[evozero]

Hi All,
Just to bring you up to speed, I didn't get any time for speakers / forums until yesterday. The weather was beautiful so out came the tools. I tried to incorporate everyone's suggestions and keep them a little WAF friendly. What do you think?







Listening last night, i was say the imaging / stage depth has increased by a large margin. I really like this Dipole effect.
I am going to have a proper listen tonight.
The plate amps and crossover part will be with me in a week of so, depending on customs.
Rudolf, did you have any more thoughts regarding the crossover?
Many Thanks
Ian

[Danny Richie]

That looks like it will give you a lot better results.

Next is that they are so close to the wall that the reflection will sound like a time smear of the original on axis response. You need to get them a good 3 feet or more off the wall so that the reflection will sound more like a delayed reflection.

[Rudolf]

Ian,
while this should be a big step forward compared to what you started with, this will NOT give you better results than the simple rectangular baffle I already recommended for the AN:



The fully circular baffle will rob you of some dearly needed efficiency below 200 Hz, it will emphasize the peak from 500-1000 Hz, and it will add a dip around 1600 Hz.

Pink is the rectangular baffle which I recommended, red is the same but with the top like in your present  baffle (if you prefer the looks).
My RCL filter will care for the region 0.4-1.2 kHz, but does not compensate for the dip.

Luckily you don't need to junk any of your work by now. When you have received and installed the filter, you could just cover the open part between OB and H frame with a rectangular piece of wood or heavy cardboard to see what kind of performance you like more. If SWMBO prefers the hovering OB you should think about a square piece of glass between H frame and OB.

More about the crossover later.

Rudolf

[Rudolf]

Hello Ian,

coming back to the crossover:
The Beta 15 should be connected to the plate amp directly. If I remember correctly, your plate amp (is it the SA240?) has a low pass of 24 dB/oct (4. order). It would be nice if you could find out, whether the crossover is of Linkwitz-Riley, Bessel or Butterworth type. This is not crucial, but could be helpful for some fine tuning. The low pass should be tuned to 110 Hz.
Most subwoofer plate amps are notorious for a bad consistency between the frequency values shown on the dials (if any) and the actual X-over frequency. This would be another reason to write an email to PE asking for some real values.

The filter network for the AN should consist of all these parts:



In this configuration (and after dialing in the right x-over frequency and boost for Beta 15) the response of the system will be like this:



The wider lines show the simulation with MJKs worksheets. The thin lines are the simulation with Boxsim. From 0.3-1 kHz you see the influence of the notch filter, which MJKs worksheet does not simulate. The narrower response of the Beta 15 in Boxsim is due to the H frame option missing in Boxsim. I had to simulate with a wider baffle, which is not the same in every respect. Both drivers work in phase (same polarity).

When everything is connected I would definitely INSIST that you toe in both loudspeakers until their driver axis' to the rear point into the room corners. The distance from the corner to the back of the H frame should be no less than 1 m. To the front both driver axis' should cross about 30-50 cm in front of your nose. Later you may position the loudspeakers any way you want - but I'm quite sure you don't want to after you have listened to that configuration.

Rudolf

[gvimhoof]

Ian!!! 

Love the photo with the desk lamp providing a single angled light source for the picture- very dramatique'

What are your plans for finishing?  Are you going to stain, paint, incorporate that fabric you started with?  I would vote for something clear and durable on the bottom of the circle to the top of the H-frame and then take that fabric you started with and cover the circular baffle. 

And I agree with Danny.  Make sure you pull those out from the wall at least 3ft.

Nice aesthetics on that!

[Danny Richie]

From my experience with these drivers, I don't see that network working well at all.

For one, if the lower woofer is powered with its own plate amp then there is no reason for the added resistive load to the full range driver.

Secondly, with a cap that large in line with it, it is going to be receiving a fair amount of low frequency information that might burn it up if played hard for very long.

Thirdly, due to the rising response of this driver, the 47uF cap value should be omitted also. This will also mean that the resistor value that by-passes the inductor can be shifted around a bit to balance the top end with the lower range. It might need to be adjusted to a level between a 4 ohm and 6 ohm.

[evozero]

Hi All,
I am ready to do what i am told!
I have some difficulty positioning the speakers 3ft from the rear wall, it blocks the door way! But i shall try playing with the position.
At the moment the AN8 is 2ft from the wall, that is why i chose to put the baffle on the front of the H frame.
from the SA240 manual;
high level output; The signal to the L/R speakers will be shaped by an internal 6 dB/octave 125 Hz high pass filter.
Frequency: Used to determine the frequency of the internal variable electronic crossover. Adjustable from 40 Hz to 180 Hz, and has a slope of 24dB per octave.
I will email PE to try and find out more.
Danny, may I ask how to deal with the small baffle, as Rudolf pointed out;
"The fully circular baffle will rob you of some dearly needed efficiency below 200 Hz, it will emphasize the peak from 500-1000 Hz, and it will add a dip around 1600 Hz."
Is it just a compromise between the small baffle / better imaging and rectangular baffle / higher output?
I will get some cardboard and try it out.
Thanks all, looking forward to trying out your suggestions.
Ian

[evozero]

Hey gvimhoof,
 I like the aesthetic ideas, i will get the wife involved, she has a good eye for design and chose the fabric.
The front baffle may well change depending what Danny & Rudolf come up with, its all a little over my head
But keep the ideas coming, Thanks.
Ian

[Rudolf]

From my experience with these drivers, I don't see that network working well at all.

