Something stirs in the Undergrowth

[IslandPink]

Interesting indeed. I had some brief correspondance with CV about this . The Radian is clearly superior from somewhere around 2-3kHz upwards . It's the area at the low end where possible the lack of 'reach' isn't working for me. Maybe though I just need to play around with the crossover some more ?
Now, the other option CV favours is running the Fostex to 1.2k low-pass and raising the high-pass for the Radian ( or GPA288 ) to 1.2k also . Maybe this will get around the problems of 'disjoint' in trying to cross at 700Hz that I've found with both drivers at various times. The only trouble with this , is that it's harder to control the 'crud' that the Fostex produces above 3.5kHz . When I ran the Fostex with an Aurum G3 ribbon a few months ago , it was necessary to have a notch group at 4kHz , in addition to the low-pass on the Fostex. That was with about 2.2kHz low-pass. At 700Hz low-pass , it doesn't seem to be a problem. I wil report back soon ....

Pic of latest set-up shown below :

[Lynn Olson]

I've gotten a lot of heat in the "other" forum about designing the AH425 around a large-format driver instead of a small-format (1" exit) compression driver. After all, the small-format drivers can easily reach 15 kHz or better, even with plain old Mylar diaphragms.

The real reason is subjective, which isn't that welcome in the other forum. There's something very special about the sound of a large-format compression driver; an easy, relaxed, and unforced midrange. By comparison, small-format compression drivers start to sound hard and spitty if you ask them to work below 2 kHz ... and to me, don't sound that great above 2 kHz, either.

This is why I settled on a crossover in the 700~850 Hz range. If the horn were larger, say, the size of an Altec 1005 or the truly gigantic 1505 multicell, then 500 Hz would sound good.

If the large-format compression driver, AH425, and crossover are working correctly, 700~850 Hz should sound unstressed and relaxed on all programme material, at all playback levels. The first hint of stress is a reminder to raise the crossover, increase the slope, or ask if the compression driver is comfortable in the 1~2 kHz range.

Not many drivers handle 700 Hz to 2 kHz successfully; this is probably the most critical region in the entire spectrum, and small faults are very audible. Although the compression driver is on a tightrope, and has to handled correctly, I'd guess it would outperform any cone driver, including Feastrex and Lowthers, in this range.

Another perspective is from the viewpoint of the direct-radiator cone driver. Below 700 Hz is asking little from a competently engineered driver, since it's in the piston band, and variations in VC inductance aren't as severe an issue as it becomes at higher frequencies.

For a cone driver, 2 kHz is approaching the top of the range, in terms of freedom from cone breakup. The little Vifa's of the Ariel sidestepped the problem with soft polypropylene cones, which traded resolution and insight for good behaviour in the HF rolloff region (thus no need for notch filters).

Drivers with more rigid cones are another matter. My personal feeling is that if a driver doesn't sound musically "right" without any crossover at all, it should be avoided (with the exception of a HP filter to protect tweeters, of course).

Notch filters are tempting, but it's a temptation I try to avoid, since there's still just a shadow of unnatural sound even with a really well-done notch filter. This might be because of correction for on-axis is not correct for off-axis, as well as IM distortion residues at the mechanical resonance of the diaphragm.

I have seriously contemplated a separately amplified electrostatic panel to cover the range above 5~7 kHz. No need to roll off the AH425 since it's so directional at those frequencies anyway. There's precious little acoustic energy above 5~7 kHz, so the panel's not going to overloaded by even the most aggressive programme material.

[cv]

MJ,
Just occurred to me, why not try:

288H on a 200Hz horn from 500-2kHz
Radian on an 800Hz horn 2kHz+

You could retain the Fostex but cross it lower, or try running the Supravox up to 500Hz (though the floor location is probably not ideal) or replace the Fostex with a high efficiency driver to cover the low mids...

On a side note, I can now produce STL CAD files for elliptical JMLC horns with any desired parameters - some mates are putting one together for the S2... will let you know how that turns out. Not a pun as it's a bit hard to turn an elliptical... but for multiple horn systems, ellipticals should stack nicely... and I suspect sound the least coloured...

[waveguide]

I've gotten a lot of heat in the "other" forum about designing the AH425 around a large-format driver instead of a small-format (1" exit) compression driver. After all, the small-format drivers can easily reach 15 kHz or better, even with plain old Mylar diaphragms.

The real reason is subjective, which isn't that welcome in the other forum. There's something very special about the sound of a large-format compression driver; an easy, relaxed, and unforced midrange. By comparison, small-format compression drivers start to sound hard and spitty if you ask them to work below 2 kHz ... and to me, don't sound that great above 2 kHz, either.

This is why I settled on a crossover in the 700~850 Hz range. If the horn were larger, say, the size of an Altec 1005 or the truly gigantic 1505 multicell, then 500 Hz would sound good.

If the large-format compression driver, AH425, and crossover are working correctly, 700~850 Hz should sound unstressed and relaxed on all programme material, at all playback levels. The first hint of stress is a reminder to raise the crossover, increase the slope, or ask if the compression driver is comfortable in the 1~2 kHz range.

