “notch” RLC circuit resultant from the bypass cap. So the three variables interacting with each other are the cap value, resistor value, and choice of attenuation taps. Finding a sweet spot between the three took a while. Fortunately measurements are straightforward in this frequency range.
As Lynn pointed out, the HF bypass cap idea probably goes back to the Altec Model 19. This is nothing new, but since the Beryllium diaphragm has good self-damping, we thought this approach should work quite well with this particular driver. Indeed it did.
What sounded best to my ears is an EQ circuit that attenuates lower frequencies by about 10 dB relative to the driver’s output at 15 kHz. The resulting response is flat-ish around 10 degrees off axis, and rises about 3 dB on axis. My hypothesis is that the rising on axis response subjectively makes up for the narrower HF dispersion pattern of this particular horn profile – which is not constant directivity.
The EQ circuit plays another equally important role. The Radian Be exhibits a bump in the “presence region” of its frequency response. This is a common flaw in an otherwise excellent driver. Some may like this sound – there is a reason why this region is called “presence.” Personally I find this objectionable and prefer the more neutral presentation obtained when the EQ circuit is in place.
Our first build was a three way prototype completed in early 2014. The filter was a 6th order electrical L-R with Zobel for the TAD - similar to the TAD TN-2 filter, and a 4th order electrical L-R on the Yamaha JA-6681B midrange-tweeter. This Yamaha uses a unique high compliance, low fatigue beryllium/copper suspension (patented) and has a 35 mm (1.38 inch) throat. For super tweeter duty I relied on RAALs high pass filtered above 10 kHz. On the bright side, the extended dynamic range conveyed a natural, unconstrained delivery on well recorded music. Somehow live recordings were more enjoyable than studio recordings.
Version 1 of the crossover had many flaws though.
1. A lack of low end.
2. Poor top end spatial integration (3 way) or dull top end (super tweeter removed)
3. Overemphasized presence/ intelligibility.
4. Blurry imaging.
5. Cupped hands coloration.
Over two years we came up with the following solutions.
1. Added baffle step compensation.
2. Chose a 2 way over a 3 way configuration, and added a resistor and a cap to equalize the response of a Radian Beryllium driver.
3. Attenuated the 3 to 6 kHz bump in the response of the Radian Be driver. The Radian is efficient enough to be equalized passively. Since it was not selected for this reason, we got lucky.
4. Stiffer cabinets, with dampening to reduce the midrange back wave. The final cabinets have a thick bamboo baffle and curved plywood sides.
5. Lowered the Q of the high pass filter and adjusted the low pass filter to equalize the response of the TAD.
The most recent high pass filter for the Radian fitted to the Azura AH425 horn is shown below.
For newcomers to this thread, the Azura Horn AH425 is a medium-size horn that is the result of a collaboration started between Lynn Olson, Jean-Michel Le Cleac'h, Bjorn Kolbrek (who performed detailed BEM simulations) and Martin Seddon. Among its virtues, opening the horn angle reduces the reflection of waves from mouth to throat. The angle of the AH425 throat matches that of the Altec 288 and similar drivers, and its size and cutoff frequency allow integration with a large diameter midwoofer. Martin built a first pair for me in the summer 2013. His horns are superb in every way.
To equalize the response of the Radian driver, instead of adding a resistive L-pad, I chose to use the taps of the autotransformer already in the high-pass circuit. Simulations provided a starting point in the right ballpark for bypass capacitor and resistor values, but empirical tests were essential here. Choosing different autotransformer taps does not only adjust the level of the high frequencies, but also the shape of the response of the “notch” RLC circuit resultant from the bypass cap. So the three variables interacting with each other are the cap value, resistor value, and choice of attenuation taps. Finding a sweet spot between the three took a while. Fortunately measurements are straightforward in this frequency range.
As Lynn pointed out, the HF bypass cap idea probably goes back to the Altec Model 19. This is nothing new, but since the Beryllium diaphragm has good self-damping, we thought this approach should work quite well with this particular driver. Indeed it did.
What sounded best to my ears is an EQ circuit that attenuates lower frequencies by about 10 dB relative to the driver’s output at 15 kHz. The resulting response is flat-ish around 10 degrees off axis, and rises about 3 dB on axis. My hypothesis is that the rising on axis response subjectively makes up for the narrower HF dispersion pattern of this particular horn profile – which is not constant directivity.
The EQ circuit plays another equally important role. The Radian Be exhibits a bump in the “presence region” of its frequency response. This is a common flaw in an otherwise excellent driver. Some may like this sound – there is a reason why this region is called “presence.” Personally I find this objectionable and prefer the more neutral presentation obtained when the EQ circuit is in place.
