Luke's Gjallarhorn/Othorn Discussion

[Nate Hansen]

I know the feeling!  Every time I visit another HT I end up making changes to mine.  

 

The GTG had a great turn out this year, and we had plenty of material to play with.  I need to get you and Mike over to hear my latest EQ settings on the mains.  

 

Bring it on! I'm always down for a bass onslaught....

[lukeamdman]

My AE L/C/R build is the first to be featured in the "DIY Showcase" on John's new website!

 

http://aespeakers.com/dual-ae-td18h-mains-and-td15m-center-channel-by-lukeamdman/

[SME]

Congrats!  Do you think you will ever upgrade beyond what you have?

 

I'm actually going with an AE + SEOS combo, but I am opting to go with the TD12M and 15" SEOS in a 2-way active configuration and crossed to subs at 80 Hz.  That's still good for 120 dB @1m, which is plenty.  I don't need full range, especially because the subs achieve better in-room response anyway.

[lukeamdman]

Congrats!  Do you think you will ever upgrade beyond what you have?

 

I'm actually going with an AE + SEOS combo, but I am opting to go with the TD12M and 15" SEOS in a 2-way active configuration and crossed to subs at 80 Hz.  That's still good for 120 dB @1m, which is plenty.  I don't need full range, especially because the subs achieve better in-room response anyway.

 

I certainly don't have any plans for any upgrade, but I'm sure years down the road something will happen.

 

Which CD are you planning to use?

[SME]

Which CD are you planning to use?

 

I'm trying to get DIYSG FL-450s, but I've been having trouble closing the deal with Erich.  If those don't work out, I'm leaning toward the DIYSG DE-250 or the Radian 475PB.  I'd consider the beryllium option for the 475PB, but I'd really like to see some data (other than Radian's own generously smoothed plots, which even Radian doesn't seem to provide).  Theoretically, beryllium is superior to almost all other materials, but this means nothing if other aspects of the design are significantly flawed.

 

Unfortunately, there is very little good data for evaluating CDs.  This is complicated by the fact that the choice of horn has a big impact on response.  One reason I lean toward the DIYSG offerings is that they've actually been tested in the SEOS horns instead of in horns with much higher loading as is the case with most manufacturers.

[lukeamdman]

I'm trying to get DIYSG FL-450s, but I've been having trouble closing the deal with Erich.  If those don't work out, I'm leaning toward the DIYSG DE-250 or the Radian 475PB.  I'd consider the beryllium option for the 475PB, but I'd really like to see some data (other than Radian's own generously smoothed plots, which even Radian doesn't seem to provide).  Theoretically, beryllium is superior to almost all other materials, but this means nothing if other aspects of the design are significantly flawed.

 

Unfortunately, there is very little good data for evaluating CDs.  This is complicated by the fact that the choice of horn has a big impact on response.  One reason I lean toward the DIYSG offerings is that they've actually been tested in the SEOS horns instead of in horns with much higher loading as is the case with most manufacturers.

 

Erich can be that way...don't hold your breath. 

 

I don't want to make a judgement without hearing them first, but I doubt Be diaphragms are a game changer.  I definitely wouldn't turn down the opportunity for a blind test though.  

[3ll3d00d]

I'd consider the beryllium option for the 475PB, but I'd really like to see some data (other than Radian's own generously smoothed plots, which even Radian doesn't seem to provide).  Theoretically, beryllium is superior to almost all other materials, but this means nothing if other aspects of the design are significantly flawed.

have you given them a call? I had a chat with Gene @ Radian recently and he provided some more accurate/recent data on their 5208 coax. The published data, at least for that model, is quite a way away from the actual spec. From that discussion I gathered that they primarily deal direct with speaker designers but are totally open to the DIYer.

[SME]

Erich can be that way...don't hold your breath. 

 

I don't want to make a judgement without hearing them first, but I doubt Be diaphragms are a game changer.  I definitely wouldn't turn down the opportunity for a blind test though.  

 

I believe Be could make a significant difference, but only if the rest of the design is executed well.  I'd rather have good data than "the opportunity for a blind test" though, because things really are a lot more complicated than that.  I don't know that I could appreciate the differences with a single blind test.  I'd need time to integrate it into the rest of my system and listen for a month or so to a variety of music and then do the blind test with my own.  The thing is, I find the blind tests are less important when you have good measurements to work with.

 

have you given them a call? I had a chat with Gene @ Radian recently and he provided some more accurate/recent data on their 5208 coax. The published data, at least for that model, is quite a way away from the actual spec. From that discussion I gathered that they primarily deal direct with speaker designers but are totally open to the DIYer.

