Didn't PWK write a paper about driver alignment not being that important

[jwgorman]

Perhaps this is merely confusion on my part, but I seem to recall that PWK wrote a paper about separating the squawkers and tweeters and they had to be pretty far away from one another before people (who could not see the speakers) noticed.  

 

I removed my Crites tweets from the normal mounting position in my Khorns yesterday, sat them on top of the cabinet and pushed them back to where the tweet/squawker drivers lined up. I mounted the tweets up on cigar boxes and covered the Khorn tops with towels to absorb any early reflections. From my listening position I could see the bottom of the tweet. I spun a few familiar albums and I think is was a bit better with a couple of things, like, locating an acoustic guitar...and it did seem like the tonal balance shifted a bit (which is what a lot of people who'd time aligned their Khorns have reported.) 

 

I realize with the phase shift of the balancing network, there are still issues with this approach, but it was interesting as a free experiment.  Anyway, does anyone else have any recollection of this paper that PWK wrote? I could not find it via searching. 

[Deang]

I don't know about "a paper", but everyone seems to know about the demo he did. 

 

This is a good thread. Enjoy. 

 

 

[Chris A]

See http://www.aes.org/e-lib/browse.cfm?elib=2039

 

This is one area where it is clear to me that PWK's views changed over time.  Klipsch has published delay settings for its professional/cinema loudspeakers for many years.

 

1 hour ago, jwgorman said:

I realize with the phase shift of the balancing network...

 

One of the aspects of crossover filters is that there is a 90 degree phase lag of the LF section per "order" of the crossover filter used (Butterworth).  So for higher order crossover filters, you must delay the HF section by the correct amount...or set the drivers farther back than the midrange driver.

 

The way to set the delay correctly is by measuring it using something like REW or TrueRTA.  Setting the delay wrong is quite audible.

 

Chris

[Grizzog]

I wondered the same thing - how far apart do the drivers have to be to hear any issue? I've never had any problems with my systems and being able to pick out specific drivers...at least at the proper listening distance. What I have noticed with having my new studio monitors, though, is that time-alignment allows for better soundstaging. The depth to the stage is very apparent. It's one thing to be able to left/center/right placement, but the depth seems much easier to pick out with time alignment. I'd have to imagine PWK would've had a pretty good idea on its effects.

[jwgorman]

6 hours ago, Chris A said:

See http://www.aes.org/e-lib/browse.cfm?elib=2039

 

This is one area where it is clear to me that PWK's views changed over time.  Klipsch has published delay settings for its professional/cinema loudspeakers for many years.

 

 

One of the aspects of crossover filters is that there is a 90 degree phase lag of the LF section per "order" of the crossover filter used (Butterworth).  So for higher order crossover filters, you must delay the HF section by the correct amount...or set the drivers farther back than the midrange driver.

 

The way to set the delay correctly is by measuring it using something like REW or TrueRTA.  Setting the delay wrong is quite audible.

 

Chris

So Chris, per the link you provided, can one infer that the time delays encountered in something like a Heresy or a Cornwall would not be particularly important, according to PWK? 

[pauln]

Let the speed of sound be 343.2 m/s

 

hz    wl in m    wl in s    wl in ms    4.1 ms %
400    0.86    0.003    2.5    164%
283    1.21    0.004    3.5    116%
200    1.72    0.005    5.0    82%
141    2.43    0.007    7.1    58%
100    3.43    0.010    10.0    41%
71    4.85    0.014    14.1    29%
50    6.86    0.020    20.0    21%
35    9.71    0.028    28.3    14%
25    13.73    0.040    40.0    10%

 

First column is woofer Hz

Second is wavelength in meters

Third is wavelength in seconds

Fourth is wavelength in milliseconds

Fifth is what percentage of the wavelength in seconds is 4.1 ms - the percentage indicates how much of a full wavelength is being delayed... 100% means a whole wavelength of shift, 164% means a little more than a whole wavelength, etc...

 

So what is interesting is that for very low frequency sounds, the 4.1 ms delay is negligible, especially since low frequencies take more wavelengths to be discriminated in the first place, and you can't determine pitch from a quarter or half wavelength, actually you have to hear quite a few cycles to get it, the lower the frequency the more cycles it takes to determine.

So the lower percent portion that is the 4.1 ms delay (potential advantage) is buried in how long it takes hear something down there anyway.

 

On the other hand, the higher frequencies are being delayed an increasing percentage of their wavelength - those exceeding 100% are delaying more than the first cycle.

But even at the high end, at 400Hz the 4.1 ms delay is only acting to shift its first approx. one and a half cycles (or maybe as if it cut off the first 1 1/2 cycles and "added" it to the end of the sound).

 

Anyway, the numbers are interesting? 

