Jump to content

Why is the ALK Universal network "universal"


Recommended Posts


These days it seems that everybody is experimenting with upgraded midrange (squawker) horns and fancy drivers to replace the K55 / Atlas PD-5VH. I can't complain since my Trachorn is part of the reason people ate thinking about it! Anyhow, I keep getting questions about moving the 400 Hz crossover frequency of my Universal network up to 500 Hz to accommodate better drivers like the Altec 902 or the JBL 2426h I use in my Belles. I hope to illustrate why this is not feasible and also explain why my "Universal" network is actually "universal". The reason for both is the same.

Here is the plot of the woofer and squawker channels of my Universal network on a log scale of 20 to 2000 Hz. Notice that the two plots cross at about 3.16 dB. This is my "definition" of the crossover frequency OF THE NETWORK. That is, the frequency point where the attenuation through the two channels is identical. With power equally divided between the two channels you will have half the total power in each. That is 3.01 dB attenuation. Mother nature always wants to take a little off the top. It's kind-of like the interest on your credit card! Anyhow, that happens because of resistive losses in the components. In this network it amounts to about 0.16 dB leaving us with 3.16 dB out each channel. That happens at the little black dot (marker) located at 377 Hz on the plots. That's a nominal 400 Hz crossover.

Note that this filter is a first order filter. 6 dB / Octave slope. To make the measurements I have terminated the midrange filter by connecting a 13 Ohm resistor between taps X and 4 on the network. This is where the K55 would be connected. The woofer is simulated by a 1 mHy inductor in series with a 6 Ohm resistor which is the complex impedance of the Klipsch K33 woofer driver. The measurements are taken across the 6 Ohm resistor and the total winding of the 3619 autotransformer. The 1 mHy inductor is part of the woofer filter in this test plot. More about that next!

post-2934-13819511525776_thumb.gif

Link to comment
Share on other sites

Here's another plot of the very same network but this time I have removed the 1 mHy inductor that represents the voice coil inductance of K33 woofer from the measurement. I did a long series of test a few years ago to determine if the voice coil inductance of the woofer driver is part of the filtering function or not. It turns out that is NOT! It does effect the impedance seen by the rest of the filter and ultimately by the amp though, so it needs to be accounted for. I normally synthesize this inductance as part of the filter. The other choice is to tune it out with a Zobel network. The two methods seem to be a wash as to which is the better way. In the Universal network, the entire filter is just a single inductor, so I just subtract the 1 mHy voice coil inductance from the 2.3 mHy inductance required for a 400 Hz crossover leaving 1.3 mHy. That's the value I use in the Universal network. This keeps the constant 6 Ohm resistive impedance when the 24+24 uF caps of the highpass channel cancel the inductance of the woofer filter plus the voice coil inductance. Nice, BUT, look what happens to the measured crossover frequency where the attenuation through each channel is equal (at the marker). It moves up to 488 Hz! So where is it? 488 Hz or 377 Hz? Notice too that the loss is now 2.05 dB rather then the 3.16 dB we had before. With less than half power (3 dB) going through each channel we seem to have more total power out of the network than we are putting in! NOT! What's going on is that the network is no longer presents a constant impedance to the amp. We have reactive "wattless" power involved. So the next question is if we don't know where the actual crossover is, how can we move it to, say, 500Hz and know it's at 500 Hz? Answer: WE CAN'T! We just have to make an assumption that the right frequency is with the voice coil inductance in the circuit providing constant impedance. With a higher order filter this is not a problem because all the extra elements overwhelm the effect of the single element represented by the woofer voice coil inductance making the crossover frequency stable and definable. In other words, this is only a major problem with a first order filter like in my Universal network.

post-2934-13819511526196_thumb.gif

Link to comment
Share on other sites


Ok, Now let's consider the change in crossover frequency from a nominal 400 Hz to 500 Hz. The idea being to allow the use of a nice wide-range driver without blowing it up. Let's also assume that we accept the idea that the voice coil inductance should be considered part of the network to define the crossover frequency.