Not many drivers handle 700 Hz to 2 kHz successfully; this is probably the most critical region in the entire spectrum, and small faults are very audible. Although the compression driver is on a tightrope, and has to handled correctly, I'd guess it would outperform any cone driver, including Feastrex and Lowthers, in this range.

Another perspective is from the viewpoint of the direct-radiator cone driver. Below 700 Hz is asking little from a competently engineered driver, since it's in the piston band, and variations in VC inductance aren't as severe an issue as it becomes at higher frequencies.

For a cone driver, 2 kHz is approaching the top of the range, in terms of freedom from cone breakup. The little Vifa's of the Ariel sidestepped the problem with soft polypropylene cones, which traded resolution and insight for good behaviour in the HF rolloff region (thus no need for notch filters).

Drivers with more rigid cones are another matter. My personal feeling is that if a driver doesn't sound musically "right" without any crossover at all, it should be avoided (with the exception of a HP filter to protect tweeters, of course).

Notch filters are tempting, but it's a temptation I try to avoid, since there's still just a shadow of unnatural sound even with a really well-done notch filter. This might be because of correction for on-axis is not correct for off-axis, as well as IM distortion residues at the mechanical resonance of the diaphragm.

I have seriously contemplated a separately amplified electrostatic panel to cover the range above 5~7 kHz. No need to roll off the AH425 since it's so directional at those frequencies anyway. There's precious little acoustic energy above 5~7 kHz, so the panel's not going to overloaded by even the most aggressive programme material.

Useful posting Lynn.

Are there any other compression drivers that will give the extension to 600Hz or below with good treble that may work well without but still be improved by a tweeter. i.e give a 2 or 3 way. We need to identify some more readily available compression drivers affordable both sides of the duck pond. There is nothing in posts that I can find that really tests a sensible selection of current CD's.

[IslandPink]

Some interesting ideas there CV . Yes, I started thinking about the bi-radial TAD horn before reading your comments about elliptical, but that would be the way to go, I guess.
The Supravox has to be near the ground to get the bass extension, so can't go to 500Hz, probably 300 Max . The Radian will also need help above 8kHz , so it's going to be a 5-way ; probably too complex.

I'll mull over this during my holiday in the Lake District.
Shame there isn't an option that's affordable, that we missed, for compression driver down to 200-300Hz....

I'll be offline for about a week now, cheerio !

[chris661]

An EV DH1a will probably do the low-end you want at domestic levels. They're meant for PA use above 500Hz, and can take on some of the best modern compression drivers and win out:

http://www.diyaudio.com/forums/multi-wa ... ation.html

They'll also get up to 15kHz if you don't mind EQing a bit.

I have a pair of these that I really wouldn't fancy selling, but might bring them to Owston if you'd be interested in seeing what they can do.

Chris

[slowmotion]

It's an interesting discussion , and as always, your tastes in music and in general will strongly influence your choices.
I've come to appreciate the 288 for midrange with a 1" comp driver on top.
I agree with what Lynn writes regarding the large format driver for midrange.

There's something very special about the sound of a large-format compression driver; an easy, relaxed, and unforced midrange.

[Lynn Olson]

I think Slowmotion is on to something. A 288 with a good 1" compression driver on top is a very traditional solution. I've even heard suggestions of using a 604 Duplex this way: large-format horn from 700 Hz to 7 kHz, than the internal horn of the 604 above that. However, with a new Alnico Duplex around US$1500 each, while a new Alnico 416 is US$360 each, that's a mighty expensive supertweeter.

It does have subtle advantages, though: the internal horn acts as a "localizer", with transients located right in the center of the 15" driver, which is a good thing. And the 700 Hz crossover from the 15" driver to the large-format horn avoids the really bad region of the 604 from 1 to 2 kHz, where both drivers are really coming apart thanks to cone reflections and the small internal horn running out of range.

I think it'll be easier to find good-quality 1" drivers than a superlative 1.4" or 2" large-format drivers, since these operate in such a critical region of hearing. Since the smaller compression driver mostly carries leading-edge transients, it should be time-aligned, with the back of the 1" compression driver maybe 1~2 cm behind the back of the 288 driver (both subjective listening and impulse measurements required).

Aside from the Radian vs 288 discussion, another advantage of a supertweeter is wider dispersion in the frequency range where you need it, and maybe crossing as low as 5 kHz. That's really easy for both drivers: the 288 is well below first breakup mode, the AH425 dispersion is still reasonable, and the 1" driver is hardly working at all. It also dramatically reduces IM distortion in the HF to VHF range, not a minor consideration.

If you follow this path, an additional backwards and upwards-facing 1" driver will open up the HF range considerably, and you have the option spectrally shaping the FR of the rear-facing driver to offset the narrowing dispersion of the front-facing drivers.

[IslandPink]

The only problem with CV's suggestion ( from a quick calc while on holiday ) is the size of a 200Hz le Cleac'h , elliptical ... I think Chris still believes I live in a converted Welsh chapel !