Attachments
Last edited:
Acknowledgements
Please allow me to acknowledge and thank the following individuals.
Lynn Olson for inspiration and sharing is wealth of experience with great patience.
Mark Kravchencko and other local Ottawa DIY enthusiasts who have been helpful throughout the project. Mark has extensive speaker design experience and guided me from the start. Together we considered many design alternatives, including an Iwata horn, Faital Pro HF144 driver, B&C 15NA100 woofer and so on. He measured the final drivers with his Smith & Larsen Audio gear. Mark built the prototype boxes and the final plywood and bamboo enclosures. I was lucky to count on his generous help and good spirits.
Jean-Michel LeCléac’h (rest in peace) for conceiving the JMLC horn profile and documenting tests of the TAD TL1601 15 inch driver in a sealed enclosure.
David dal Farra, also a local DIYer, gave me wonderful advice on crossover design. Well, no. He rescued a clueless beginner. David has a well-trained ear, and can pick up flaws in a speaker design within a few minutes of listening to recordings he knows. He diagnosed the need for baffle step compensation, and led me to build a BSC test box that I inserted between preamp and power amp to test various compensation curves. In the final filter, the compensation is implemented as a resistor across the first coil in the low pass filter. Sometimes simplicity works like a charm. David was able to hear that the Radians Be have some useable top end and would not need a super tweeter. He was right of course, and following his advice I brought up the top end through passive EQ and settled on a two way design.
Martin King for making wonderful fiberglass horns, and sharing his measurements. The Yamaha A6681B response on the AH425 is simply incredible.
Gary Dahl who, working with Lynn, shared his crossover topology early as well as first impressions of the then newly released Beryllium Radian drivers. For his good advice and encouragements too.
Please allow me to acknowledge and thank the following individuals.
Lynn Olson for inspiration and sharing is wealth of experience with great patience.
Mark Kravchencko and other local Ottawa DIY enthusiasts who have been helpful throughout the project. Mark has extensive speaker design experience and guided me from the start. Together we considered many design alternatives, including an Iwata horn, Faital Pro HF144 driver, B&C 15NA100 woofer and so on. He measured the final drivers with his Smith & Larsen Audio gear. Mark built the prototype boxes and the final plywood and bamboo enclosures. I was lucky to count on his generous help and good spirits.
Jean-Michel LeCléac’h (rest in peace) for conceiving the JMLC horn profile and documenting tests of the TAD TL1601 15 inch driver in a sealed enclosure.
David dal Farra, also a local DIYer, gave me wonderful advice on crossover design. Well, no. He rescued a clueless beginner. David has a well-trained ear, and can pick up flaws in a speaker design within a few minutes of listening to recordings he knows. He diagnosed the need for baffle step compensation, and led me to build a BSC test box that I inserted between preamp and power amp to test various compensation curves. In the final filter, the compensation is implemented as a resistor across the first coil in the low pass filter. Sometimes simplicity works like a charm. David was able to hear that the Radians Be have some useable top end and would not need a super tweeter. He was right of course, and following his advice I brought up the top end through passive EQ and settled on a two way design.
Martin King for making wonderful fiberglass horns, and sharing his measurements. The Yamaha A6681B response on the AH425 is simply incredible.
Gary Dahl who, working with Lynn, shared his crossover topology early as well as first impressions of the then newly released Beryllium Radian drivers. For his good advice and encouragements too.
Last edited:
I agree, your explanation makes a lot of sense to me. In fact, this is very similar behaviour to some well-regarded "full-range" cone drivers.
Instead, with my radial horns, which are almost constant-directivity in the horizontal plane from ~300 to ~6k Hz, I find that I need an opposite, downward tilt for them to sound balanced.
Marco
Last edited:
I see that you ended up using similar 4th order slopes on the LP and HP branches.
From the picture you posted, however, I see a significant front-to-back offset between the woofer and the compression driver.
My question is: how did you get the two drivers to "phase track" over a reasonable frequency range around the crossover point?
In fact: did you?
In my experience, achieving good phase tracking with that kind of offset typically calls for a very asymmetrical crossover (like the TAD's 6th order LP and 2nd order HP, for instance).
Marco
I think this is the typical case of "short blanket" (you win something, you lose something).
In my experience, no matter how good Be is, a properly integrated (*) quality super-tweeter makes for a more realistic result, also thanks to the resulting more even directivity.
