Jump to content

Eliptical Filter


Recommended Posts

John, where have you been? haven't heard from you in a while.

Thanks for asking this, I would also like to see it. It was also said you can clip it on the newer models, just how new ? which x-overs can be clipped & which ones can't ?

John can you take a stab at my question on the new subs ?, it's under technical questions on the 12th.


Link to comment
Share on other sites

The unofficial mod list: All K-horn's,Belle's and LaScalla's that have networks newer than

the AA type will have the eliptical filter and will benifit from the

removal of same(lift one lead).All units with AA will benefit from

adding 100~125uf in parallel with the K-33E woofer and 230uH in series

and 4uF in parallel with the K-55V midrange.

I can email you a photo of where to snip if you contact me. Actually I should post a schematic I drew of the x-over as well. I shall endeavor to scan that.


Link to comment
Share on other sites


Look at the tweeter caps. There is an assembly, a diode etc. on the AL3, that is parallel with the caps. Desolder the end that is toward the front of the speaker and

run it to the tweeter terminal on the rear

of the barrier strip.There is a spade lug on this terminal that you move out of the way.It comes from the rear of the filter. It is a visual thing. Look at the caps and it becomes self evident.It is instantly reversable. The assembly is the tweet protection, elliptical filter? or

whatever. The highs became better, less edgy

on my system. If this does not help, it is time for Filmofreddy to step up to the plate.

Link to comment
Share on other sites

I clipped the "red wire" that connects in the series 2uF capacitor and 500uH inductor which is in parallel with the tweeter in my Klipschorns (1999 vintage, AK-3 network). I now have one speaker "Dynamatted, rope-caulked and clipped". I still plan to fill the upper end of the enclosure with polyesther fiberfill, and then sit back and listen for a while.

After clipping the red wire, I listened to Celine Dion, Barbara Streisand, Loreena McKennitt, Kenny G, and Reba McEntire. This session was done in monaural, with lights out, and my bass control turned down 12 dB (this effectively begin to gradually drop out frequencies below say 500 Hz . . . at 200 Hz -3dB, 100 Hz -6 dB, etc.). This allowed me to focus on the changes I had made.

The upshot is I must thank our good friends John Albright and JWGorman for their guidance and pioneering efforts . . . these tweaks are worthwhile. Obviously, we are all splitting hairs at this point, but there is a perceived improvement. Let me recommend that anyone taking these changes on do a serious A-B comparison after one speaker is completed . . . let me further recommend Celine Dion's "My Heart Will Go On", Reba's "For My Broken Heart", and Loreena McKennett's "The Highwayman" . . . simply breathtaking!

I have been in contact with Klipsch to get a more precise definition of the capacitor-inductor arrangement . . . I want to understand precisely what it does from Klipsch's point of view . . . I will share what I learn when I get a response.


Link to comment
Share on other sites

I received a response from Klipsch on the L-C in parallel with the tweeter in the AK-3 network . . . while not an exhaustive explanation, it was something:

"You can remove the branch from

the network, BUT THAT WILL LEAVE YOUR TWEETER UNPROTECTED. At a given frequency,

the roll off goes from 6db to about 50db.

Be sure not to over drive the tweeter with it removed!

We also do not know what the change will do to the sound of the system."

I am positively not worried about overdriving the speakers . . . I am running 100 watts per channel, the loudest I have ever played these is around 1-1/2 watts, and I have measured peaks of about 103 dB on the A-scale.

I have also brought this subject up to five electrical engineers, and no one can precisely explain what the series L-C in parallel with the tweeter does (how it works) . . . I am fascinated and need to find out!

filmofreddy, you have called this a "state variable" resistor . . . what is that and how does it work? What would you feel happens to the sound as a result of its addition or deletion? Where can I learn more?

One electrical engineer called it a low pass filter . . . does this make any sense?


