Crossover puzzle

[pauln]

I have been pondering this for a while. There must be something I am missing or misunderstanding. Any thoughts appreciated.

It takes a little background to explain the puzzle so I will just lay it out as best I can.

Starting facts:

1] The old La Scalas (like Khorns) were rated at 104dB/W/m with the old networks up through AL-3 for the LS and AK-3 for the Khorn.

2] The last La Scalas (like Khorns) from 2005 were rated at 105dB/W/m with the AL-4 and AK-4 networks.

3] The latest versions of the squawker and tweeter (K-55-X and K-77-F) are about 1.5dB less efficient than the older ones.

Assumptions:

1] In all the big Heritage the woofers are unpadded and the squawker and tweeter are padded down to balance the woofer.

2] Since the woofer efficiency varies in the Heritage models, the padding of the high end is different in the models; least for the Khorn.

3] In order to increase from 104 to 105 with the new AX-4 networks, the padding on the high end must have been reduced about 2.5dB to get a compensatory gain in the high end of 2.5dB to continue to match the woofer with the quieter squawker and tweeter.

4] The old type A network is designed to anticipate use in the old style models utilising the original greater padding of the high end with the louder squawkers and tweeters.

Expectations:

Placing type A networks into the latest LS using the K-55-X and K-77-F should result in a relative decrease of 1.5dB in the high end.

Outcome:

When I put the type A into one of my LS (other one still had the AL-4) I immediately not only noticed it was clearer and more beautiful, but it was quite noticibly louder. Not just a just noticable difference but a strong noticable difference.

Puzzle:

If my LS were 105dB with AL-4, what happened when I put in the type A? Did the LS become more efficient? They certainly became louder.

Am I mistaken that the new K-55-X and K-77-F are 1.5dB quieter? Could it be that they are louder? I am sure I read that they are quieter and one of the features of the new AX-4 series networks was to compensate for this with less padding.

Any clarification on this is appreciated. I have to be missing something - it is hard to imagine my LS have increased in efficiency with the installation of type A's, but I know what I heard when I compared them to the AL-4; they got louder.

[rcarlton]

No real puzzle. Measuring the efficiency of speakers had changed over time. Klipsch just recently remeasured the Heritage series and found the LaScala II was 0.5 dB more efficient than the K-horn. On the spec sheets they upped the efficiency rating for both to 105 dB. They (K-horns and LaScala's) probably were that efficient for years. In other words a K-horn with an AK-2, AK-3 or AK-4 probably was at 105 dB/W/m. This information came from the 2006 Klipsch Pilgrimage. I can get you exact quotes if you would like.

[meagain]

When I put the type A into one of my LS (other one still had the AL-4) I immediately not only noticed it was clearer and more beautiful, but it was quite noticibly louder. Not just a just noticable difference but a strong noticable difference.

Pauln - I'm a bit surprised you like the A over the AL-4, but I've never heard an A so.....

I preface this because it's coming from "me", but I had the same result going from an AA type to the AK-3. I had to turn the AK-3's volume up a notch or 2 to equal the volume I was used to with the AA. I think what you're hearing volume-wise is the difference in the 'type' of network.

The AL-4 (or AL-3) is a steeper slope than the A and I think that's what's resulting in your volume discrepancy. Someone correct me if I'm wrong.

[Deang]

It's all smoke and mirrors.

[pauln]

Someone must have an explanation.

It would be great if someone could just explain the increase in efficiency of the A's over the AL-4's.

A just noticable difference would be 3dB. 108db LS would be great with my SETs, but is this what happened?

[BEC]

I need to get around to some testing on the AL-4.  I do have a pair of them and a pair of AK-4s.  But, it makes sense to me that the AL-4 network is likely not as efficient as the Type A or other networks.  The Type AL-4 does away with the autotransformer for sensitivity matching of the drivers.  The autotransformer should have almost no loss in the way it attenuates.  The AL-4 goes to resistors to match up the driver sensitivities.  In other words it achieves the desired attenuation by dissipating power as heat in the crossover.  

 As to the K-55X and K-77F drivers, they are not more or less sensitive than the K-55V or the K-77M.  The K-55X and the K-55V (before 1980) are the same driver, the Atlas PD-5VH.

 The K-77F which is a clone of the K-77M is pretty close to the K-77M in performance.   