I have marked up the first plot with two markers (red) that represent the attenuation through each channel at 500 Hz when the point of equal attenuation (the crossover frequency) is roughly 400 Hz. It comes down to 2.0 dB through the squawker channel and 4.4 dB through the woofer channel for a grand total of change of 1.1 dB difference through each channel when we change the measured crossover frequency. WHY BOTHER? What do you gain?

To address more directly, the purpose of doing the crossover frequency change is to protect the light metal diaphragm of the wide-range driver from the powerful low frequency energy getting through the squawker filter. Lets look at the attenuation at 100 Hz. With the crossover at 400 Hz the attenuation will be about 12 dB. If we move the crossover up by 100 Hz to 500 Hz. you will then have the same attenuation at 100 Hz that you had before at about 80 Hz. That's a big fat 1.9 dB more! Again: WHY BOTHER!

(Clarification:
Butterworth attenuation = 10 * Log10(1+w^(2*n))
w = stopband position n = number of elements (filter order = 1)


stopband position w = 400Hz / 100Hz = 4
10 * Log10(1+4^(2*1)) = 12.3 dB
stopband position w = 500Hz / 100Hz = 5
10 * Log10(1+5^(2*1)) = 14.2 dB

improvement = 14.2 - 12.3 = 1.9 dB)

SO: If you want to upgrade your tough-as-nails K55 driver to a better one you need to have a higher order woofer / squawker filter to protect it from the lows. Look at the specifications on most wider-range drivers and they will tell you the same thing. They will tell you that you can move the crossover lower with a higher order filter. Most do not recognize that filters beyond 12 or 18 dB / octave are available. This means that you can safely move a wide range driver much lower with a network having slopes as high as 120 db / Octave. Now, where on Earth can you get one of those, I wonder!

If you moved to a 2nd order filter (12 dB / octave) with a 400 Hz crossover you would have:
w = 400 / 100 = 4 n = 2
10 * Log10(1+4^(2*2)) = 24.1 dB at 100 Hz for an 11.8 dB improvement.

The final point: With the crossover frequency and attenuation slope of a first order filter, like in my Universal network, being so sloppy and undefinable, how can you say it must only be used in the Khorn with a 400 Hz crossover or a Belle or LaScala that can be crossed over at 500 Hz or even higher? This is why my Universal network, the Klipsch A and AA networks are all "universal". PWK put the AA in the Khorn, Belle and in the LaScala. The horns determine the acoustic crossover, not the network. The first order network is so sloppy that you don't even need an inductor in series with the woofer at all for it to function. The AB network used in the Belle for a while had none! PWK also left it out of some of the speakers he experimented with. The woofer voice coil inductance alone can serve to route the highs to the squawker. All a first-order network can do is allow the two drivers to operate smoothly across the acoustic transition. That is, to provide constant impedance to the amp.

If you want to take this to the next step, find the acoustic plots of the Khorn woofer and the K55 on any horn you like and plot it over the curves of the network. The plot of the network won't amount to much in comparison!

Al K.

post-2934-13819511526566_thumb.gif

Link to comment
Share on other sites

Hey Al I've been wanting to try your Universals on my Khorns. I've heard very good things about them.

I'm currently using these on my JubClones.

IMG_1165_2.JPG

I keep flip flopping on wether to go three way but I think I'd like to give a try at some point. When I do is it possible to get a different crossover point than what is currently posted if needed?

Link to comment
Share on other sites

Moving the crossover on any correctly designed crossover network means a total rebuild. All of the parts need to change value. You need to set it aside and start over! If the one in the picture is the one I think it is, you don't need to change it to go 3-way. Just feed the high output into a second crossvoer at 6000Hz or whereve.

Al K..

Link to comment
Share on other sites

I couldn't resist! Here's a computer plot of the lowpass channel of the ES400T network on the same frequency and attenuation scale as the measured plots of the Universal network. Attenuation at 100 Hz through the squawker channel is roughly 50 dB,

Al K.

