Jump to content

K-Horn crossovers - from AK-3 to Universal to ES


Recommended Posts

Just read the discussion on Brick Wall crossover filters and at the risk of being beaten around the ears by the Klappenstick I will ramble on about my experience with K-Horns and crossovers.


I bought the speakers new in 1990 and had not altered them. Early last year a visitor, who is a professional jazz singer, commented after listening to Erykah Badu's Baduism CD that the speakers were certainly dynamic but were lumpy in their response. This prompted a revisit to the Klipsch forums and a quick look around indicated that the 26 year old crossover caps needed replacing. I ordered the kit from Crites Speakers and while waiting for the caps to be delivered I started reading what people had done to old K-Horns.
The Sonicap equipped AK-3 crossover certainly sounded different, the squaker squaked and the highs were all there. It was like a blanket had been covering the top hat.


I read that the best thing to re-vitalise an old K-Horn was replace the 401 mid horn with a wooden tractrix horn. A few emails showed that was an expensive path - $380 for freight alone for the Volti horn. The Eliptrac freight is over $200 for the DIY panels. Then there is the driver cost plus freight.
Too expensive, next best was going to be crossovers.


I read that the stock Klipsch K-401 mid horn induces harshness at higher volumes above 4500Hz so I ordered Bob Crite's CT120s to replace the stock Klipsch K-77 tweeters and built an Al Klappenberger (ALK) designed Universal crossover, the 4,500Hz version, to replace the re-capped AK-3 crossover. I had found the circuit diagram on the Klipsch forums.


The ALK Universal crossover's performance in my home and for my listening levels and music was well worth the effort. Compared to the AK-3 there is an improved soundstage, it is easier to pick out individual instruments and the upper midrange is far less harsh especially when the volume is turned up.


I have a NAD C725BEE receiver so there is plenty of power for the swamping resistor on the autotransfer. The room is 36' (11m) by 22' (6.7m) with 8' (2.5m) ceiling, heavily damped by couches and carpet with kitchen / dining area one end, K-Horns in the corners at the other end , TV and slow combustion fire between the speakers and couches in the middle of the room. Music is served by a Logitech Squeezebox server and Raspberry Pi player or from ABC classic FM.


As an aside, the NAD amp has pre out and main in RCA sockets on the rear allowing a passive attenuator to replace the supplied strait-thru jumpers so that the volume control doesn't permanenty live around 7 to  8 o'clock when driving highly efficient K-Horns.


My searching through the Klipsch forums indicated that for a big well damped room the Extreme Slope networks from Al Klappenberger would work well. Tri amping and DSP electronic crossovers were ruled out in favour of the simplicity of passive networks and a single amplifier.


Some years ago I had downloaded from ALK's site all of the available DIY instructions which included an Extreme Slope 400Hz woofer/midrange crossover for the K-Horn. These were retrieved from an old archive drive and re-read. ALK has removed these DIY instructions from his site due to someone altering a component or two and then selling the design as his own.


On ALK's site (http://www.alkeng.com/klipsch) there is a paper entitled "A case for extreme-slope crossover networks", which puts forward the case for an extreme slope network as an alternative to time alignment of speaker drivers, especially in a 3-way horn loaded speaker. The paper is an interesting read (and re-read a couple of times to understand it) and supports ALK's comments in the earlier mentioned Brick Wall crossover slope thread.


The paper also includes a worked example of a 7,500Hz crossover and explains how to calculate component values for other frequencies along with strong advice not to DIY these networks due to performance being easily compromised by mutual inductance from bad  placement of the inductors.


I built a crude spreadsheet to calculate the components for a 4,500Hz network and confirmed the performance by modeling the networks in LT-Spice. I then rounded each of the component values to available values - e.g. 0.439222mH to 0.45mH and checked that the predicted outcome in LT Spice would not introduce too many bumps in the frequency response. I also used LT Spice to simulate the ES400 networks especially as I wanted to modify the woofer leg of the network to utilise the existing AK-3 4mH inductor and associated capacitors rather than buy new components.