(*) Again, in my experience, at super-tweeter frequencies, "proper integration" only requires that one obtain a smooth transition and the overall desired frequency response when measured with 1/3rd octave smoothing.
While phase matching is critical at lower frequencies, it just doesn't seem to matter beyond 5kHz or so.
So, which were the factors that led you to prefer the Radian compression driver in the end?
"Just" the top end extension? Or also a better timbre in the mid frequencies?
Or other reasons?
That may depend on your hearing or your tastes.
I almost always found that super tweeters sound fake to my ears. They may sound fun and add some certain "something" to the overall feel, but it's not something I hear in natural sounds. They sound fake, perhaps pleasingly so, but fake.I do agree that hearing phase differences way up high is difficult, if not impossible.
However, as I've gotten much older and my ears much worse, super tweeters don't bother me as much. Go figure.
One of my guilty fantasies (no, not those) is somehow combining an Altec/GPA 604 Duplex with the AH425/Radian 745P. I admit, it's mostly because it would look really cool with the 16" AH425 sitting in free air directly above the 604 Duplex, with another 416 sitting below the Duplex. A triple 16" stack, as it were.
The hand-waving rationalization is the 604/AH425 combination lets me use the 604 horn as a supertweeter (coming in around 5 kHz), and lowpassing the 604 woofer around 650~750 Hz, which avoids the really rough 1~2 kHz crossover region of the Duplex. I know the 604 HF horn is hardly a supertweeter in any sense of the word, but since the leading-edge transients would be coming from it, it would act as localizer that's conveniently concentric with the 604 woofer. It would also improve dispersion of the AH425 (well, a little bit, anyway).
This is one of those head versus heart things. I'm intuitively drawn to the idea, but mostly because it's kind of offbeat and zany-looking. I could probably force-fit the crossovers, but the phasing between the three drivers would be tricky, since the front-to-back distance of the 604 HF horn is fixed in space. The free variables are the crossover and the adjustable front-to-back distance of the AH425. It might even sound best with the 604 HF horn working in parallel with the AH425 horn (above 5~7 kHz).
Is this idea ridiculous, or is there any merit to it? It wouid certainly sound Altec, of course, but could it be tamed into a decent loudspeaker?
The hand-waving rationalization is the 604/AH425 combination lets me use the 604 horn as a supertweeter (coming in around 5 kHz), and lowpassing the 604 woofer around 650~750 Hz, which avoids the really rough 1~2 kHz crossover region of the Duplex. I know the 604 HF horn is hardly a supertweeter in any sense of the word, but since the leading-edge transients would be coming from it, it would act as localizer that's conveniently concentric with the 604 woofer. It would also improve dispersion of the AH425 (well, a little bit, anyway).
This is one of those head versus heart things. I'm intuitively drawn to the idea, but mostly because it's kind of offbeat and zany-looking. I could probably force-fit the crossovers, but the phasing between the three drivers would be tricky, since the front-to-back distance of the 604 HF horn is fixed in space. The free variables are the crossover and the adjustable front-to-back distance of the AH425. It might even sound best with the 604 HF horn working in parallel with the AH425 horn (above 5~7 kHz).
Is this idea ridiculous, or is there any merit to it? It wouid certainly sound Altec, of course, but could it be tamed into a decent loudspeaker?
Last edited:
I think the idea has merit in principle. However, I'm not sure the Altec compression driver that is part and parcel of the 604 coax 'duplex' would make for a particularly good supertweeter (because of diaphragm break-up, imperfect phase plug, etc.).
Now, if you could somehow replace that part of the 'duplex' with a TAD TD-2001, or even better a TD-2002... then you'd really be on to something! (those two are IMHO the only compression drivers that can really compete with the best dedicated supertweeters in the top-most octave).
Marco
[EDIT] P.S.: I just googled "TAD coax" and came across these. Not sure what make the woofer is (definitely not a TAD, and I don't think an Altec either). But it might be worth a punt, perhaps.
Now, if you could somehow replace that part of the 'duplex' with a TAD TD-2001, or even better a TD-2002... then you'd really be on to something! (those two are IMHO the only compression drivers that can really compete with the best dedicated supertweeters in the top-most octave).
Marco
[EDIT] P.S.: I just googled "TAD coax" and came across these. Not sure what make the woofer is (definitely not a TAD, and I don't think an Altec either). But it might be worth a punt, perhaps.
Last edited:
I took the liberty of reprinting the full PierreQuiRoule post because it has such a lot of excellent content, and hard-won experience (complete with measurements).