Danny,
it doesn't help neither Ian nor me, if you only tell us what is wrong from your point of view. How about telling Ian how you would do a better crossover?
 

For one, if the lower woofer is powered with its own plate amp then there is no reason for the added resistive load to the full range driver.

OK, I'm not married to that resistor. Leaving it off will have no major effect on my crossover design.

Secondly, with a cap that large in line with it, it is going to be receiving a fair amount of low frequency information that might burn it up if played hard for very long.

Ian has lived with that cap before and not burnt the AN. So I hope it will be safe for the future too. Problem is: If we move the crossover point for the AN to a higher frequency, the Beta will have to follow. But by doing that, the peak frequency of the subwoofer section will get into the 100 Hz region - something I want to avoid.

Thirdly, due to the rising response of this driver, the 47uF cap value should be omitted also. This will also mean that the resistor value that by-passes the inductor can be shifted around a bit to balance the top end with the lower range. It might need to be adjusted to a level between a 4 ohm and 6 ohm.

Danny,
we are constantly talking about different things here. The RLC filter is NO low pass against rising highs. It's ONLY purpose is to notch the dipole peak of the OB at 0.4-1.2 kHz. It has NO effect beyond that.
You would be very welcome to suggest some values for an ADDED low pass against the rising highs. If you don't know which values should be best - who else will.

from the SA240 manual;
high level output; The signal to the L/R speakers will be shaped by an internal 6 dB/octave 125 Hz high pass filter.

Ian,
if you could use this high pass filter in addition to the 150µF cap this would be helpful. But it is restricted to a 'high level' input and output. I don't see how you could use it without an additional amplifier stage. If it is just a cap in line with the plate amp output, you can`t use it directly connected with the filter network.

Rudolf

[Danny Richie]

Mmmmmmm, there are clearly several misunderstandings about what I posted.

I am really busy right now with getting orders out. I will come back to this at the end of the day and be very specific. I will posted some measured responses, some crossover typologies, and show cause and effects.

Hang in there and I'll be back.

[Danny Richie]

Okay, let me address a few things then on to the meat.

Secondly, with a cap that large in line with it, it is going to be receiving a fair amount of low frequency information that might burn it up if played hard for very long.

Ian has lived with that cap before and not burnt the AN.

I wasn't talking about burning up the AN. If you can keep it from reaching its mechanical limits then it will handle plenty of power.

I was referring to the resistor in line with the big cap value.

A resistor can only handle its rated power level for short bursts if run full range. Long term power levels at or above its rated power limit, using a full range, signal with quickly burn it up.

If you put a cap in front of it then that will block the lows and it will handle considerable more power. The smaller the cap, the more lows are blocked and the greater the power handling. When the cap gets pretty large, like the value used here, then it lets quite a bit of the lows pass through it and power handling is very limited.

we are constantly talking about different things here. The RLC filter is NO low pass against rising highs. It's ONLY purpose is to notch the dipole peak of the OB at 0.4-1.2 kHz. It has NO effect beyond that.

Oh, I know, but that is not what it needs.

from the SA240 manual;
high level output; The signal to the L/R speakers will be shaped by an internal 6 dB/octave 125 Hz high pass filter.

If you are thinking of running the signal to your main speakers through the sub woofer plate amp then DON'T. All it does is put an electrolytic cap in line with it. This is NOT what you want your signal passing through.

I have to go grab so supper and I will get to the rest of the information as soon as I can.

[Danny Richie]

Okay lets take a look at one of these full range drivers mounted in an open baffle. For now let's just say that this 12" version is in the same family.

Here is a factory made measurement. Nothing like a 120db range to make a curve look good huh.



Note the rising response.

Now here it is on a baffle not much bigger than the driver.



Looks a little rough doesn't it?

The little bump around 750Hz is likely the result of the baffle width like Rudolf was pointing out early.

You can still keep the baffle narrow and maintain the better imaging and still not have this little hump if you work with the length of the support wings on the sides that help support the front baffle. A little deeper on those sides (back side) and you can shift the little peak down into the range where there is a little dipped area around 400 to 500Hz. Be careful though. If the sides get too deep then you get a little cavity resonance to deal with.

