[scorpion]

Martin,

I am looking forward to see what you arrive at.

David,

The Kef 139 was and is a very good element even for upper bass. I do not think I have seen any TSP-parameters for it, have you any clue ?
But really 4 Kef B139s would not equal much more than one 15" unit ?

/Erling

[Russell Dawkins]

Erling,

I don't know how far you consider upper bass to extend, but according to Lynn Olson, the B139 has a stubborn peak at 1.5 kHz that he found audible except with fairly brute crossover design, like 3rd order filters.

You might find this page interesting:

http://www.nutshellhifi.com/ME2txt.html

(see almost halfway down the page, just under the first illustration)

[D OB G]

Erling,

Kef themselves apparently showed a peak at 1000 Hz.
I and others have measured, what I presume to be the same peak to be at 800 Hz.
Lynn Olsen might be measuring the same peak at 1.5 kHz, as he specifically mentions one, but in any case, there are heaps of peaks above 800/1000/1500 Hz.

That's why I cross actively at 24 db/ octave at 125 Hz.
It's purely a bass driver- a very good linear piston (good for transmission lines too).

[D OB G]

Martin,

I have half! a photo-copied data sheet for the Kef B139.

Fs 25Hz Qm5.5 Qe0.4 (no Qt on the page- you'll have to calculate yourself).

Vas 164 litres, and just for good measure Bl 12.3

David

[D OB G]

Martin,

Qt 0.37 ??

David

[D OB G]

MJK,

"I was referring to the electrical impedance peak around driver resonance, this will be quite a tall peak. How do you size the inductor to account for this very non constant electrical impedance right in the middle of the intended range of operation?"

Maybe the theoretical answer might have something to do with the phase change that occurs at resonance.
There is a zero phase angle?

???????

Still looking for some further theoretical input to explain the empirical results.

David

[D OB G]

Martin,

In my book a 15" driver has an effective cone diameter less than 15", and the B139s have a moving diameter a bit less than 10 inches, so I reckon it works out pretty closely.

David

[scorpion]

Russel,

Thanks for the link. So long time since B139, B110 and T27 were state of art. What I meant was a bit over sub-frequencies. I think the B139 used
to be crossed in the 300-400 range.

D OB G,

Given Qm=5.5 and Qe=0.4 then Qt should be 0.37.
The Kef B139 has a cone area of about 56 sq inches or 350 sqcms (edited with values from Kef spec-sheet). A 15" could be about 150-155 sq inches or about 950 sqcms. Yes, the B139 is more like a 10 ".

/Erling

[D OB G]

Hi Erling,

I don't want to get petty about it, but yes, 4 X Kef B139s have a moving area of 1400 sqcms.   However a 15" moving diameter (excluding rolled surround) might be 31 cm (at least measured on my emi 1550s- maybe more with some non-double roll surrounds). This means 2 X 15" might have a moving area of 1510 sqcms (say a bit more for non-double rolled surrounds).  So really not so very different.

I agree that it has been a long time since the B110 was considered SOTA (and then only by transistor afficienados, and those in favor of complex crossovers ala BBC monitors).  And I'm not sure that the T27 ever was.

I still maintain that the B139 has qualities (except efficiency) that compete with today's transducers.  (check the price out on the used market!)

David

[maxro]

Enough about the oblong Kef drivers. Back to the subject of the OP:

David,

I had thought about using such a method with a line level passive XO. Simulations in Basta were less than promising, with phase issues causing a major null at the "crossover" point.  Although, a first order filter in its "source/active filter" option probably does not react the same as a line level passive.

Well, anyway, I tried modelling the passive option with a pair of Peerless SLS 12" and a CSS FR125s on a 35cm x 90cm baffle. With a 10mH inductor and 1.24 ohms in series with the parallel woofers and a 100uF cap in series with the fullranger, we end up with a very smooth graph, -3db at around 35Hz. (Am I correct in assuming the resistance I ought to put in the 1st order lowpass filter box in Basta is the DCR of the inductor?)

I think there's something about the way the inductor reacts with the driver's impedance which allows it to work passivly and not actively. 

I'll post a pic if I can figure out how. Do I need some sort of on-line image hosting?

Max

EDIT: I posted a pic in the gallery, album name "maxro".

[D OB G]

Thanks Max,

Now we're getting somewhere.

I don't know Basta, although I downloaded a demo.  Might be better if I ask you, if you don't mind, doing another simulation.
Yes, I would expect the resistance would be the DCR of the inductor.

Thanks for the pic.  I don't know what on-line image hosting is.

Interesting how your graph ends up so smooth.  In agreement with my measured responses.
Interesting to see you can end up with such good, smooth, LF extension with such a small baffle.

Would you mind doing a simulation, using the same drivers (what is the Fs of the SLS 12" ?) using an inductance of 35 mH, and a resistance of 2.2 ohms?
We will probably find that the Fs will limit the LF extension in spite of the large inductor.

Basta doesn't seem to take the rising impedance towards resonance as an issue??
i.e. MJK's concern may still not be addressed.
I think we still need someone to come up with a definitive theoretical answer.