The idle speculation I posted about the 604 Duplex/AH425 combination got me thinking about PierreQuiRoule's experience with baffle-step compensation. What's really going on with BSC is that the woofer radiation is widening out from about 90 degrees at the very top of its range to a full 360 degrees in the bottom octaves. This is unavoidable due to the increasing size of wavelengths relative to the fixed size of the emitting surface (the cone). With the speed of sound about 344 m/sec, that means 800 Hz has a wavelength of 43 cm, or 16.9". I've measured the paper cones of a pair of GPA 515's I have in the basement, and they are (surprisingly) 12.125" or 30.8 cm across (this excludes the surround, which contributes no desirable radiation at 800 Hz). The emitting surface is about 0.72 wavelength at 800 Hz, so dispersion somewhat wider than 90 degrees is to be expected.
Step down three octaves to 100 Hz, and the wavelength is now 8x bigger (expanding spherically into 3D space), at 3.44 meters, or 135". Both the woofer cone and speaker enclosure are much smaller than this, so radiation is effectively 360 degrees into a sphere. This spreads out the power, so less reaches the on-axis listener (a light bulb instead of a flashlight).
There's no way around this; one of the awkward things about audio is the extremely wide bandwidth compared to color vision (which is less than an octave) or radio transmissions, where a bandwidth of an octave is considered extremely wide. A woofer ten feet across isn't practical, and only a few really far-out horn builders have built bass horns with mouths that size.
But when we look a little closer at radiation into a sphere at low frequencies, we have to consider reflections off the nearby walls .. the wall behind the speaker, the wall to the left or right, and the floor. At 100 Hz, they are nearly perfect reflectors, with less than 10~20% loss, and carpeting has very little absorption at those frequencies. If you could see it, it would look like a hall of mirrors, with one real speaker and 7 reflections.
At higher frequencies (like 800 Hz), these reflections are incoherent and in random phase. At 100 Hz, though, they are starting to become coherent, and adding together in phase. As the distance to the nearest reflection starts to become small relative to the wavelength, the woofer also "sees" its mirror image as another driver, and acoustic coupling is improved. By 50 Hz, everything is in-phase and the emitting driver sees all of its nearby reflections, unless the room is very large.
This is the reason for so-called "room gain". The same thing would happen outdoors if the speaker were in a basketball court with two side walls and the ground of the court ... at low frequencies, it's simply a corner reflector, or a 3-sided conical horn.
What's the effect on baffle step correction? It means the only time we need a complete, full-on BSC is if the speaker is going to be hanging in free space, 20 feet off the ground. Once there are nearby walls, "room gain" partially offsets BSC and eventually overpowers it at the lowest frequencies. In real rooms, there can be sort of a rough dip in the 100~300 Hz region, and this is where a BSC circuit comes in handy. But in reality, response is going up and down in this frequency range, as the relative phasing of the first-order reflections changes with frequency. At higher frequencies (typically above 500 Hz), the reflections become so dense they average out.
This long digression aside, I was thinking about the two ways to implement BSC. One is passive filtering, which is subtractive (it decreases efficiency over part of the passband). The other approach is bringing in another LF driver, adjusting the lowpass filter so it compensates for BSC. This is room-dependent, so the lowpass filter will need to have taps, or be adjustable. In practice, this is a subjective adjustment, since every recording has a different set of balances in the bass region. (Recordings have a final master-balance phase, and the rooms where this is done have a variety of monitor speakers with their own set of equalizers in place.)
Personally, I think it's long overdue for the return of bass and treble tone controls, this time around with a limited +/- 6 dB range, and good phase tracking between channels.
Last edited:
Big Ugly Speakers
Here's my BUS ... Big Ugly Speakers. Quite a mess to look at, rather nice to hear.
Bass cabinet somewhere around 17 cu ft. (480 liters)
604 coax bass and tweeter
Altec 803 multicell horn + 1.4" Radian driver - early 90s vintage
Altec 811B horn with Altec 1" 803 driver
Sometimes sealed, sometimes vented. I didn't use both mid horns at once, but swapped off. Always used the 604 tweeter.
Here's my BUS ... Big Ugly Speakers. Quite a mess to look at, rather nice to hear.
Bass cabinet somewhere around 17 cu ft. (480 liters)
604 coax bass and tweeter
Altec 803 multicell horn + 1.4" Radian driver - early 90s vintage
Altec 811B horn with Altec 1" 803 driver
Sometimes sealed, sometimes vented. I didn't use both mid horns at once, but swapped off. Always used the 604 tweeter.
