The Klipschorn Woofer Polarity Challenge

[wvu80]

17 minutes ago, garyrc said:

The window behind the curtains doesn't provide enough firmness (unless, maybe, it is very thick plate glass, and not too large), and there isn't supposed to be a gap between the Khorn and the wall(s).

You are correct, the window (left) is not strong at all.  There is 2 feet of wall then the large picture window but it's just not a good corner.  Here is a pic of my LS's so you can see L/R and the couches which are in the way.

 

My wife thinks there's too many speakers.  I think there's too much furniture. 

 

 

[John Warren]

9 hours ago, Wirrunna said:

wvu80, here are some Cliff's notes on my earlier graphs.
First, the purpose of the graphs was to show that an Al Klappenberger (ALK) Extreme Slope (ES) network all but eliminates the interaction between the woofer and squaker and therefore the polarity of the woofer becomes irrelevant.

  

 

A signal at 300Hz, within the range where the bass horn is operating, has even and odd harmonics that extend well into the into the midrange. 

 

Consider impulse type signals from a single source with sound energy over a wide range of frequencies.  The low stuff, say below 400Hz, and the high stuff, say above 400Hz, arrive at the listener at oddly different times.  It's the stuff you listen too that's within the range where the drivers are operating is where the medicine needs to be applied. 

 

 

[BeFuddledinMn]

In my experience, both listening to and measuring Klipschorns, both room and corner issues can be decisive. Long vs short wall positioning, total room length and width, ceiling height and furniture placemeant all come into play. Of course, sufficient corners with about 4'+ of clean wall space is very important. In my experience with the Klipschorn, real world measurements from listening distances in any room, will vary considerably from factory response measurements. 

 

Getting back to the Klipschorn woofer polarity puzzle, I think there have been a number of good points raised in the discussion, including bass horn length and phasing, bass horn behavior across the woofer range with peak horn efficiency, and the ability of crossover slopes to mitigate the corresponding issues in the midrange transition. However, I believe the woofer polarity challenge is far from resolved. Further, dozens of measurements with several,woofer and midrange horn combinations appear to confirm that there is a lot going on with this in the Klipschorn. I don't have the time to post all of the measurements and what I've learned, but it's clear from testing that woofer polarity does impact performance - both measured and heard ( particularly with material that emphasizes the affected range of ~250hz- 500hz) and, the optimal polarity alignment is dependent upon several factors, including crossover point and midrange horn used, among others. Both the degree and location of "destructive interference" between the woofer and midrange is in play, along with the correspondent degree and location of measured distortion. As someone said, there's a lot going on here. Moreover, understanding this challenge should, in my opinion, be of interest to anyone interested in "upgrading" or optimizing Klipschorn performance. 

 

I had a chance to pose the issue to Jim Hunter and Roy Delgado in a recent historian thread " Klipsch Licensed Enclosures" 

 

Here's a copy of my comment there on April 29th:

 

" Great stuff! One of the most interesting things about BOTH the Patrician and the Georgian is the fact that EV specified and used a reversed woofer polarity arrangement. A few years ago, I read an EV manual description of exactly why this was done - to align phasing in a Klipsch-based design where the folded horn travel length created phasing challenges when integrating with the midrange horn. The EV Patrician assembly documents that WMcD has once again so kindly posted, point that out as well, although they don't go into detail. I've posted a thread on the Klipschorn Woofer polarity issue, but this Patrician discussion should be very instructive IMHO.

 

We know that Klipsch didn't observe any polarity specification at all on the Klipschorn until 1958, and then instituted a matched polarity alignment between all drivers. Interestingly, the copyright dates on the EV documents WMcD has provided, demonstrate that EV was well aware of, and specifying the need for phasing alignment through REVERSED woofer polarity with the Klipsch-based design - at least as early as 1955! More still, the documents I read years ago described a very audible difference to EV engineers with particular reference to a real deficiency in a key octave range if the polarity / phasing issue were not addressed properly. I will look through all of my old documents to find that discussion. 

 

Meanwhile I'm pondering a few thoughts/questions:

 

1) Did PWK really not hear a difference or believe that the polarity/phasing issue mattered until 1958? 

 

2) Why, when the Klipschorn driver polarity did become specified in 1958, was the matched polarity chosen as opposed to the EV perspective/testing?