Living in Australia imposes a few challenges for speaker DIYers, we are a small market with a limited choice of local crossover component suppliers, however Speakerbug is the Asia Pacific agent for Jantzen and carries good stock and provides speedy service. Importing inductors doubles their cost and takes time (3 weeks minimum), so I tend to prototype using what I have on hand or can get locally from Speakerbug.


The K-Horn top hat has plenty of room for crossovers (and in my home the top hats served as a hiding place for rarely used but quite valuable silverware and jewelry, now they are full of crossovers) and while I appreciate that for a commercially produced crossover a minimum board size is an important design criteria, but for a home DIYer prototyping a design to try out, a generous sized ply base board allows the component layout to follow the schema and makes for easy swapping of components without requiring a complete rebuild when a component changes, especially if the prototype was built with lowest cost components (read non polarised electro caps) and it sounds so good that you decide this is a keeper and therefore you should swap in those expensive overlarge film caps. A generous sized base board also makes it simpler to minimize mutual inductance of the inductors. So no photos as I don't want to give ALK another target for his dart board.


I first built the squaker side of an ES400 woofer/squaker crossover and hooked it up so that the AK-3 handled the woofer roll-off and the Universal crossover managed the squaker/tweeter crossover at 4,500Hz. There was a subtle but noticeable difference with mids being just that bit clearer making positioning of instruments on the soundstage more definite.


Then I built an ES4500 squaker/tweeter crossover using solid wire air cored inductors and Jantzen Standard Z-Cap capacitors. This required removing the autotransformer from the universal crossover board and building a board with the speaker terminal block and autotransformer on it so that the Universal or ES could be swapped in or out.
Listening tests with the ES4500 networks reveal even more detail. And when you turn the volume up, wow! No harshness, just crystal clear.


Finally I built the ES400 woofer roll-off portion of the network on the woofer hatch cover. This means that the woofer has an extreme slope roll-off 3db down at 400Hz and the fuses and feed to the top hat remain standard.


So, how does it all sound?


Sensational ! Very good soundstage and can go very loud and stay crystal clear. Extreme slope networks are criticised for sounding "thin" when turned down low, well speakers 6 metres apart are going to sound thin when turned down low if you stand between them but from the other end of the room they are fine.
Here are some REW sweeps measured at the speaker terminals - the woofer response is lumpy, I suspect because the measurement is taken with the woofer hatch open. Apart from the woofer lumps, the actual measured response is very close to the LT-Spice predictions. The midrange is 7db down as set on the autotransformer. Most Klipsch crossovers have the mid between 3 and 6 db down. It's a matter of taste.

REW Capture of ES outputs at speaker terminals.jpg

  • Like 1
Link to comment
Share on other sites



Wirrunna,

You investigation on my Klipschorn extreme-slope network it very close to what I expect. Well done...

Your plot taken across the speaker terminals is the correct way to evaluate what the network is actually doing. I have attached a computer plot done on the ES400 lowpass using complex impdeance data taken on a Klipsch K33 woofer in a wood box simulating the back chamber for a Khorn or LaScala. The rough response is what you get when you load a filter with a woofer driver directly. The other drivers are attenuated. That isolates the uneven load of a driver from the filter.  The increased loss approaching 5 dB at the 400 Hz crossover is suspicious. The measured loss in a typical ES400 lowpass has been consistently about 1.5 dB.  You might have smaller gauge wire that I use.  That might account for the extra loss.

I just completed building an ES400 and will have it here for a few days. I could do fresh testing using the K33 in the box if you want to do some comparisons.

Of course, shelving is the reason for the midrange level being lower than the woofer. I normally set it to 6 dB. Your plot shows more like 10 dB. I suppose it's a matter of taste or a different midrange driver.