Al,

Will the ES400 work with a 4 ohm woofer (and a 16 ohm midrange)? I have a pair of speakerlab bass bins that I think use a 4 ohm woofer and JBL 2470 16 ohm midrange drivers.

Does the ES400 need a 5800 hz crossover up top or does it not matter? I have a pair of JBL 2404's that I'd like to cross over lower, at 4500. I'd like to use the 4500 HZ from your universal as attached and add a pair of your tweeter attenuators.

post-10861-13819511550534_thumb.gif

Link to comment
Share on other sites

Al,

Will the ES400 work with a 4 ohm woofer (and a 16 ohm midrange)? I have a pair of speakerlab bass bins that I think use a 4 ohm woofer and JBL 2470 16 ohm midrange drivers.

The manufacturers of woofer drivers never give you the full info on their drivers. Juat beasue it says "4 Ohms" doesn't mean it is! In other words, I don't know becasue we really don't know what the true imedance really is. Chances are it's close enough to work ok but you can't be sure.

As to the midrange. Any of my designs will work with virtually any quality midrange driver of any impedance.

Does the ES400 need a 5800 hz crossover up top or does it not matter? I have a pair of JBL 2404's that I'd like to cross over lower, at 4500. I'd like to use the 4500 HZ from your universal as attached and add a pair of your tweeter attenuators.

As long as the blue 24 Ohm resistor is still on the ES400 it will match fine. If it's a DIY ES400 you should make sure it's there. If it's not there the match will not be quite as good but will still work

Link to comment
Share on other sites

Al,

Will the ES400 work with a 4 ohm woofer (and a 16 ohm midrange)? I have a pair of speakerlab bass bins that I think use a 4 ohm woofer and JBL 2470 16 ohm midrange drivers.

The manufacturers of woofer drivers never give you the full info on their drivers. Juat beasue it says "4 Ohms" doesn't mean it is! In other words, I don't know becasue we really don't know what the true imedance really is. Chances are it's close enough to work ok but you can't be sure.

AL,

I was wrong. I have the W1508S woofers that are rated at 8 ohms. Will these work with the ES400 or do all of the values need to be changed (and to what values)? I need to build the ES400 netowrks because I am using a JBL 2470 and want to be sure that it can handle the task.

Link to comment
Share on other sites

Al,

Will the ES400 work with a 4 ohm woofer (and a 16 ohm midrange)? I have a pair of speakerlab bass bins that I think use a 4 ohm woofer and JBL 2470 16 ohm midrange drivers.

The manufacturers of woofer drivers never give you the full info on their drivers. Juat beasue it says "4 Ohms" doesn't mean it is! In other words, I don't know becasue we really don't know what the true imedance really is. Chances are it's close enough to work ok but you can't be sure.

AL,

I was wrong. I have the W1508S woofers that are rated at 8 ohms. Will these work with the ES400 or do all of the values need to be changed (and to what values)? I need to build the ES400 netowrks because I am using a JBL 2470 and want to be sure that it can handle the task.

hey there....I have built and have used both the ES400T and ES500T networks along with the ES5800 with a few different woofers other then my Belles...the other cabinets are Altec Model 19 and Onken (which are about 17ft3)....

JBL 2205H, University C15W, Altec 416-8A and Altec 515B (16ohms) all with either 8 ohm mids like JBL 2470, Altec 292 and the AER, which are 16 ohm....all sound excellent....I don't think Big Al realize how "universal" all his networks are....I can't say I have measured anything but as AL also says "whatever sounds right is right"....or something like that.....Al - thanks again for making your designs public....

Link to comment
Share on other sites

I did a few response plots of one of the low ES networks with several different complex loads simulating a wide range of drivers. There is a difference in smoothness as you move away from the 6 Ohms + 1 mHy that represents the K33, but it's not a lot. Considering the respones when measure in the acoustic output is slightly different anyhow, using unknown drivers isn't a big problem.

Al K.