David

[maxro]

I don't know Basta, although I downloaded a demo.  Might be better if I ask you, if you don't mind, doing another simulation...

Would you mind doing a simulation, using the same drivers (what is the Fs of the SLS 12" ?) using an inductance of 35 mH, and a resistance of 2.2 ohms?
We will probably find that the Fs will limit the LF extension in spite of the large inductor.

Done and done (see gallery). I think you will see why I chose 10mH for the inductor. The LF extension is indeed controlled by the driver's Fs. I captured the TSPs in the screen shot for reference. The Fs according to Basta is 24.4Hz, whereas the factory spec is 28Hz. I don't actually own these drivers, so I can't say who is correct, not that they are that far apart.

Basta doesn't seem to take the rising impedance towards resonance as an issue??
i.e. MJK's concern may still not be addressed.
I think we still need someone to come up with a definitive theoretical answer.

I don't know nearly enough about this stuff to begin theorising. I was just playing around with simulations of these Peerless drivers as I have read good things about them for OB and the Fr125s because I own some.

The 1st order slopes are a bit worrisome as Peerless' pdf for them shows a nasty resonance at 2kHz and the FR125s are not getting much relief from low frequency input with a 100uF cap.

Max

[el`Ol]

Hello David!

I would like to hear your opinion which of the many 18" drivers from Precision Devices would be your choice.


Oliver

[D OB G]

Max and el'Ol,

Thanks Max.
Note that a Qt of 0.45 produces the LF hump.
Using this technique with typical inductors, to get the lowest response needs a large inductor, with the APPROPRIATE Qt speaker.

This means a Qt depending on circumstances of, usually, 0.3 or less, which would apply to 18" drivers as well.  Max this cone breakup is indeed an issue with 6 dB per octave.  I'm using Peerless 830491s which have a breakup below 0 dB.  This virtue is hard to find in 12" speakers, rare in 15" speakers, and, I'd say, impossible in 18" drivers.

el'Ol, I'll have a look at Precision Drivers for you.

David

[D OB G]

Hi el'Ol,

I'd never heard of them, but Precision Devices sure do seem to make some magnificent drivers.
I'm not qualified in the slightest to make any judgement on the matter, but maybe I can offer suggestions as to which might work better with my design technique.
I'd rule out any with an Fs higher than 30Hz (especially if you're going to the trouble of 18" drivers)- that seems to knock three off the list.
There are two that have highish Qts- they might be too high to compensate.
That seems to leave the PD184, PD186 and the PD1850.
The response graphs don't seem to be represented in an absolutely equivalent way. e.g.the PD1850 graph stops- what is the high freq. like? Should you take a chance?
That leaves for me the PD184 or the PD186.
PD186 is better made, with better parameters in other respects.
It's probably one of the dearer ones??

I emphasise again that I am totally unqualified to judge a manufacturer's products that I have never seen or heard.
So that's it for what it's worth.

David

[el`Ol]

Thanks, David.

Then I guess the Ciare HW450 would also work. The frequency response seems to be smoother (Or is just the graph smoothed?).

http://www.ciare.com/pdf/catalogo/HW450.pdf

[D OB G]

Hi el'Ol,

Yes, looks good, but as you say, can graphs be compared from model to model?  It does have a rising peak at 2000Hz, which the Precision Devices seem to avoid?
I'd be interested to know the comparative prices.

David

[D OB G]

Max, 

I hope you can help me.

I'm still waiting on the key from Basta, so I'm working on the demo version.

Why do the graphs show such a sudden HF roll off for the LF drivers (has happened with all I've tried)? This would skew the results- they basically show a "native" peaked response, rather than the flattish (above IEC baffle size) that occurs in real drivers.
The response shows no difference between default baffle size, infinite baffle and closed box?
Is this because I don't have the unlocked version?
Obviously it won't let me simulate a baffle roll off.
Why does it calculate a negative Qm?
I obviously have a lot to learn, and I can't find a user's manual that goes through step by step anywhere on the net.
Maybe you know of one?
Hoping you can help.

David

[D OB G]

Max,

I am really confused about Basta.
If I plug in all the ticked requirements for the Kef driver, including Qt, it generates a Qm of 1.07, and yet Kef's own figure is 5.5, which was needed to calcuate the plugged in Qt value in the first place !!
It shows the response in a large closed box (v = infinity) (which I did when mounting into a large garage) to be -3dB at 35-40 Hz.  Yet I measured -3 dB at 22 Hz.  It shows 22 Hz at - 8 dB.
It also shows a high frequency slope again, which I have not measured, nor have I seen it on published graphs.
What am I doing wrong?

David

[maxro]

David,

Play around with the Le and Le loss parameters in the driver tab to get an infinite baffle response that matches your real life measurements.

I believe I read something about getting a negative Q in the Basta instructions: http://www.tolvan.com/basta/Basta!UsersGuide.htm

Make sure you have "enable" checked in the baffle step field of the box tab. It could be that the demo version does not allow this.

Beyond this I don't think I can be a ton of help as I'm new to all this.

Have fun.