Anyhow, it all looks good to me.

Al K.


 

es400Woof.gif

Link to comment
Share on other sites

Al,
First, thank you for taking the time to reply.


Second, I was half expecting a blast as I have broken what I percieve as three of your fundamental rules in respect to K-Horn crossovers - changed the mid/high crossover to 4,500Hz, admitted to building old fashioned breadboard style and finally not using the parts you specified. I offered explanations for each of these breaches in my post.


The ES400 woofer re uses the AK-3 components. For lower DC resistance I used Jantzen P-Core (http://speakerbug.com.au/index.php?route=product/category&path=18_45)  15AWG coils for the higher values and an Erse 14AWG air core for the 0.62mH. So the DC resistance in series with the woofer is 0.21, 0.15 and 0.16 ohms. I will send you the LT-Spice .asc file by PM, however I will have to change the file type to conform with rules, so simply change the file type back to .asc and you will be able to display it in LT-Spice.


I will also include the ES4500 spice model.


After hearing the difference that the ES4500 made I ordered 14AWG Litz wire inductors that are in series with the squaker, they took 6 weeks to get here and the postage cost more than the coils.

Link to comment
Share on other sites

6 hours ago, Al Klappenberger said:

 The increased loss approaching 5 dB at the 400 Hz crossover is suspicious. The measured loss in a typical ES400 lowpass has been consistently about 1.5 dB.  You might have smaller gauge wire that I use.  That might account for the extra loss.

 

Oops, I just found that the bass control is sitting at about 2 o'clock on my NAD amp, has been for years judging by the dust. So its not so much a loss toward 400Hz as a boost toward 25Hz.

Bugger, for my own peace of mind I'm going to re-measure with the tone defeat button pressed.

Link to comment
Share on other sites

The bass control turned up would cause the "sag" at the crossover which is really just a boost at extreme low bass.

 

I just did a measurement plot of the ES400 woofer and transformer output set at 6 dB. I'll post that later.

 

Changing to the 4500 Hz tweeter cross is ok.. It is promoted as a way around the poor sound of the K400 mid-range horn. I advocate leaving it a 6 KHz and upgrade the horn instead!

 

Al K.

 

Link to comment
Share on other sites



 

Here is the measured response of the ES400 looking through the woofer filter and through the squawker channel with the transformer set to 6 dB.

The woofer measurements shows "gain" at 60 and 270 Hz. That's impossible of course. It's voltage gain because of the reactive load the woofer I'm using gives the filter at those frequencies. The woofer is an Eminance Kapp15C rather then a K33 though.   

I measured the DCR of the inductors in the woofer filter (General radio 1666 bridge):

Input inductor: .248 Ohms
Notch inductor: .243 Ohms
output inductor: .172 Ohms

Around the loop: .661 Ohms.

I looked at the SPICE files. I haven't looked at one of those in years! I'll have to do some deciphering to figure it out!

 

es400.gif

Link to comment
Share on other sites

I wasn't able to easily decipher your SPICE file so I used an old program I wrote to convert PCFILT ".dzn" file to spice files.  It's attached and renamed the ES4500 file to ".pdf" as you did. Let me know how it works out and compares with the values you came up with.  The article you took the values from is very old.

 

Al k.

 

ES4500.pdf

Link to comment
Share on other sites

  • 3 weeks later...

I contacted Al Klappenberger via email so I could send a copy of the Extreme Slope network schema that I had derived from his designs. Al kindly ran the network through his software and suggested 4 changes to components. I made these changes and here is a REW measurement at the speaker terminals before attenuation for the mid is applied. The woofer hatch was open for these measurements, hence the lumpy response.

If you are interested in the Extreme Slope networks read the paper entitled "A case for extreme-slope crossover networks" (http://www.alkeng.com/klipsch).

 

ES Outputs.jpg

  • Like 1
Link to comment
Share on other sites

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...
To Top