 

I hadn't thought to do that.  What kind of measurements were you able to get data for?

 

I will emphasize again the difference the horn makes in the results.  From what I've seen, CDs don't extend as low or play as loud in SEOS horns as they do in the horns that vendors typically use for testing.  At the same time, as the wavelengths get small, the on-axis response and dispersion become much more sensitive to the interaction between the CD components (diaphragm, phase plug, and exit) and the horn.  The dispersion patterns are often very unusual and may exhibit rapid variations in directivity that are only apparent with polar response measurements but are most likely audible unless this octave is rolled-off a lot, as it often is.

 

The other thing I'm interested in seeing is characterization of break-up resonance.  Frequency response measurements alone don't seem sufficient to reveal or compare break-up resonance.  The horn itself contributes quite a bit of variability in the frequency response, so you won't be able to see resonance peaks there like you might for a woofer.   Impedance measurements, which might be more helpful, seem to be rarely posted for CDs.  A more useful set of tools are waterfalls, spectrograms, and other kinds of time-frequency analysis.  The trouble with these is that the visual emphasis of break-up depends a lot on the settings used.  What might also be helpful are sweeps at successively higher levels like Ricci publishes for subs to assess compression.  I suspect break-up resonances may tend to behave non-linearly, as I have observed as much myself.  This means that the resonances will become more accentuated with increasing output levels, and such a trend may be revealed by sweeps at different levels.

 

I had the crazy thought that we might need a "data-tweeter" site for measuring CDs and other HF drivers.

[lukeamdman]

I believe Be could make a significant difference, but only if the rest of the design is executed well.  I'd rather have good data than "the opportunity for a blind test" though, because things really are a lot more complicated than that.  I don't know that I could appreciate the differences with a single blind test.  I'd need time to integrate it into the rest of my system and listen for a month or so to a variety of music and then do the blind test with my own.  The thing is, I find the blind tests are less important when you have good measurements to work with.

 

 

I hadn't thought to do that.  What kind of measurements were you able to get data for?

 

I will emphasize again the difference the horn makes in the results.  From what I've seen, CDs don't extend as low or play as loud in SEOS horns as they do in the horns that vendors typically use for testing.  At the same time, as the wavelengths get small, the on-axis response and dispersion become much more sensitive to the interaction between the CD components (diaphragm, phase plug, and exit) and the horn.  The dispersion patterns are often very unusual and may exhibit rapid variations in directivity that are only apparent with polar response measurements but are most likely audible unless this octave is rolled-off a lot, as it often is.

 

The other thing I'm interested in seeing is characterization of break-up resonance.  Frequency response measurements alone don't seem sufficient to reveal or compare break-up resonance.  The horn itself contributes quite a bit of variability in the frequency response, so you won't be able to see resonance peaks there like you might for a woofer.   Impedance measurements, which might be more helpful, seem to be rarely posted for CDs.  A more useful set of tools are waterfalls, spectrograms, and other kinds of time-frequency analysis.  The trouble with these is that the visual emphasis of break-up depends a lot on the settings used.  What might also be helpful are sweeps at successively higher levels like Ricci publishes for subs to assess compression.  I suspect break-up resonances may tend to behave non-linearly, as I have observed as much myself.  This means that the resonances will become more accentuated with increasing output levels, and such a trend may be revealed by sweeps at different levels.

 

I had the crazy thought that we might need a "data-tweeter" site for measuring CDs and other HF drivers.

 

With a blind testing I'm referring to a setup where the only difference is a Be diaphragm vs polyester/titanium/whatever.  Everything else, including the basic response at the LP, would be identical.  

 

There will be plenty of things that a microphone will pick up and show in great detail that we'll never be able to perceive with our ears.  Obviously I spend a lot of time taking measurements and tweaking things, but ultimately I trust my ears and hey, I like what I like.  I disregard changes all the time that look better in REW but don't float my boat with the sound and my subjective preference.   

[tuxedocivic]

SME, have you seen my testing of Be? It was a 4" phragm though. On an Iwata 300. The difference in listening was fairly apparent to me. I didn't post CSD because I find this data troublesome to reveal. It's easily misunderstood.

[3ll3d00d]

I hadn't thought to do that.  What kind of measurements were you able to get data for?

 

in this case I was just looking for accurate TS specs & accurate info on xmax (as published values varied from ~0.1mm to 3mm which is just a bit of variation!), they supplied this quite happily and I had a decent chat with him about various issues around speaker design & what sort of thing they're doing. I don't know whether they'll have the data you want but they were v helpful when I spoke to them & there is v little data in the public domain so it can't hurt to ask.