[JL Sargent]

Its an interesting subject because there are so many great sounding speakers out there with no time alignment.

[Chris A]

Sounds like a lot of guys don't want loudspeaker time alignment to be audible.  You must hear it (A-B) to believe how real and important it is. 

 

Here is a thread on tri-amping and time-aligning a Belle that you can use to understand just how much:

I found that phase delay mismatches as little as 30 degrees at the center crossover frequency (bass bin-midrange and midrange-tweeter) were quite audible and affected the overall timbre of the loudspeaker strongly.

 

Chris

[Don Richard]

There are several effects caused by temporal offset in the drivers of a loudspeaker system. The first is hearing two distinct sounds around the crossover frequency, normally heard when the offset exceeds 10 ms., depending on the frequency. Another effect is increased side lobing in the loudspeaker's polar response. Another effect is steering the dominant lobe of a particular driver within a loudspeaker system in the direction of the delayed driver. This causes irregularities in the overall frequency response of the system as that particular driver's dominant lobe moves around. Then there is the comb filtering.

 

This is not to say that a loudspeaker must be time aligned to produce enjoyable sound, but when it is properly aligned the frequency response will be smoother and the sweet spot much wider. However, in a system with vertically stacked drivers the alignment will be correct for one vertical height only. If you stand up or sit on the floor you are no longer listening to a system where the sound from the different drivers in the speaker arrives at the same time. The ultimate solution would be a time aligned system with coaxially mounted drivers or a unity horn arrangement.  

 

 

[Chris A]

If you read the Belle tri-amp thread linked above, you will see the degree of sensitivity of the time alignments were to the timbre of the sound.  This was a very stable phenomenon whether sitting or standing. 

 

I was completely caught off guard by how sensitive it was, especially for the tweeter-midrange...a few microseconds (one or two clicks of the Dx38 delay setting dial) were enough to drive me out of the room when playing a sine wave at the center crossover frequency.

 

It was from that experience that I learned why eliminating one crossover was PWK's goal in both the original Khorn and the next-generation Khorn (i.e., the Jubilee).  Getting the crossovers right in terms of correct delays, frequency/phase response, and matching coverage angles in the crossover region is much more difficult than most people want to believe.  I found that minimizing their number is critical to achieving a neutral and balanced sound.

 

Chris

[Schu]

I would love to learn more about delay and more importantly about phasing as it relates to XO's.

[Quiet_Hollow]

7 hours ago, pauln said:

So what is interesting is that for very low frequency sounds, the 4.1 ms delay is negligible, especially since low frequencies take more wavelengths to be discriminated in the first place, and you can't determine pitch from a quarter or half wavelength, actually you have to hear quite a few cycles to get it, the lower the frequency the more cycles it takes to determine.

So the lower percent portion that is the 4.1 ms delay (potential advantage) is buried in how long it takes hear something down there anyway.

Testers often focus on the audibility of specific delay, but that's only half the problem. The other half is the resultant shift in power. Line up all the impulses correctly in time, and the difference is not only audible, but also very tactile. The energy in the sound climbs rapidly as frequency decreases.

[Chris A]

On 8/21/2016 at 11:39 AM, Schu said:

I would love to learn more about delay and more importantly about phasing as it relates to XO's.

 

Candidly, I was made specifically aware about the 90 degree phase shift/filter order from Tom Danley, for instance a much more recent posting on the subject: http://www.audioasylum.com/audio/speakers/messages/35/351039.html.  The Rane tech note expands on that much more (see figure 10): https://www.ranecommercial.com/legacy/note160.html

 

 

It's not as if filter designers didn't already know that there would be a phase shift, it's that this fact simply gets lost in the all the discussions about "summing for flat FR in the crossover band"...and also higher order filter types, such as Linkwitz-Riley (Butterworth squared) filters.  Everyone seems to be worried about the FR when they should be as worried about the phase shift--LF to HF.  I found that simply going from 24 dB/octave L-R crossover filter (digital) to 48 dB/octave introduced additional phase shifts in the Jubilees when I used the Yamaha SP2060, which had the higher order filters available than the EV Dx38, introduced a dip in response at the 425 Hz crossover frequency that caused me to shift the published HF delay (from Roy) to a bit more delay to compensate the FR dip.

 

I believe that one reason that many people like the sound of lower order crossover filters is that they introduce less phase shift than higher order filters. (Another reason is that there are typically large coverage angle mismatches between the crossed drivers/horns that tends to get rounded out in the crossover band using lower order filters, but at a high price in terms of distortion of the imaging of the loudspeaker, and loading of tweeter and midrange drivers well below the crossover center frequency that has power handling and harmonic distortion consequences.)