 

3) Is it possible that EV engineers knew something about the Klipschorn that PWK didn't, ( hard to accept ) or that PWK simply didn't want to address it? (EV designed and built the EV 15wk woofer used in the Klipschorn and Georgian, so presumably understood that better) (May reflect school of perfection vs school of "good enough")

 

4)I wonder which way Vitavox, the English Cadillac of Klipschorn copies, went with woofer polarity?

 

My own testing suggests there is a lot going on here, but I'm reserving judgement.

 

Does Jim Hunter or Roy Delgado have some perspective they'd like to share on this? "

 

-BeFuddledInMn

 

 

 

 

 

 

 

[BeFuddledinMn]

It took some digging, but I've found an old copy of the EV instruction manual from 1955 - on building the Georgian - the EV version of the Klipschorn, with specific instruction on Woofer polarity alignment below:

 

 

This corresponds to what I had read years ago in another EV report that described the issue in more detail. The Georgian crossover point from woofer to midrange was 300hz, not the 500hz of the Klipschorn at that time, but the EV engineers felt pretty certain that the woofer polarity alignment 1) made an audible difference and, 2) had a definite perspective on which choice was optimal, and why. As I recall, the x336 crossover EV used here, incorporated a 2nd order high pass filter on the midrange as well. In any event, the phasing and transition challenge presented by the Klipschorn bass horn length should be the same. No?

 

-BeFuddledInMn

[WMcD]

Sorry this is long.  I didn't have time to make it short.

 

Sorry  I’ve been away.  There have been a lot of distractions at work and home.

 

It looks like BFM might have been asking for permission to contact me but I took it as just polite phrasing.  Of course I’m always ready.  Though I think open communication on the site is best.

 

On the subject of testing. I’ll just say that it is difficult to find what is accurate. If you put the microphone a meter on axis in front the mid, you’re offset from the woofer.  Maybe the microphone should be put on the floor for testing the woofer and then you’re offset from the mid.  With these variations in path length the dips at the crossover freq(s) is going to move around. 

 

Move three meters out in the room the paths from the mouth of the bass bin and the mouth of the mid are pretty much equal, but room reflections hide the dips.  (Which might be why PWK was not overly concerned.)  Overall, it is tough to find a place for the microphone which reveals everything you want to know.

 

In my view:  Perhaps we can agree that an issue is the phase of the test signal as it is presented at the mouth of the bass horn and the mid horn over, say, an octave around the crossover freq.  Assuming they are both sounding. 

 

If this is a polarity issue (reversing the feed lead) we may be lucky if the sound is 180 degrees out of phase at some freq, and then we correct by reversing polarity.

 

However we can’t generalize between the K-Horn and the Georgian.

 

Essentially: Delay from length causes phase changes. Linear phase change with freq.

 

Also, any roll off causes phase changes. The roll off may be caused by the electrical filter, the mechanics of the driver, or even the response of the horn.  IIRC this is described though a Hilbert transform.  PWK mentioned it a little bit in an article.

 

Going back to BFM’s analysis.  It appears to me that you’re assuming that EV Georgian was correct (in polarity reversal) and PWK in the K-Horn was not correct (in no polarity reversal) . But we have to dig deeper.  The Georgian and the K-Horn are different in important ways. EV and PWK might well both correct for their respective and different designs.

 

As mentioned EV used  second order slopes in the crossovers.  PWK used first order slopes. That is going to cause a relative phase change in what the drivers are fed at the crossover point.  These have different phase changes at their respective, and different, cross over points.

 

Further, EV used its “reentrant” mid.  (I like to point out that this mid horn is shown in M*A*S*H.)  This is sort of what we see in bull-horns.  I’ll let you guys look that up.  But is a type of folding.  The driver is at the front, the sound goes to the back, makes a 180 degree turn, and comes out to the mouth at the front.  Effectively, the path is twice the overall length of the horn structure. Therefore, the EV mid horn is twice as long as it seems and has twice the delay of a similarly sized K-5 or K-400.   (Actually, bull-horn have a second fold which is why the driver is at the back.)

 

The bottom line:  If the output of the mid horn is 180 degrees out of phase with the woofer near the crossover and causes a dip, polarity reversal can cure it.  None the less, the situation is caused by phase issues caused by a number of factors which are unique to the crossover and length of the mid, plus the roll off of the horn and driver. This is relative to the bass horn, which has similar phase issues.       

