Jump to content

New extreme-slope crossover requiring NO Zobel


Recommended Posts

Guys,

Here's an entirely new crossover design being developed for Shawn Fogg

(Sfogg). I expect to make it available for use in 2-way or 3-way systems

based on the Belle Klipsch or LaScala woofers. This will include the stock

Belle and LaScala. I will also be posting the design itself, for the DIY

people, including the schematic in later posts. I want to delay posting the

layout drawings because it will be critical because of inductor coupling

problems. I also expect to change the layout a bit on the next prototype to

address the problem.

The design crosses over at 600 Hz (actually about 605 Hz) and is an extreme-

slope design. The woofer filter reaches 22 dB down at about 680 Hz. The

highpass filter reaches 26 dB down at about 530 Hz. Rejection is equal to or

greater than that as you go away from the crossover. This gives a window of

only about 50 Hz centered at 605 Hz where both drivers are less than 10dB

down and can interfere with each other.

The new innovation with this design is that it compensates, and actually

utilizes, the inductance of the woofer voice coil WITHOUT the use of the

usual Zobel R-C combination. It was found that the Belle woofer complex

impedance (and I assume the LaScala woofer) actually measures 5.9 +j4.01 Ohms

at 600 Hz. This is equivalent to 6 Ohms in series with a 1.06 mHy inductor.

The lowpass filter in this network was synthesized to have a 1 mHy inductor

at the output. This inductor was simply left out of the filter and becomes

the woofer inductance. The results of the combination measures a very flat

resistive 6 Ohms impedance all the way from the woofer resonance to beyond

the crossover. The reactive component of the impedance measures about 1.4

Ohms right at the crossover input. This is the worst case and represents a

"VSWR" of only 1.27:1 in a 6 Ohm system. "VSWR" can be thought of as a

measure of impedance mismatch which includes complex impedance. Anything

below 1.5:1 is considered good. 1:1 is a perfect match. The speaker will be

transformed to a 6 Ohm speaker which should operate very nicely from the 4 or

8 Ohm taps of any power amp.

The machine-run plot of complex input impedance is attached. The plot was

done with the Belle woofer connected to the lowpass filter and an 8 Ohm load

resistor terminating the highpass filter.

More in the next post.

Al K.

IT SEEMS THE IMPEDANCE PLOT GOT LOST. I HAVE ATTACHED IT DOWN A FEW POSTS BELOW.

post-2934-13819250078928_thumb.jpg

Link to comment
Share on other sites

Guys,

I have attached a photograph of the new network. As you can see, it is quite

large and requires 4 inductors and 9 caps. The base board measure 13.5 X 8

Inches. It's a little difficult to see, but the large inductor at the lower

left corner is jacked up 1 inch to center it with the inductor next to it

(above it in the picture). This area will be laid out differently in the next

prototype. The two coils wanted to talk to each other messing up the slope of

on of the filters.

Note that there is a removable strap connected between the 4th and 5th screw

terminal on the barrier block. This connects the output of the highpass

filter to the input of the autotransformer and swamping resistor. The strap

may be removed so that an external crossover from a squawker to a tweeter may

be connected with the squawker output fed back to the transformer. This will

allow 2-way or 3-way operation. The transformer taps are connected to plugs

so that the level can be adjusted just like with my type "A" replacement

network. Because the transformer could also be located on the external

squawker / tweeter network, the transformer and tap plugs will be optional.

Al K.

post-2934-13819250079928_thumb.jpg

Link to comment
Share on other sites

Colin,

The wires you are seeing are the leads from each of the transformer taps with a plug on the end. They plug into shove-on-terminals connected to the high output of the netwrok. You move them to adjust the output level for various drivers. This is identical to the arrangement on my "stock" replacement network.

A network for the Khorn will be a bit more trouble. That's down the road!

As to pricing, I'm not worried about that yet. For now, it is just a developement project.

Al K.

Link to comment
Share on other sites

Al, here is a chart John Warren presented prior of an unbaffled K-33 vs. baffled in a Khorn (IF I'm not mistaken). I'm amazed, but maybe I shouldn't be, as to the > 2 ohm delta between your Belle measurements and the Khorn and wonder how much of a spread is possible between the Belle and the Lascala??

While the spread may not seem to be much in the 'real world', once again I wonder of its impact on XO design parameters. Could one really hear the difference of a couple of ohms Z, and the subsequent change in values in L/C??

I read your paper on extreme XO's, and have questions, but will email you at a later time when I have a firmer grasp as to exactly what I'm asking...

Thanks

TC

post-13284-13819250081948_thumb.jpg

Link to comment
Share on other sites

TC,

I am courious about the difference between John's plot and mine. I think his was of the raw k33 driver OUTSIDE the hron (I think). His plot is also simple impedance. My plot includes the reactive (inductive) component. I think there is also some difference between horns too. I really wish I had a LaScala and a Khonr here to do measurements on!

The significance of the impedance matching really relates to the performance of more complex filters. A simple 6 dB / octive filter can operate into anyhting. It's all over the place anyhow! A more complex filter, even a 12 dB / octave, starts require a good terminaiton to act as it should. It also matters to some amps. I suspect a multi-million dollar Krell wouldn't care what it's terminate with, but a little SET "one lunger", as I call them, should see a good load. Their source impedance is pretty high (poor damping factor). In my case though, it's just "my thing" to work to a falt impedance. It comes form the absolute need to do it at microwave frequencys where I "grew up"!