 

There are also other subjects that are equally obscure...at least to me...like why many like the sound tube amplifiers with high efficiency loudspeakers, because of their "current source" transconductance characteristics that are less likely to respond to loudspeaker driver nonlinearities that voltage-source amplifiers with extremely low output impedance will amplify more readily - and also coupling characteristics at loudspeaker high input impedance bands.  Perhaps a separate thread on that subject might be interesting to explore.

 

Chris

[Chris A]

If you look at a couple of the plots in the Belle tri-amping thread, you'll see an instance where I didn't set the delay correctly, and you see symmetric "side-band distortions" in the phase response, such as this plot:

 

 

This is a case where I was learning about how to measure and set correct delays, and Greg Oshiro patiently and calmly suggested that I take a some time to "get the hang of it".  That exercise was very instructive and informative as it turns out.  I wouldn't trade that experience now.  Using REW (freeware), a calibration microphone (about $100, with small mixer to provide phantom power to the microphone and to digitize its output back to my laptop) and a digital crossover to dial in changes and see the effects--was a very useful experience.  I highly recommend it to anyone that's a bit more serious about their loudspeaker performance if they get a chance to use a digital crossover such as an ElectroVoice, Ashly, Xilica, etc. and stereo amplifiers to tri-amp their 3-way loudspeakers.  The events around producing that Belle tri-amping thread was a fun and very useful experience.

 

Chris

[mikebse2a3]

Chris just to clarify aren't you locating the Mic at ear height in the primary listening position when setting the delay time..?

 

miketn 

[Chris A]

16 hours ago, jwgorman said:

So Chris, per the link you provided, can one infer that the time delays encountered in something like a Heresy or a Cornwall would not be particularly important, according to PWK? 

Sorry for the delay in response on this one. 

 

What I've found is that, at least in the case of the Cornwall 1.5 (1979 model), and perhaps just by chance...but probably not, the Cornwall balancing network seems to be "phase aligned", in other words, the balancing network and the position of the drivers in the cabinet pretty much adjusts the phase delay of the midrange and tweeter to be multiples of 360 degrees of phase to one another.  While this isn't as good as full time-alignment in terms of the listening experience, it definitely cuts way down on the phase/time delay distortions that would otherwise detract more seriously from the listening experience. 

 

So from this observation, it's clear to me that PWK knew about phase/time delay issues and actually worked to minimize their effects-probably going back to the beginning--whatever year "AD" that happened to be. (By the way, PWK himself mentioned that phase and time delays are the same thing in the referenced article he penned.)  I assume that PWK also took steps to do the same with the other Heritage loudspeaker models.  If you can't fully correct the time delay issues, then you should at least phase align the drivers via the balancing/crossover network at multiples of 360 degrees from each other. 

 

Chris

[Chris A]

38 minutes ago, mikebse2a3 said:

Chris just to clarify aren't you locating the Mic at ear height in the primary listening position when setting the delay time..?

When setting delays for each loudspeaker driver, the REW calibration microphone is on-axis, X-Y, at 1 metre from the front face of each loudspeaker. 

 

Later, when setting delays for each channel in the 5.2 array, the preamp calibration microphone is located at various trial locations at the two main listening positions (three places across the two seats).  For this role, Audyssey takes out the slack in the delays of the loudspeakers, their distances from the listening position, and the various delays of the digital crossovers.

 

Chris

[Chief bonehead]

On 8/21/2016 at 4:34 AM, JL Sargent said:

Its an interesting subject because there are so many great sounding speakers out there with no time alignment.

are they hornloaded?

[Chief bonehead]

On 8/21/2016 at 5:47 AM, Chris A said:

Sounds like a lot of guys don't want loudspeaker time alignment to be audible.  You must hear it (A-B) to believe how real and important it is. 

 

me hears nothing.  me sees nothing.  me knows nothing...

[Chief bonehead]

On 8/21/2016 at 6:17 AM, Don Richard said:

There are several effects caused by temporal offset in the drivers of a loudspeaker system. The first is hearing two distinct sounds around the crossover frequency, normally heard when the offset exceeds 10 ms., depending on the frequency. Another effect is increased side lobing in the loudspeaker's polar response. Another effect is steering the dominant lobe of a particular driver within a loudspeaker system in the direction of the delayed driver. This causes irregularities in the overall frequency response of the system as that particular driver's dominant lobe moves around. Then there is the comb filtering.

 

This is not to say that a loudspeaker must be time aligned to produce enjoyable sound, but when it is properly aligned the frequency response will be smoother and the sweet spot much wider. However, in a system with vertically stacked drivers the alignment will be correct for one vertical height only. If you stand up or sit on the floor you are no longer listening to a system where the sound from the different drivers in the speaker arrives at the same time. The ultimate solution would be a time aligned system with coaxially mounted drivers or a unity horn arrangement.  

 

 

using the cone as horn or a horn loaded driver?