Link to comment
Share on other sites

AW,

So have you tried the ES400 with an 8 ohm woofer? If the design was based on 6 ohms, and ignoring teh other complex impedance factors, teh 8 ohm woofer shoudl work as well as the 4 ohm version. I will build the ES400 because worst case scenario I will have to replace the woofers. I just hate to because the Speakerlab is know for having more bass than a K Horn.

Link to comment
Share on other sites

Guys,

I just did some computer simulations on the ES400 woofer filter using various complex loads to simulate different woofers. It seems to tollerate between 4 and 8 Ohms and a voice coil inductance between .5 and 1.5 mHy. If you go out of that range the passband gets pretty rough. A 16 Ohm load looked very band. From the quick simulations I did it looks like the ES400 load should be 4-8 Ohms in sereis with .5 - 1.5 mHy. I wouldn't know what drivers that represents without testing each driver. I know it rules out the K24, but that's a 12 inch driver from the Heresy II so I don't think anybody is going to be trying to use that one!

AL K.

Link to comment
Share on other sites

AW,

So have you tried the ES400 with an 8 ohm woofer? If the design was based on 6 ohms, and ignoring teh other complex impedance factors, teh 8 ohm woofer shoudl work as well as the 4 ohm version. I will build the ES400 because worst case scenario I will have to replace the woofers. I just hate to because the Speakerlab is know for having more bass than a K Horn.

I am using the ES400T and ES 5800 now with the University CW15 woofers set to 8 ohm (this woofer can also be jumpered to 16 ohm) in my Onken .... they sound excellent and I just had my Altec 515B in the cabinets and they sounded excellent....you will be fine....build away....

Link to comment
Share on other sites

Guys,

I just did some computer simulations on the ES400 woofer filter using various complex loads to simulate different woofers. It seems to tollerate between 4 and 8 Ohms and a voice coil inductance between .5 and 1.5 mHy. If you go out of that range the passband gets pretty rough. A 16 Ohm load looked very band. From the quick simulations I did it looks like the ES400 load should be 4-8 Ohms in sereis with .5 - 1.5 mHy. I wouldn't know what drivers that represents without testing each driver. I know it rules out the K24, but that's a 12 inch driver from the Heresy II so I don't think anybody is going to be trying to use that one!

AL K.

Hi Al....I realize the sim you did indicates poor results for a 16ohm woofer ....as I said I can't measure stuff but the 515B's sounded very good.... here are some specs I could find...not sure if this helps with any further simulations....thanks -Al

Thiele-Small parameters
Model Number: Xmax(inch) Re (ohms) Vd(cu. in.) Fs(Hz) Vas(cu. ft.) Ref(%) Qts Qms Qes Vid
515B 0.18 12.00 23.00 24.70 19.70 4.60 0.17 7.50 0.17 0.22
Link to comment
Share on other sites

Guys,

OK, now you did it! You got me going! I happen to have an ES400T network sitting here for a local customer that I am going to install for him in a week or two. I did some plots with various loads on the woofer filter. They are all assuming a 1 mHy voice coil inductance. It looks like a 16 Ohm load is OUT! 4 -8 Ohms looks ok.

Al K.

post-2934-13819511570062_thumb.gif

  • Like 1
Link to comment
Share on other sites

I'm the local customer Al referred to in the previous post. He's scheduled to come over October 31 to install his extreme-slope networks and Trachorns in my 2003 Klipschorns. When he does I will take and post a few pictures in a new string. He lent me a pair of Trachorns for a week a while ago that I screwed the Khorn squawker drivers onto and sat them atop the speakers for a good trial listen (see picture). The improvement with the Trachorns alone was very noticeable. That led me to save up for the whole upgrade. With the Trachorns + ES crossovers the improvement should be even more, though Al has cautioned me that it won't be as dramatically more better (sic - my words not Al's) than it was with the Trachorns alone.

With the upgrades I imagine I will have a pair of what will truly be the "world's best loudspeakers" or among the very best. Quite a dream come true for this 67 y.o. who got started in this audiophile hobby at the age of around 12 building Heathkits and homemade speaker enclosures as a little boy in Boston in the 1950's!

post-10590-13819511834382_thumb.jpg

  • Like 1
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...