Bob Crites

 

[Erik Mandaville]

Bob:

"The Type AL-4 does away with the autotransformer for sensitivity matching of the drivers. The autotransformer should have almost no loss in the way it attenuates. The AL-4 goes to resistors to match up the driver sensitivities. In other words it achieves the desired attenuation by dissipating power as heat in the crossover. "

But do you think the fact that resistors are used rather than an autoformer would be responsible for a reduction in output? I agree that heat would be a byproduct, however it seems that the values of the fixed resistors would be what is responsible for the attenuation -- and not simply the fact that resistors are used in place of an autoformer. In other words, and regardless of how much heat is produced, it would be possible to select values of resistance that would provide exactly the same amount of attenuation as a given tap on an autoformer.

The reduction in output may be the result of a subjective and collectively-agreed-upon voicing of the speaker using the crossover/drivers/enclosure design in question, and/or I think Lisa may be absolutely correct about insertion losses related to a more complex, higher order network.

Just my view,

Erik

[Marvel]

My guess is that Paul doesn't play music nearly as loud as Lisa, so the lower order crossover will work better for him, whereas Lis likes the higher order. Most have agreed that the higher order crossovers need more power, need to be played louder, for them to sound better and open up.

That's why the first order DHA2 crossovers sound so good to me with my 2A3 amps. I don't play music loudly.

Bruce

[Erik Mandaville]

As a sort of inversely related 'aside,' I had a pair of Wright 2A3 monoblocks here that needed repair, and on inspection found, among some other problems, that the output transformer secondary on one amp was set for 4 ohms, where the same on the other channel amplifier was on the 8ohm tap. The difference in output between the two was obvious, and the resulting overall sound using the pair that way......just kind of strange.

All of the 2A3/300B/45 single ended triode amps I've used with La Scalas and Klipschorns preferred the most simple of dividing networks. Actually, the amps didn't *prefer* anything. It was my personal preference, only.

Erik

[BEC]

Bob:

"The Type AL-4 does away with the autotransformer for sensitivity matching of the drivers.  The autotransformer should have almost no loss in the way it attenuates.  The AL-4 goes to resistors to match up the driver sensitivities.  In other words it achieves the desired attenuation by dissipating power as heat in the crossover. "

But do you think the fact that resistors are used rather than an autoformer would be responsible for a reduction in output?  I agree that heat would be a byproduct, however it seems that the values of the fixed resistors would be what is responsible for the attenuation -- and not simply the fact that resistors are used in place of an autoformer.  In other words, and regardless of how much heat is produced, it would be possible to select values of resistance that would provide exactly the same amount of attenuation as a given tap on an autoformer.

The reduction in output may be the result of a subjective and collectively-agreed-upon voicing of the speaker using the crossover/drivers/enclosure design in question, and/or I think Lisa may be absolutely correct about insertion losses related to a more complex, higher order network.

Just my view,

Erik

 Erik,

When you attenuate with an autotransformer, the power produced by the amp stays the same for a given acoustic output of the driver.  Say you want 1 watt into the driver, then only one watt is produced by the amp.

With a resistor attenuation, say you want one 1 watt into the driver but you want the driver attenuated 3 db in relation to the woofer.  You could use a series resistor equivalent to the impedance of the driver.  Then the amp would product 2 watts into the resistor/driver combination and each component (resistor and driver) dissipates  half the power.  The one watt dissipated becomes heat and the 1 watt going to the driver becomes sound.

So, with an autotransformer, the excess power never leaves the amp (or we could say the excess power is never produced by the amp).  With resistor attenuation, the excess power is produced by the amp and wasted as heat in the crossover.

 Bob

 

[NOSValves]

Pauln - I'm a bit surprised you like the A over the AL-4, but I've never heard an A so.....

His system is very different from yours.......... I bet with his upstream components the type A does sound better.

[NOSValves]

It's all smoke and mirrors.

Kinda like those Jubs Heh............

[NOSValves]

In reality I bet if they retested an original Heritage Lascala or Khorn with a fresh set of Type A crossover by todays standards and procedures the original Heritage would be higher then the modern Heritage. The testing method and equipment has changed so the numbers do not correlate. Kind of like the horsepower ratings change back in the early 70's for the automotive industry.

[Deang]

"When you attenuate with an autotransformer, the power produced by the amp stays the same for a given acoustic output of the driver."

I'm probably misunderstanding the above statement, because I know you know this. Power is cut in half for every 3dB you attenuate. When you drop 3dB, the impedance that the amplifier "sees" doubles, which means that while the "8 ohm" unattenuated drivers are getting a watt, the attenuated driver is only getting a half watt to do the same amount of work.