Al K.

Link to comment
Share on other sites

Guys,

Well, the say: in for a penny, in for a pound. I might a well include the

plot of the network by itself. Here's the high and low outputs plotted into

resistive loads with an external 1 mHy inductor between the lowpass section

output and the load resistor. The insertion phase is also shown. If anybody

is interested in Group delay, it looks to be somewhere between 2.5 and 3.0

mSec at the crossover. It goes way down on either side of the crossover very

quickly. I can't measure it accurately though. I can only calculate it from

the phase data at each frequency point. In any case, I am quite sure it is

below the limits set by the well known B&L criteria. I have come to believe

group delay is not important anyhow.

Al K.

post-2934-13819250082488_thumb.jpg

Link to comment
Share on other sites

John,

The impedance method you are used is quite valid. It's described in a

book I got here by Dickason as the "Voltage divider method". I used to do it

that way myself. The only problem is it doesn't tell you what part of the

impedance is reactance.

The instrument I am using is very old but fancy beast made by

Spectral Dynamics called a "Dynamic analyzer II" model SD375. I got it on

eBay. It cost me roughly $70! I also am using a HP 7470a digital plotter,

also from eBay for less the $10! Remember, my main business is filter design

software. Crossover networks began as a hobby too. I never believed when I

started doing this that anybody would actually want me to build these things

for them and actually pay me to do it!

Anyhow, The SD375 is a dual FFT spectrum analyzer that maintains

phase data between the A and B channels. That gives it the ability to compute

and display all sorts of information about the two including transfer

function between A and B channels. I have a bridge set up with two 3300 Ohm

1% resistors and a 10 Ohm resistor as a reference on one side. That runs

channel A. With the unknown on channel B, it displays complex impedance as

the ratio of A to B in the transfer function mode. The scheme is fairly

accurate so long as the unknown doesn't get more than about 2 times the 10

ohms reference. It depends on a constant current. That goes to pot if the

unknown gets large enough to unbalance the bridge. I think it's actually

within the ball-park up to about 50 Ohms.

I have attached a picture of the SD375. The big down-side to the

thing is that it weighs 55 Lbs! Lugging it down the steps from the office to

the listening room is a real drag!

Al K.

post-2934-13819250083438_thumb.jpg

Link to comment
Share on other sites

Shawn,

I think you will be quite pleased. I could only compare it to the 3-way arangement of my Belle + JBL 2426h and Beyma. pink noise had a slightly higher pitched quality through the 3-way system than the 2-way using the Altec 902 driver. I can only assume this is due to the better dispersion of the Beyma up high compared the the Altec 811bs dispersion up high. I expected that though. I will say tis though, the JBL 2426h woudn't make a bump on an Altec 902s posterioe running 2 way! The jebl really NEEDS the help of the tweeter!

I am very anxious to hear how you like it especially if you try it 3-way using my standard network as the squawker/tweeter crossover. When you do that I need you to short out the 39 uF cap and leave its woofer port open.

AL K.

BTW.. The crossover and driver were shipped this morning by FedEx ground in seperate boxes.

Link to comment
Share on other sites

Al,

" pink noise had a slightly higher pitched quality through the 3-way system than the 2-way using the Altec 902 driver."

It will be interesting to see how this compares to a t35. The Altec looks like it goes higher but maybe not with quite as good of a dispersion. I'll try and doing some testing of that when I get a chance. If the high end is a little soft my pre/pro can do a little digital EQ to bring it up a touch if needed.

"I am very anxious to hear how you like it especially if you try it 3-way using my standard network as the squawker/tweeter crossover. When you do that I need you to short out the 39 uF cap and leave its woofer port open."

So just feed the standard network from the high pass section of the ES600 and leave the standard networks woofer out completely open or should I terminate it with a resistor?

One of the things I want to do is build a relay box that will let me swap between the standard crossover used as a three way vs. ES600 crossover as a two way via remote control from across the room. Easy enough to do.. just need to find the time build it. It could also be used to compare other three way crossovers or even to compare ES600 two way vs. stock 3 way setup with AA for example.

Looking forward to receiving the network. Thanks again for all the work on this.

Shawn

Link to comment
Share on other sites

Shawn,

No, don't terminate the woofer output port. You want the 400 Hz woofer crossover totally out of action. The ES600 will perform that job. That consists of the 2.4 mHy inductor and the 39+1uF cap. Open the inductor (no termination) and short out the cap. This leave the input connected directly to the 6000 Hz squawker / tweeter crossover. You will then be using the autoformer in the convention "A replacement" network to run the squawker. In 2-way operation you will need to use the transformer on the ES600.

Al K.

Link to comment
Share on other sites

I received the ES600 from Al earlier this week but haven't had any chance to play around with it till now. I haven't lisened to it too much so far but my initial impressions are favorable. This is using the LaScala as a two way with the regular bass horn but from 600hz and up is handled by an Altec 902-8B on a 511b horn.

The attached is a FR measurement with the mic in the 511b horn to get an idea on how sharp of a crossover this is.

Shawn

post-12845-13819250084508_thumb.jpg

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...