 

 

[WMcD]

I should add:  The AK-4 crossover has much more more steep filters which probably address the issue of the bass and mid having overlapping output and phase issues.

 

My guess is that Roy used the Linear-X system of LMS and LEAP to design this. 

 

WMcD

[BeFuddledinMn]

WMcD,

 

I've always appreciated your contributions and this is no exception. (And I rarely have the time to get on here either) 

 

Based on my own experience I can't disagree with anything you've said. Measuring Klipschorn performance is a real headache in the home for all the reasons you've mentioned and then some, so after doing separate measurement positions per driver, I try to measure the whole package at the distance and level that I would typically listen - believing that this best reflects what a listener's ears will hear. Having said that, about a month ago I asked a very knowledgeable colleague who has 50 years of Klipsch experience and professional measurement equipment and measuring experience to investigate my observations with his own tests. He reported woofer polarity results similar to my own. 

 

Your point on differences between the Georgian and Klipschorn are valid of course and I was simply drawing attention to the common challenge with folded bass horns - even though, as you correctly suggest in my opinion, there may be different solution paths with each. Admittedly, I may have leaned a bit into the wind here to stir the discussion, but my aim here has been to get all the knowledge and perspective we can out on the table on an issue that seems seldom noted and even less understood.

 

More importantly, I believe you've identified and described one of the key variables in determining the "correct" polarity choice - the midrange horn used - and the point at which it's crossed. In fact your discussion is confirmed by the testing I've done with three different midrange horns in combination with three different woofers and two different motorboard slots, etc. In short, my testing has shown that the woofer polarity choice largely depends upon which midrange horn is used - followed by crossover point and motorboard slot. From what I've uncovered and measured, the optimal woofer polarity choice is not uniform but rather, "customized." Conversely, depending upon the woofer/midrange mix used in the Klipschorn, aligned or reversed woofer polarity appears to matter.  My best "hunch" at this point is that this explains the different polarity choice between PWK and EV. (I'm still a little puzzled by the fact that PWK apparently sent Klipschorns out the door until 1958 with random driver polarity wiring - given what contemporary measurements and listening tests reveal)

 

On a practical level, given the amount of effort and interest in "upgrading" the Klipschorn - particularly the replacement of the K400/401 with one of the tractrix midrange horns, I would have to believe this discussion is all the more worthwhile.

 

There is a lot more to cover here, but I wanted to get back to you quickly in the time that I have. 

 

-BeFuddledInMn

 

 

 

 

 

 

 

[garyrc]

Measured in my room (with REW & a calibrated mic) at the Main Listening Position, which is about 14 feet away and at ear height, my Left AK4 Klipschorn (didn't measure the right, but the room is symmetrical) has about a 5 dB dip centered at 350 Hz, compared to the nearby neighbors of 200 Hz and 500 Hz.    With Audyssey Flat room correction, the 350 dip disappears, so I guess this means it is not a room null, or Audyssey wouldn't be able to fix it (?).

 

 

[WMcD]

Gary had me going – good thoughts by him re Audassity.  After some noodling:

 

The criticism of room correction goes like this.  We have a dip caused by a duplicate out of phase source from a room reflection (or another driver).   If we goose up amp power in the narrow frequency response area of the measured dip, we’re still screwed.  This is because we’re increasing the level of the duplicate source which is causing the destructive interference too.   Call it a stubborn magic hole.

 

However, the dip is localized and in the fr curve  (importantly) we have regions of lower and higher freqs which are not affected by two interfering  sources (because they are simply  out of their range and they don’t overlap).  They have magnitudes which are higher than the dip.  But with enough signal processing we can affect them without affecting the dip.

 

Sooooo. My theory is that we don’t try to lift the dip (because that doesn’t work).  Rather, we lower what is above and below the dip in fr. 

 

In a thought experiment, we can’t fill the stubborn  magic hole in the soil, but we can  dig out the surrounding non-magic area in the soil to make the surrounding ground, level . . .  down to the depth of the magic hole which we can’t fill in.

 

Granted, now we’ve got sort of a level and  flat -bottomed hole in the ground.  We can go back and just lift up all of it.

 

I have no idea whether this is what the processor is doing.  But it makes some sense, maybe.