We have several things to deal with here. First we have a rising response. Then we have a couple of bad peaks that need tamed. We also need to control this without the impedance getting out of whack.

So to straighten it out we need something like this:



The resistor and inductor at the front will fold over the top and balance it out. This puts resistance in the upper ranges but leaves the lower ranges alone.

So anything we do to notch out the peaks needs to be in shunt to lower the impedance in those ranges and bring it back down to a consistent level. Also keeping the LCR networks in shunt keeps them out of the signal path. So nothing that you hear comes up through those components.

The result:



Want to see an overlay of that?



It went from un-listen-able to not bad....

It still has quite a bit of stored energy in those peaked areas (break up) that I just can't stand. But it is a whole lot better than it was before the correction circuit.

[DanTheMan]

Danny, what are you getting off axis?

Dan

[Danny Richie]

Like any large full range driver, the off axis falls apart quite quickly.

[DanTheMan]

Do you take that into consideration when you EQ it?

Dan

[Danny Richie]

Do you take that into consideration when you EQ it?

If you note my notes on the network diagram...

"An 8 ohm resistor yielded the most accuracy but it sounded better with a 7 to 6 ohm resistor."

This was from listening evaluation. I also preferred the speakers to not be aimed straight at me but about 5 to 10 degrees or so off of that alignment. That worked best in my room.

[jhm731]

I'veeing testing the AN S8 Alnico.

Here's it response on Scott Faller's OB:  http://www.enjoythemusic.com/magazine/equipment/1009/lowther_high_efficiency.htm




Here's it's response in the CSA monitor(24"H x 13.5"W x 11.25"D) :





No networks were used with either set up.

Dan

[Rudolf]

Yes Danny,

we were really talking about different things most of the time. Thanks for clarifying that with your last posts.

Regarding the 4.6 Ohm resistor - Ian should keep an eye (or better hand) on it and see (sic!) how hot it gets. Or simply leave it out.

If the 6 dB filter in the plate amp really is only a cap in the output - which is to be feared - Ian should avoid that output, yes.

Regarding anything above 1 kHz you obviously have the hands-on experience with the ANs - which I don't. When Ian started this thread, he didn't complain about that region. So I did not care for it. But I have seen detailed non-factory measurements of the AN 6.5 S and AN 10 S which point in the same direction as yours. I can see a problem in fixing that midrange peak. Obviously all AN drivers have this 'feature', but every one peaking at a different frequency. So where is that peak for the AN8? Are you saying that your network (shown in your diagram) will work for the AN8?

If your network is the right one, the dipole peak below 1 kHz would be part of the solution and my 0.4-1.2 KHz notch filter should NOT be installed - right?

 

[Danny Richie]

Regarding the 4.6 Ohm resistor - Ian should keep an eye (or better hand) on it and see (sic!) how hot it gets. Or simply leave it out.

There is no need for it to be in there if the lower woofer is powered by its own amp.

So where is that peak for the AN8?

Judging from the factory measurements there is a pretty good sized (about 6db) peaked area in the 1500Hz range. There may be others above it as well but you'd really need to measure it to see how bad it really is. The LR network at the front will fold over that upper end output and balance it to the bottom end. So that will help a lot.

Are you saying that your network (shown in your diagram) will work for the AN8?

The notch filters used to control the peaks, in the larger one that I measured, will be different than any needed for the Super 8, but the LR network at the front will be essential.

If your network is the right one, the dipole peak below 1 kHz would be part of the solution and my 0.4-1.2 KHz notch filter should NOT be installed - right?

If the baffle is too small then it will still have a peak that might need a notch filter, but it needs to all be in shunt (parallel) and not in series with the driver.

If the baffle shape is just right then that peak can be pushed down in range to lift the area where output is lost in the lower region.

I recommend a baffle be no wider than it has to be so as not to hurt the imaging. However, it can be narrow and tall. That can give back or push the peak lower in range without effecting the imaging too much.

Some type of side panels to the back of the driver can be used to help push the peak lower in range as well.

It is a balancing act and it really helps to have some real measurement gear to see where it all is.

[DanTheMan]

If you note my notes on the network diagram...

"An 8 ohm resistor yielded the most accuracy but it sounded better with a 7 to 6 ohm resistor."

This was from listening evaluation. I also preferred the speakers to not be aimed straight at me but about 5 to 10 degrees or so off of that alignment. That worked best in my room.

All I can say is "that is confusing."  Only b/c it seems like you are saying inaccurate sounds better.  What you are probably saying is that it sounds more accurate with the 6-7 Ohm resistor 5-10 degrees off axis.  Id sure like to see a polar plot so we can know what it sounds like.

Any case, I have found that OBs tend to do better nearly straight ahead, but slightly toed in to the middle.  The U-baffles I've made didn't have this same character.  I'd really like to see some one build an L-baffle so that you might be able to get a radical toe in to improve the IACC.

Dan