SAM2LF unboxing... to the bone

[pos]

Yes, the metal cones have their challenges. You've set yourself a tougher job than mine though, since I'm crossing at 800Hz. You may need a fairly high order crossover (like LR4). Perhaps if you cross at 1.7kHz (at -6dB), that will work out.

Francis

Yes, JBL cross them at 890Hz in the SAM system...
I will probably end up with a LR8 acoustical crossover, depending on the smoothness of the horizontal off axis response.
Examining the response curves it looks like the breakup are not as severe (lower Q and gain) in the JBL Aluminum as in the SEAS magnesium you are using, so I have hope.
(I think the JBL are using CMMD cones)

The paper cone are much more forgiving when it comes to breakups, especially when additional damping is used (aquaplass), and JBL typically cross them less than one octave below their breakup peaks (1500AL, etc.)...

[pos]

Attachment 63112


seems like the same basket as Revel Ultima 8 inch drivers

Yes they seem to be using the same frames (same goes for their sub).
That particular driver has a 8" Ti cone and a big surround, and is used below 150Hz, and they are using 4" Ti cones to reach the tweeter at 2.3kHz...

[fpitas]

Yes, JBL cross them at 890Hz in the SAM system...
I will probably end up with a LR8 acoustical crossover, depending on the smoothness of the horizontal off axis response.
Examining the response curves it looks like the breakup are not as severe (lower Q and gain) in the JBL Aluminum as in the SEAS magnesium you are using, so I have hope.
(I think the JBL are using CMMD cones)

The paper cone are much more forgiving when it comes to breakups, especially when additional damping is used (aquaplass), and JBL typically cross them less than one octave below their breakup peaks (1500AL, etc.)...

It all sounds workable. If you cross at 1.7kHz (let's say), even with LR4 you'll be pretty far down (42dB by my calculations) at the 5kHz breakup. If you add a notch you can push the breakup peak down even more. 8th order will put it out of consideration.

Francis

[pos]

The problem with that peak is that even if you do not trigger it with input signal (by low passing if low enough) it will still get triggered by distortion products (3rd obviously, but maybe also IMD).
Even if the the 3rd distortion is low in level (-46dB for a 96dB input at 2kHz resulting in a 50dB output at 6kHz) I am afraid it can be audible on some signals because of the sharp Q and associated long time domain oscillations of that peak.
The only way to reduce that effect would be to filter low enough so that the 3rd distortion cannot trigger that peak (ie lower than 1/3 of the frequency of that peak), or use physical filtering in front of the driver (foam or band pass...).

[fpitas]

The problem with that peak is that even if you do not trigger it with input signal (by low passing if low enough) it will still get triggered by distortion products (3rd obviously, but maybe also IMD).
Even if the the 3rd distortion is low in level (-46dB for a 96dB input at 2kHz resulting in a 50dB output at 6kHz) I am afraid it can be audible on some signals because of the sharp Q and associated long time domain oscillations of that peak.
The only way to reduce that effect would be to filter low enough so that the 3rd distortion cannot trigger that peak (ie lower than 1/3 of the frequency of that peak), or use physical filtering in front of the driver (foam or band pass...).

A passive notch filter may be in order. If the peak is like that of the SEAS, it can vary in frequency a percent or two depending on the production lot, so you may have to measure and tweak the notch to be accurate.

Francis

[pos]

Alas a notch on the input signal (active or passive) will not solve the problem of that peak being triggered by distortion products coming from the driver itself.
The only way to attack this is on the output signal of the driver (dense foam or physical 4th order bandpass) or by modifying the driver itself (aquaplas plastering? )

Regarding frequency response, I was very impressed on how well these matched the measurement from EDS: the peaks are exactly at the same positions and the two drivers are very well matched.
Kudos to Jerry Moro for the design and GGEC for the production!

[fpitas]

Alas a notch on the input signal (active or passive) will not solve the problem of that peak being triggered by distortion products coming from the driver itself.
The only way to attack this is on the output signal of the driver (dense foam or physical 4th order bandpass) or by modifying the driver itself (aquaplas plastering? )

Regarding frequency response, I was very impressed on how well these matched the measurement from EDS: the peaks are exactly at the same positions and the two drivers are very well matched.
Kudos to Jerry Moro for the design and GGEC for the production!

I see; you were positing that the distortion from the driver itself might excite the peak. That may be the case, but I wonder whether that isn't revealed on the distortion graphs for the driver at 1/3 the peak frequency, 1/5 etc. I'll have to ponder that one. Were it me in your position, I'd give it a try with your best effort at filtering etc. and see how it works out. I doubt you are eager to start physically modifying the drivers with aquaplas etc.

Francis

[pos]

I wonder whether that isn't revealed on the distortion graphs for the driver at 1/3 the peak frequency, 1/5 etc

It is
Here are a few examples, observe 3rd distortion behavior at 1/3 of the fundamental:

JBL Synthesis 8 :


SEAS w22ex001 :


JBL 1501AL-2 :

[fpitas]

So, to me it seems that reducing drive at or below 1.7kHz will do the trick. Since it's third-order distortion, it should follow a cube law for amplitude, meaning (for example) a 6dB reduction in drive at the trouble spot will result in an 18dB reduction in distortion products. At least the devices I work with follow that behavior.

Francis

[pos]

Filtering with a steep low-pass at 1.7kHz should indeed reduce the problem (the peak has a low Q and is centered around 6kHz) as far as 3rd distortion is concerned (I don't know what to expect with IMD...).

I don't think this kind of distortion would follow a cube low like excursion related distortion would.

For example with the 1501AL-2:
@ 100dB: http://www.audioheritage.org/vbullet...1&d=1350659878
@ 115dB: http://www.audioheritage.org/vbullet...1&d=1350659917

Excursion related distortion (down low) does raise quite a bit relative to fundamental, as well as 2nd order distortion in the breakup region, but the 3rd order distortion double peak around 500Hz (related to the 1500Hz double breakup peak in the response curve) stays quite constant at -45dB.

[fpitas]

Filtering with a steep low-pass at 1.7kHz should indeed reduce the problem (the peak has a low Q and is centered around 6kHz) as far as 3rd distortion is concerned (I don't know what to expect with IMD...).

I don't think this kind of distortion would follow a cube low like excursion related distortion would.

For example with the 1501AL-2:
@ 100dB: http://www.audioheritage.org/vbullet...1&d=1350659878
@ 115dB: http://www.audioheritage.org/vbullet...1&d=1350659917

Excursion related distortion (down low) does raise quite a bit relative to fundamental, as well as 2nd order distortion in the breakup region, but the 3rd order distortion double peak around 500Hz (related to the 1500Hz double breakup peak in the response curve) stays quite constant at -45dB.

That may be the case with this form of distortion. I'll admit I haven't measured the effect on my SEAS at their 1/3 of peak frequency; I just made sure the drive was quite small by then. Short of damping the cone though, I'm not sure what you might do. And, it may not be an audible problem. Once again my advice, since you have the drivers, is to give it a try, and measure and listen. At worst, you might need to use a larger waveguide to push the crossover frequency lower.

[pos]

That is the plan
I hope I will be able to keep the small 6" 90x90 PT waveguide and 2407 driver as this combo is really impressive in the upper decade...
The ideal match would probably be an 8" PT waveguide, but these only exist in 120x60 and 90x50 versions, and are much deeper and convoluted than this one, and I like the symmetrical directivity...

One possibility would be to scan and model a PT-H1010 waveguide and scale it down from 1.5" to 1", resulting in a 8" waveguide...