[IslandPink]

OK, finally some progress after a few busy weekends .
I've been meaning to try out Chris's suggestion ( & JMLC's advice I believe ) to go higher with the horn crossover to keep output away from the 'unloading' region where the phase is rotating .
So, I dug around for caps & resistors and managed to get a 1200Hz low-pass on the 'Quasar' section and a 1300Hz high-pass on the Horn ( 288 first ) . This sounded pretty good. I was expecting more 'crud' from the Fostex in the 2K+ area given it's only a 1st-order roll-off there, but it sounded pretty good . Good integration .
After a while I was hearing some of the old roughness & distortion of the 288 on the horn, so I swapped in the Radian 745Be . This sounds very very nice I think this is the best combo I've yet had . The slight lack of output & 'depth' of the Radian below 1000Hz becomes a non-issue, and its extra resolution & tone comes through very well . I think the vocals are better by allowing the Fostex to run to 1200Hz without the awkward break to the horn . So this will be the way forward, probably with a bit of EnABL on the 208Ez once I've trialled it on the 103E's .

For info, the filters are achieved ( in my 4P1L/300B amp ) by the following :
(i) For quasar section, add 5.6k resistor in series from the 4P1L ( 7.5k Zout itself ) then 0.01uF ( copper PIO ) to ground ; this ahead of the normal 0.47u/100k coupling .
(ii) For the horn, reduce coupling cap to 1.5nF ( Philips polyprop/foil 250V 'blue' ) with 82k to ground grid-leak on 300B .

[Lynn Olson]

Thanks for your posting, IslandPink! Just for review, what's the order of the highpass filter for the AH425?

I'm contemplating an OB-mounted Altec/GPA 416 (Alnico), operating mostly through the baffle-peak region, with an additional 416 in a closed-box beneath it, so I'm watching your experiments with great interest.

The Edge simulations indicate that an asymmetric baffle can provide a nice double-peak with a central dip of about 0.5 to 1 dB, which in the overall scheme of things isn't much. The net acoustic gain of this OB layout is about 1 dB in the working passband, which is about 200 Hz to 800 Hz.

[IslandPink]

Lynn, it's a first-order high pass for the AH425 . After further listening I do have misgivings about the FE208Ez with that 1300Hz low-pass , on vocals, so that will need some more thinking probably to add a notch in the amp to knock down the 3.5/4k peak .

[Lynn Olson]

You could adopt a hybrid filter for the Fostex. By hybrid I mean a filter that's partly in the amplifier, and partly at speaker level.

For example, there could be a 1st-order lowpass in the amplifier, and a synchronous-tuned inductor going to the Fostex driver, with the inductor followed by a passive notch filter. (A parallel notch filter is going to require some inductance ahead of it to work correctly.)

Synchronous-tuned filters have a very low Q of 0.5, and do not overshoot, unlike a conventional Butterworth 2nd-order filter, which has moderate overshoot.

The notch filter will alter the tuning of the speaker-level inductor, but won't have any effect on the LP filter in the amplifier. I think you could knock together a filter setup like this in an hour or two, and make your own evaluation on how well it works.

P.S. I should also mention a hybrid filter is also a simple method to get a 2nd-order, Q=0.5 highpass filter on the AH425 if desired. Since the two poles are independent, it's easier to tune than a passive speaker-level crossover.

I've had good luck with this kind of filter topology, since they are free of peaking and overshoots, and have more flexibility than in-amplifier filtering.

[waveguide]

Lynn, it's a first-order high pass for the AH425 . After further listening I do have misgivings about the FE208Ez with that 1300Hz low-pass , on vocals, so that will need some more thinking probably to add a notch in the amp to knock down the 3.5/4k peak .

The problem as always is moving into a realm of complexity. You have the basis for what in my view is a good two way. If it sounds good do not bother with trying to get good? technical measurements. Geddes seems to have some good real world designs, but I can see it is good fun trying to pass up to even potentially better performance with Lynns project.

I do not like the idea of flea sized amps valve or ss as they are probably better suited to headphone use where another level of finess is possible straight into the ears.

[IslandPink]

Sorry I didn't comment on the last two emails . Been somewhat busy but also wanted to get some more listening done and some basic measurements to see where I stand currently . I've been tinkering with the little MLTL speakers with the Fostex 103E drivers, and they have also been showing up the 'big' system in some areas, so I got a bit demoralised .

Here however are some measurements , not 'industry-standard' but hopefully have some useful content. The first two are the whole ( 3-way ) system , from two slightly different mike positions , each about 1.9m from the speaker . The distance is rather arbitrary due to the relative immobility of the computer and my desire ( lazy ? ) not to move the speaker from where I've been listening to it. I'm most interested in the 500kHz to 5kHz region, so didn't try to do much to cure floor bounce. These are unsmoothed as you can probably tell .

The left amp channel drives the bass & mid with a 1.2kHz low-pass, and the right amp channel with a 1.3kHz high-pass drives the Radian/AH425 ; I tried to get the levels approx correct , but the horn is still a little strong in this combo.

There's a fair amount of choppiness through the vocal region , some of which changes with mike position so there's room-mode content there. The Radian holds up well to 10kHz then drops fairly gracefully .

More to follow .