Tweet
Boy, am I going to catch h@ll from Eric about this one... 
This is one of those little projects that just kind of sneaks up on you... the precursor is the EM7 I did in 2003 for my daugher, a small MT with the HiVi D6.8 woofer and XT25 speaker, to replace the speakers we built when she was about 12 that were damaged at her mother's house. I took those speakers to the Northern CA DIY in 2003, and with the moderately prodigous bass output the little HiVi Dynaudio Klone woofer has (a clone right down to 3" VC and neodyminum magnets), it was probably the most full range small system exhibited. But there were things I wasn't totally happy with in the upper midrange and presence region. This was speaker an attempt at doing some sort of series crossover, just for fun.
Now, what I'm going to show next is more a design study than a project, as it's just a few hours work riding home on the train in the evening, using data from one of the Christmas time measuring sessions, for the Peerless 850439 and SEAS 27TDFC. I picked up two pairs of the Peerless woofers last year for about half price, and one pair is destined for a speaker like this for an old friend back east. The other set may wind up in an MT for rears or bedroom speakers or whatever. They'll use the same type PE cabinet as my daughter's but this time I bought Cherry cabinets.
Strangely, I've had several requests recently about an MT with the Peerless or Dayton 7's, something not quite as "complicated" as the Modula MTM. With the upper range breakup modes in the aluminum cone Daytons, I don't think they're a good candidate for a simplified speaker. But perhaps I've come up with something that will work, with reduced component count.
The crossover is shown below- it's somewhat different than the one for my daughter's speaker. Well, actually, it's a lot different; one of the things about LspCAD 6 is that it's completely free-form in network drawing, and is free of the limitations version 5 has for modeling strange configurations in a series crossover. So, my imagination could cut loose a lot more with this one. Actually, I managed to retain a "trademark" feature- this crossover does use a partial cauer-elliptic filter section on the low pass, to help suppress the upper end peak of the midwoofer at 4.2 kHz.
I wasn't really sure whether this was going to work out, as with a series crossover the components interact in their affects on both drivers, but I'm somewhat optimistic now, especially given the robust characteristics of the 27TDFC, and the modeling possible with LspCAD6.
The acoustical target for each driver is third order, but with L-R coefficients, not Butterworth. As designed for the physical offset of acoustical centers, if the tweeter is connected out of phase, a deep null appears at the nominal crossover point of 1850 Hz. This isn't a Butterworth alignment, because both polarities would work for on axis response, with the usual lobe off axis. Also, the main crossosver lobe in the vertical direction is almost straight forward- also a good thing. But doing this requires some compromise in the front axial response, as suggested by Robert Bullock for the conventional 4th order L-R.
The tweeter and woofer data used was taken in a 9" wide cabinet, but smaller and not ported, so the LF response is not as the final design in the PE cabinet. Due to the pressure zone B&K capsule, there's something of a measurement artefact hard to avoid someitmes, so ignore the roll off above 16 kHz.
The intended use for these speakers is for a non-audiophile friend, and more for background music listening or off axis use, so the power response is much more important than the front axial response- with this higher crossover frequency, and using a modified series network, it takes some care to optimize this, and the final results will not be as ideal as with one of my more complex parallel cauer filters. But the component count and cost is about half, a not inconsiderable benefit. So, be aware of the trade-offs.
L4, C8, and R9 together with L2 and C1 and R8 determine the LF roll off characteristic for the midwoofer, including baffle step compensation; L4, C8, and R9 form the elliptic portion of the filter. C1, L3, R3, R4 and R8 determine the tweeter filter high pass - no notch or elliptic filter here.
This plot gives an overview of axis and off axis curves, and the net predicted power response.
How many of you remember when Luxman tried to introduce a speaker, and brought out a two way model around 1976, using a kevlar midwoofer? B&W, which had their distribution in Japan through Luxman, was furious, and pressured Luxman into withdrawing that speaker from the market. It was a very interesting little bookshelf- much better sounding than the comparable B&W models we had in the shop at that time- very nice imaging, too. It didn't measure all that flat on axis, but the overall power response was quite smooth...
Here's a plot of the predicted impedance curve; this looks pretty ideal for the average solid state amp; 5-10 ohms over much of the range, pretty much ideal for getting juice of out midrange gear. Fairly easy phase angle, too.
The highish (by my standards) crossover point results in some narrowing of the polar plot in around 1500 Hz, just before the crossover, but compared with most conventional two ways, the predicted performance looks quite good. It would suffer only in comparison with something like the Modula MTM, I think, with it's lower crossover point.
This won't have the prodigous low bass output capability of an Extremis woofer, but then it's about 4 dB more efficient, and should have nice characteristics down to the high 30's, expecting about -4 dB at 34 - 36 Hz, using a 2" flared Precision Port. :B
Every now and then I have to throw a curve ball just to keep you guys on your toes...
Thanks to Dennis H for catching some mistakes in my post!