[garyrc]

4 hours ago, WMcD said:

Gary had me going – good thoughts by him re Audassity.  After some noodling:

 

The criticism of room correction goes like this.  We have a dip caused by a duplicate out of phase source from a room reflection (or another driver).   If we goose up amp power in the narrow frequency response area of the measured dip, we’re still screwed.  This is because we’re increasing the level of the duplicate source which is causing the destructive interference too.   Call it a stubborn magic hole.

 

However, the dip is localized and in the fr curve  (importantly) we have regions of lower and higher freqs which are not affected by two interfering  sources (because they are simply  out of their range and they don’t overlap).  They have magnitudes which are higher than the dip.  But with enough signal processing we can affect them without affecting the dip.

 

Sooooo. My theory is that we don’t try to lift the dip (because that doesn’t work).  Rather, we lower what is above and below the dip in fr. 

 

In a thought experiment, we can’t fill the stubborn  magic hole in the soil, but we can  dig out the surrounding non-magic area in the soil to make the surrounding ground, level . . .  down to the depth of the magic hole which we can’t fill in.

 

Granted, now we’ve got sort of a level and  flat -bottomed hole in the ground.  We can go back and just lift up all of it.

 

I have no idea whether this is what the processor is doing.  But it makes some sense, maybe.

 

It's conceivable that is what is happening.  I wish I could post the two curves, but they are on a crashed computer.  But the Audyssey Flat curve was elevated by about 5 dB over the No Audyssey curve -- corresponding, perhaps to your final step of lifting up all of it.  The frequencies below the dip were already raised up.  The newly elevated part of the curve (after Audyssey) is pretty flat up to about 12K, where it begins to fall (my MLP, where I measured with REW, is slightly off-axis ... the path from the Khorn tweeters cross in front of the listeners).  Audyssey also cut down peaks between 100 and 200, and at 8K..

[BeFuddledinMn]

It's the middle of summer, but I thought I'd throw a couple more thoughts on this topic:

 

Given what appears to be an organic challenge of the Klipschorn design, it would be interesting to hear from Roy Delgado on the balancing act between folded horn woofer polarity, midrange horn type, crossover point and room factors (vs. factory anechoic chamber).

 

In practical terms, as long as I can remember, the biggest knock on the Klipschorn after effective corners, was that in many rooms, the Klipschorn just didn't "work." Of course, the reasons are many and some even amusing, but optimization of the Klipschorn "in the field", is nonetheless a perennial topic and the expectations of someone who drops $15k on a new set of speakers are understandably high.

 

PWK never allowed for user controls in his speakers because he (correctly in my opinion), knew that until recently, it would be nearly impossible for most consumers to objectively measure/and or evaluate user settings. However, with the hardware/software tools now readily available for in home sound customization, it would be interesting to ponder the usefulness and viability of a future Klipschorn feature that offered a Klipsch designed hardware/software based optimization kit in conjunction with an advanced, variable setting crossover - providing a direct Klipsch based hardware approach to user optimization, vs. say an Audessy based approach. Risks? Efficacy? Cost? 

 

Of course, that might be considered a new version of heresy and it will inevitably beg the question as to whether this is a problem in search of a solution, or a solution in search of a problem.

 

-BefuddledInMn

[garyrc]

As I said earlier, my Klipschorn (I only tested the left one) has a dip at 350, rather than the 450 crossover point (it's the AK4 model), without Audyssey (Audyssey fixes that).  Could this be the 450 crossover dip being somehow shifted down to 350?   Does that kind of thing happen?  The room has good corners, the Khorns are pushed in to the corners against the recommended gasket material.  Length: 25 feet, width 16.75 feet, ceiling is sloped up from 8.5 feet in the front (where the Khorns are) to 11.84 feet in the back.  The volume is 4,195 cu ft.

 

I'm feeling that the next big improvements need to occur in recording practices, rather than in speakers.  Hi End speakers are available at up to more than a million dollars a pair, and every time I hear one of the more reasonable ones (hard to find), they sound to me like they have veiled the sound to prettyfie it.  Given a good recording, the Khorns sound more like the live sound of acoustically played music, IMO.  I haven't heard the Jubilees.

 

Audyssey seems to be slowly improving over the years, now has an app to customize the curve a bit, and the expensive alternative (Dirac (?) or some such) is reputed to be even better.  You raise an interesting question, BeFuddledinMn.  I think any Klipsch based hardware/softwear approach would have to measure and optimize the room, as well as the speakers, in both the frequency and time domains to improve on current room optimizers.  Some advocate manually equalizing with REW.