Link to comment
Share on other sites

Dear Phil, working backwards from the tweeter terminal on type AK-B-L networks you have a "state variable" resistor in series with + on the tweeter. This resistor increases its value as excess power is applied going to a maximum of, I believe, 100ohms. This works in concert with the "eliptical filter" in shunt with + - of the tweeter leads. This was Klipschs' answer to the complaints of the filthy mcnasty things those zeners in the tweeter circuit of the type AA networks could do to some solid state amplifiers. Plus, under drive (well within the safe capabilities of the tweeter) they sounded bad. So what happens when you "over drive" the tweeter with AK-B-L type networks? The resistance increases in the variable resistor, that increased resistance in conjunction with the "eliptical filter"(500uh & 2uF in shunt) acting in concert with the pre-ceading tweeter x-over components( 2uF in series 125uH in shunt followed by another 2uF in series)to RADICALLY increase the rate of slope( Klipsch said 50db, Believe it!) to stave off the destruction of the tweeters voice coil from excess midrange energy. I think thats a fairly accurate description. My advice still stands, That network is there to protect the speakers from abuse, NOT neccessarily the best from a sonic standpoint. Go back to a simple single capacitor about 3.5 to 4 uF in series with the + on the tweeter and live happily ever after(Until we can "upgrade" to a single horn top end). Until then thats my half-baked explanation and advice. Peace(and thanks to AL K. for the correction of my phrasology and the plot of the "eliptic filter" in action. Now I'm convinced more then ever GET RID OF IT!

Link to comment
Share on other sites

Well, I have taken filmofreddy's advice of going beyond the removal of the "elliptical filter" from the AK-3 network on one of my Klipschorns (the same speaker that has been Dynamatted, rope caulked, and stuffed with polyesther fiberfill). I have now removed the 125 uH inductor which is in parallel with the tweeter, leaving me with two 2 uF capacitors in series with the tweeter.

After a short audition, I am quite pleased. I will continue listening for another week and decide whether to replace the two capacitors with a single 3.5 - 4 uF Hovland (or other such upgrade).


Link to comment
Share on other sites

I shall post both a schematic and a photo for all those interested in this mod and would rather see photos. If you're reading along, check the LC sections in your book. All of you who have emailed me lately, I will get back to you soon. I've been extremely busy lately.

Where shall I post these photos/jpegs? Suggestions?


Link to comment
Share on other sites

  • 22 years later...

There was another post I bumped shortly after this one, so I could find it in ‚Äúmy activity streams‚ÄĚ, ‚Äúcontent I posted in‚ÄĚ and it has disappeared, like a witness for...


Anyway, it was started by ALK, and another member had just posted some plots after I bumped it.


I hope he’s ok! 

Link to comment
Share on other sites

14 hours ago, Curious_George said:

I love (loath) a technical thread with no pictures, graphs or charts to show you what is going on...


I've read quite a few threads about this filter.  Here's my uneducated take on it...


I believe this is a three pole Chebyshev "Constant K" filter combined with a null pole.  As I understand it, "Constant K" implies that the filter is designed for a circuit with equal source and load impedances. That's why the two series capacitors are equal.  When this filter is used with a source impedance much lower than the load impedance, you get the familiar Type AA tweeter transfer function.


I found this calculator which works great.  Took a lot of trial and error to make it work.  Someone smarter than me should be able to explain why it works.




For a high pass filter, I multiply the desired crossover frequency (-6db point) by 1.6.  For a 6000hz cross, that's 9600hz.  The screenshot shows the data entered.  Note that the input impedance and output impedance are both set to 8 ohms (Constant K?).  The Klipsch elliptical filter closely matches the computed component values.  The 125uh inductor appears lower the crossover by around 300hz.


The values for the null pole are pretty straight forward.  Use the same value capacitor in series with an inductor that's 4 times the value of the Chebyshev shunt inductor.  That would be a 2uf capacitor in series with a 460uh inductor, or in Klipsch's case, a 500uh inductor.


The second screenshot shows both the Chebyshev filter alone and the Chebyshev filter with the null.  I used the calculator's values for the simulation.


The same calculator can be used to design a low pass filter as well.  Instead, though, you divide the desired crossover frequency by 1.6.  For the null, use the Chebyshev calculated inductor value in series with a capacitor that's 25% of the Chebyshev shunt capacitor.


Again, I'm no EE.




Screenshot (141).png

Screenshot (142).png

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.

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...