This is one of those little projects that just kind of sneaks up on you... the precursor is the EM7 I did in 2003 for my daugher, a small MT with the HiVi D6.8 woofer and XT25 speaker, to replace the speakers we built when she was about 12 that were damaged at her mother's house. I took those speakers to the Northern CA DIY in 2003, and with the moderately prodigous bass output the little HiVi Dynaudio Klone woofer has (a clone right down to 3" VC and neodyminum magnets), it was probably the most full range small system exhibited. But there were things I wasn't totally happy with in the upper midrange and presence region. This was speaker an attempt at doing some sort of series crossover, just for fun.
Now, what I'm going to show next is more a design study than a project, as it's just a few hours work riding home on the train in the evening, using data from one of the Christmas time measuring sessions, for the Peerless 850439 and SEAS 27TDFC. I picked up two pairs of the Peerless woofers last year for about half price, and one pair is destined for a speaker like this for an old friend back east. The other set may wind up in an MT for rears or bedroom speakers or whatever. They'll use the same type PE cabinet as my daughter's but this time I bought Cherry cabinets.
Strangely, I've had several requests recently about an MT with the Peerless or Dayton 7's, something not quite as "complicated" as the Modula MTM. With the upper range breakup modes in the aluminum cone Daytons, I don't think they're a good candidate for a simplified speaker. But perhaps I've come up with something that will work, with reduced component count.
The crossover is shown below- it's somewhat different than the one for my daughter's speaker. Well, actually, it's a lot different; one of the things about LspCAD 6 is that it's completely free-form in network drawing, and is free of the limitations version 5 has for modeling strange configurations in a series crossover. So, my imagination could cut loose a lot more with this one. Actually, I managed to retain a "trademark" feature- this crossover does use a partial cauer-elliptic filter section on the low pass, to help suppress the upper end peak of the midwoofer at 4.2 kHz.
I wasn't really sure whether this was going to work out, as with a series crossover the components interact in their affects on both drivers, but I'm somewhat optimistic now, especially given the robust characteristics of the 27TDFC, and the modeling possible with LspCAD6.
The acoustical target for each driver is third order, but with L-R coefficients, not Butterworth. As designed for the physical offset of acoustical centers, if the tweeter is connected out of phase, a deep null appears at the nominal crossover point of 1850 Hz. This isn't a Butterworth alignment, because both polarities would work for on axis response, with the usual lobe off axis. Also, the main crossosver lobe in the vertical direction is almost straight forward- also a good thing. But doing this requires some compromise in the front axial response, as suggested by Robert Bullock for the conventional 4th order L-R.
The tweeter and woofer data used was taken in a 9" wide cabinet, but smaller and not ported, so the LF response is not as the final design in the PE cabinet. Due to the pressure zone B&K capsule, there's something of a measurement artefact hard to avoid someitmes, so ignore the roll off above 16 kHz.
The intended use for these speakers is for a non-audiophile friend, and more for background music listening or off axis use, so the power response is much more important than the front axial response- with this higher crossover frequency, and using a modified series network, it takes some care to optimize this, and the final results will not be as ideal as with one of my more complex parallel cauer filters. But the component count and cost is about half, a not inconsiderable benefit. So, be aware of the trade-offs.
L4, C8, and R9 together with L2 and C1 and R8 determine the LF roll off characteristic for the midwoofer, including baffle step compensation; L4, C8, and R9 form the elliptic portion of the filter. C1, L3, R3, R4 and R8 determine the tweeter filter high pass - no notch or elliptic filter here.
This plot gives an overview of axis and off axis curves, and the net predicted power response.
How many of you remember when Luxman tried to introduce a speaker, and brought out a two way model around 1976, using a kevlar midwoofer? B&W, which had their distribution in Japan through Luxman, was furious, and pressured Luxman into withdrawing that speaker from the market. It was a very interesting little bookshelf- much better sounding than the comparable B&W models we had in the shop at that time- very nice imaging, too. It didn't measure all that flat on axis, but the overall power response was quite smooth...
Here's a plot of the predicted impedance curve; this looks pretty ideal for the average solid state amp; 5-10 ohms over much of the range, pretty much ideal for getting juice of out midrange gear. Fairly easy phase angle, too.
The highish (by my standards) crossover point results in some narrowing of the polar plot in around 1500 Hz, just before the crossover, but compared with most conventional two ways, the predicted performance looks quite good. It would suffer only in comparison with something like the Modula MTM, I think, with it's lower crossover point.
This won't have the prodigous low bass output capability of an Extremis woofer, but then it's about 4 dB more efficient, and should have nice characteristics down to the high 30's, expecting about -4 dB at 34 - 36 Hz, using a 2" flared Precision Port. :B
Every now and then I have to throw a curve ball just to keep you guys on your toes...
Thanks to Dennis H for catching some mistakes in my post!
ops: Must be taking dufus pills this weekend! :roll: 
. 
Comment