With that in mind, how do the speakers with the resistors in the networks end up with the same or better sensitivity rating. And how does a 20+ element filter end up with less insertion loss as one with four parts total? There's no free lunch, you either the raise efficiency of the system, change the way you measure, or change the way you interpret the data.

[Deang]

It's all smoke and mirrors.

Kinda like those Jubs Heh............

lol, yeah, just like that.

I think what you said in that last post is true.

[BEC]

"When you attenuate with an autotransformer, the power produced by the amp stays the same for a given acoustic output of the driver."

I'm probably misunderstanding the above statement, because I know you know this. Power is cut in half for every 3dB you attenuate. When you drop 3dB, the impedance that the amplifier "sees" doubles, which means that while the "8 ohm" unattenuated drivers are getting a watt, the attenuated driver is only getting a half watt to do the same amount of work.

With that in mind, how do the speakers with the resistors in the networks end up with the same or better sensitivity rating. And how does a 20+ element filter end up with less insertion loss as one with four parts total? There's no free lunch, you either the raise efficiency of the system, change the way you measure, or change the way you interpret the data.

 

Think back to the way I said it.  A watt into the driver gives a certain acoustic output.  Now what power does the amp have to output to give that acoustic output? 

The answer with an autotransformer attenuating 3 db is still one watt. (ignoring for the moment very small losses of the circuit).

The answer for the resistor attenuating the same 3 db is 2 watts.

Bob 

[Erik Mandaville]

Thus, the new generation Klipschorn and La Scalas are capable of 105dB/1watt (but ONLY if one's amps are capable of at least 2 watts). Sorry single-ended 45 and SE OTL guys!

Erik

[garyrc]

Pauln,

A just noticeable difference would be 3dB. 108dB LS would be great with my SETs, but is this what happened?

The magnitude in dB of a "just noticeable difference" may depend on whether one is listening to audio oscillator tones or a complex stimulus like music. Back in college, everyone in our Discover Your Ears class (about 20 students) could clearly discern 1 dB differences in complex music with a lot of high and mid frequency content (when switched between two electrical channels, which either did -- or did not -- incorporate a dB difference, single blind), but it took more like 2 + dB (depending on the student) to hear differences in a 300 kHz tone. We expected a 1K Hz tone to be easier than the 300., but I forget whether it was for most of the students.

[WMcD]

There are a lot of things going on with testing conditions and writing specs. It is not really a matter of being dishonest or better efficency overall. It is just very difficult to deal with the peaks and dips in the best of conditions.

- - - -

I built something like a K-400 from plywood, but with flat pieces on the top and bottom. This was similar in configuration to the trac horns people make or what Bruce Edgar did with his SpeakerBuilder midrange. I used an EV driver.

The best test of the unit was outside, with the horn on a ladder, about 6 feet up and with the mike 6 feet up, and the LMS gated to ignore the bottom bounce. It fit in a plus and minus 3 dB window from 400 to 6000, no smoothing . . . but with one little data point in a dip which was 1 dB outside the window of 6 dB. Drat.

Let us suppose, for argument, there were one or more dips which made the peaks and troughs fit in an 8 dB window. I'd have to say the response was, 107 dB, plus and minus 4 dB. The 107 would be the center and I call that the efficency, just because it was the middle of the peaks and dips.

But, if that dip went away, because of testing conditons, or anything else? Then the midpoint (which we call efficency) would go up by a dB or more. It is not dragged down by the few anomalies at the, dip, low points.

Nothing in the driver changed.

= = = =

It could be the same in redesigning a crossover in a test of the entire system of three drivers. It could be there is a dip caused by the mid to tweet crossover point, or mid to woofer. If you redesign the crossover, and get rid of that, then, per the "pick the middle of the peak and dip for efficency" then the overall efficency spec goes up.

Again, nothing has changed except a more favorable set of testing conditions.

- - - -

Gil

[Guest " "]

"Any clarification on this is appreciated. I have to be missing something - it is hard to imagine my LS have increased in efficiency with the installation of type A's, but I know what I heard when I compared them to the AL-4; they got louder."

its simply a function of the orders of the xover.  

Gentle sloping xovers have less insertion loss than higher sloping xovers

Gentle sloping xovers offer greater overlap in xover points

Gentle sloping xovers offer less driver protection

etc