 

Last time I talked to Klipsch customer service, Steve Phillips hinted that the official position of Klipsch was that the Klipschorn was flat enough, and that fooling around with it with EQ optimizers could sometimes do more harm to the sound than help it.  I remember Roy Delgado, back in the late 1980s, telling me, "Mr. K doesn't like equalizers."  But the technology has changed a lot since then, and the Khorn has also changed.  I'm no expert, but here is what I think:     

• My Khorns sound fine without Audyssey, and even better, markedly, with Audyssey (speakers this efficient need a work around for Audyssey to turn them down enough without maxing out -- "mining out" -- at -12 dB on most preamp/processors or AVRs.  This doesn't affect the EQ/time domain compensation, just the overall levels per speaker, and is easily remedied with a test disk in the player, but never with the built-in pink noise in preamp/processors or AVRs, because the noise does not go through the finished Audyssey EQ, and therefore often produces incorrect results.).
• Bi-amping or Tri-amping with the right equipment can help address the internal time misalignment of a Khorn, which Audyssey can't touch.  Chris A and others have written about this on the forum, and I think Roy Delgado has addressed it in regard to the Jubilee.  To me, the Khorn sounds excellent without it, but, if I later Bi/Tri amp I may see an improvement.
• Variation in recording techniques, the "loudness wars," and screwing up recordings (at least pop/rock/alt/metal, less so for classical, jazz, and DTS HD Master or Dolby Tru-HD Blu-ray movie soundtracks) make any Khorn phase/polarity problems seem minor.  Chris A has written about re-mastering solutions on the forum.  Speaking of phase, one corporation offered a device in the '70s and '80s that would deliberately throw the sound out of phase, thinking that the result would be an improvement.  I think they called it the "Aural Exciter," or something like that. Do I remember this correctly?  I'm not sure.  Meanwhile, a studio in Oakland received shipment of a set of Altec 604Es that were accidentally wired out of phase.  Most of the recording engineers liked them better than the old, correctly wired ones, and, I guess, wondered how Altec improved them.  One engineer, who didn't like them, conceded that they were, "Close enough for Rock and Roll." 
• One persistent conundrum in room EQ is whether to optimize for one listening position or for several.  Even without any EQ, if you move your head about a foot or two, the sound changes.  Yes, the brain will partly compensate for this, thank goodness.  Someone sitting next to you and you will get different sound, and one position may measure better, or be preferred, and these two are not necessarily the same.   With only one microphone location, placed at the Main Listening Position, this is still true.  Audyssey XT and XT32 use 8 mic positions and use a "fuzzy logic" assessment, rather than an average, to get around this.  But those who usually listen alone are advised to locate all 8 mic positions around the location of their (one) head.  
• The other persistent problem is whether to try to get the frequency response flat.  Holt, the founder of Stereophile, wrote an article called "Down with Flat."  His stance is obvious.  Audyssey provides a "Flat" option, and a "Reference" option.  Reference provides a gentle roll off in the extreme highs, -2 dB at 10K and -6 dB at 20K.  They claim this is the best curve for most living rooms and HTs.  This curve also provides a slight dip at just about 2K and a little above.  The CTO and founder says he has never heard a speaker that was not improved by the addition this dip.  With a good, clean, low distortion recording, I tend to prefer Audyssey Flat, but with a recording with garbage at the top, I prefer Audyssey Reference, to get "garbage in, clean sound out."   The "Harmon curve" basically is about 10 dB higher at the bass end than at the treble end.  The notorious X curve in movie theaters uses a more extreme treble roll off, that is a disaster in a well treated room, and, after 40 plus years, is falling out of favor in the cinemas themselves (Google it).   

We all try for the sound we like, and if altering the internal phase anomalies that might be in the Khorn helps us achieve this, so be it.  If all else fails, there is alchemy. 

[John Warren]

On ‎7‎/‎9‎/‎2017 at 8:13 PM, garyrc said:

Steve Phillips hinted that the official position of Klipsch was that the Klipschorn was flat enough

 

The AK-4 and -5 nets implement the notch filter in the bass horn to reduce the peaking between 100-300Hz.  It's a big change given 50 years of manufacture without it.  Operating bass horns on band-pass filters is another improvement.  They sound better operating where the horn acts as a horn.  The low stuff should be handled by a sub.