 

 

I will emphasize again the difference the horn makes in the results.  From what I've seen, CDs don't extend as low or play as loud in SEOS horns as they do in the horns that vendors typically use for testing.  At the same time, as the wavelengths get small, the on-axis response and dispersion become much more sensitive to the interaction between the CD components (diaphragm, phase plug, and exit) and the horn.  The dispersion patterns are often very unusual and may exhibit rapid variations in directivity that are only apparent with polar response measurements but are most likely audible unless this octave is rolled-off a lot, as it often is.

 

The other thing I'm interested in seeing is characterization of break-up resonance.  Frequency response measurements alone don't seem sufficient to reveal or compare break-up resonance.  The horn itself contributes quite a bit of variability in the frequency response, so you won't be able to see resonance peaks there like you might for a woofer.   Impedance measurements, which might be more helpful, seem to be rarely posted for CDs.  A more useful set of tools are waterfalls, spectrograms, and other kinds of time-frequency analysis.  The trouble with these is that the visual emphasis of break-up depends a lot on the settings used.  What might also be helpful are sweeps at successively higher levels like Ricci publishes for subs to assess compression.  I suspect break-up resonances may tend to behave non-linearly, as I have observed as much myself.  This means that the resonances will become more accentuated with increasing output levels, and such a trend may be revealed by sweeps at different levels.

I have quite a bit of data inc polar response (horizontal and vertical) on my SEOS10/BMS4550/AE TD10H combo. Smaller horn than you're using but can share the raw data if you want it.

[BeastAudio]

My AE L/C/R build is the first to be featured in the "DIY Showcase" on John's new website!

 

http://aespeakers.com/dual-ae-td18h-mains-and-td15m-center-channel-by-lukeamdman/

 

 

Good work John....Who is going to want to show THAT build up? He kinda just doomed his "Showcase" section....

[SME]

SME, have you seen my testing of Be? It was a 4" phragm though. On an Iwata 300. The difference in listening was fairly apparent to me. I didn't post CSD because I find this data troublesome to reveal. It's easily misunderstood.

 

I haven't.  Can you post a link?  Was it a Radian CD with Be diaphragm?  If so, do you have measurements of the same CD with a non-Be diaphragm?

[SME]

in this case I was just looking for accurate TS specs & accurate info on xmax (as published values varied from ~0.1mm to 3mm which is just a bit of variation!), they supplied this quite happily and I had a decent chat with him about various issues around speaker design & what sort of thing they're doing. I don't know whether they'll have the data you want but they were v helpful when I spoke to them & there is v little data in the public domain so it can't hurt to ask.

 

Interesting.  Maybe I should give it a try.  Or maybe just send an email.

 

I have quite a bit of data inc polar response (horizontal and vertical) on my SEOS10/BMS4550/AE TD10H combo. Smaller horn than you're using but can share the raw data if you want it.

 

That'd be great!  I'm open to considering that driver as well.  What crossover do you use?

[tuxedocivic]

Oops, meant to: http://www.avsforum.com/forum/155-diy-speakers-subs/1800353-king-bouma-s-info-eventual-build-thread-2.html#/forumsite/3207/topics/1800353?page=1

[SME]

Oops, meant to: http://www.avsforum.com/forum/155-diy-speakers-subs/1800353-king-bouma-s-info-eventual-build-thread-2.html#/forumsite/3207/topics/1800353?page=1

Thanks.  Wow, that really is a night and day difference!

 

I wish I could find something similar for Be vs. non-Be on a 1" Radian driver, but there's not much.  This thread on diyaudio seems to have mostly negative reviews of the Radian + Be combo with one user suggesting that the Be version may use the same suspension, which is not optimized for Be use.  I have no idea, but considering the cost, I'd like to have some good data to back it up.  I'm also concerned in general about the extension capability of the Radian's.

 

Right now, the DNA-360 is looking like my second choice.  It looks like it can cross lower than the BMS 4550, or most other 1" drivers for that matter.  I haven't decided where to cross yet, and it will depend on what measurements reveal.  Having the option to cross lower is a real plus.

[tuxedocivic]

The B&C DE550 might be worth a look for a beefy 1".

[SME]

The B&C DE550 might be worth a look for a beefy 1".

Interesting.  I can't find any data at all on using this with an SEOS horn.  I actually like the look of the DE500 more because the shorting ring improves the top octave response.  For some reason though, it has a recommended crossover of 1.6kHz vs. 1.2 kHz for the DE550, but I can't see why by looking at the published response curves.

 

All this makes me think I should maybe just order one driver of each kind I'm considering (including the Radian Be and FL-450, if I can get it) to test them all.  Hopefully I could sell anything I don't use.

 

Edit: Just read that the DE500 uses Ti diaphragms.  Yuck.  I wonder why they don't put a shorting ring on the 550?

[3ll3d00d]

 

That'd be great!  I'm open to considering that driver as well.  What crossover do you use?

 

there's some data in https://drive.google.com/folderview?id=0BxdmSMpV-t3GQ28yc1FJZ2ozYjA&usp=sharing

 

the folder named directivity has 0-60 degree data for each driver with no crossover mounted in my enclosure, the folders beneath that are processed into sonograms using a script I hacked up (https://github.com/3ll3d00d/directivity-utils)

 

the xo_take1 folder has more detailed directivity (-90 to +90 horizontal and -45 to +45 vertical) but that's with my 1st pass at a XO in there 

 

I have rolled my own passive XO btw, the results of the (possibly) final iteration is in the xo_take3 folder

 

in case you haven't seen it, there's a load of data in http://www.diyaudio.com/forums/multi-way/212240-high-frequency-compression-driver-evaluation.htmlas well

[SME]

there's some data in https://drive.google.com/folderview?id=0BxdmSMpV-t3GQ28yc1FJZ2ozYjA&usp=sharing

 

Thanks for the data!  That CD does look very good.  It appears to be a very good geometric match with the throat of the SEOS-10 horn.   The normalized sonogram looks very good.  Where I would be concerned is that the break-up resonances that appear around 15.5 and 19 kHz may not clean up as nicely as with EQ as the normalized sonogram suggests.  It depends on how linearly they behave at the particular listening level chosen.  In your "outside" measurements, the break-up appears to be controlled fairly well at the output level you measured and is probably quite linear.  As SPL increases, things could get ugly, but I can't find any data to tell me at what point things get especially non-linear.

 

There's also some uncertainty about the low end performance.  I did see the studies in the link you provided, but I didn't study it in-depth or download any of the samples.  The author states it has very high distortion and shouldn't be kept under 1 W at 500 Hz.  That suggests that a 12 dB/octave crossover at 1 kHz would make its effective "power handling" something like 20 W.  At the same time though, the CD is producing like 12 dB less output even without the crossover, so that fundamental at 500 Hz that demands 1W from the tweeter (which would play at like 111 dB in-room) would produce a 2nd harmonic at about 3%.  So maybe that 12 dB/octave 1 kHz crossover is fine and dandy!

 

 

About non-linear diaphragm break-up resonance, check out the picture above I have from my front left speaker.  These sweeps are all close miced right in front of the horn at what would be around 75 dB (IIRC) at the MLP. with the only change being the duration of the sweep / FFT length.  From blue to red, they are 128k, 256k, 512k, and 1024k.  Absolutely no smoothing or windowing is used.  In theory, these should all appear to be virtually identical because even though the longer sweeps have higher resolution, the display resolution is still way lower than the measured resolution for even the 128k sweep.  Resonances are apparent at about 7.2 kHz, at around 15 kHz, and then again at 19 kHz and above.  Note that the one at 7.2 kHz is more severe and exhibits substantial variation between sweeps whereas the other resonances don't really vary at all.  Even though there is a deep notch at 15 kHz, the response is still very linear there,, at least at the levels I did my sweeps.  I would not be surprised to see the different sweeps exhibit divergence at 15 kHz and maybe 19 kHz if the sweep levels were increased enough.  I don't currently have the measurements to back it up, but subjectively, the 7.2 kHz resonance sounds much worse as the playback level is pushed up.

 

All my speakers have Ti diaphragms, which are known for break-up problems.  In this particular instance, the resonance is severe and very audible, due to either a manufacturing flaw or damage to the driver in-use.  I'm guessing other high frequency drivers exhibit this phenomenon in some form or another, more typically at higher drive levels.  The question then is not just at what frequencies the break-up occurs but at what point does it begin to behave non-linearly.  This point is crucial when using DSP correction.  At the output level I measured, DSP EQ could eliminate most of that dip at 15 kHz because it's essentially the same across multiple seats and it responds quite linearly.  That business at 7.2 kHz is a different story, and in fact trying to correct it with DSP is likely to only make it worse!

[lukeamdman]

Here's a track I chose to test the voltage/amperage on during music playback:

 

 

 

The very first note at the bottom is 27.5hz, so I created a 6 second particle and looped it: