Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Peter Morris]

A few observations:
1. Mark Engebretson - the author of the white paper - knows what he's doing. Nobody has said or implied otherwise.
2. The white paper in question has nothing to do with the subject of this thread.
3. Intentional polarity reversal of one element in a two-way loudspeaker is very common practice. When that is the practice you've adopted - as QSC has done here - it ill-serves any legitimate purpose to assert that the tweeter is in polarity. That's exactly what Bob Lee did in the QSC forum. It is not in polarity, and you can establish that with a simple clicker. Whether the speaker works better with the polarity corrected is arguable, as I acknowledge in the thread in the FAS forum (http://forum.fractalaudio.com/amps-cabs/37266-qcs-k10-tips.html). Whether it is actually in polarity is not a matter of interpretation or opinion, however.

ps - You should hear the story Don Davis tells - and Mark Gander confirms - about Mark Engebretson and acquiring loudspeaker response data back in the early 1960s.

You missed the point… anyway ...  the K series uses Intrinsic Correction™ - that means they are combining both IIR and FIR functions. They are able to directly weight the delay taps  that derive the FIR response. As such they can modify amplitude, phase and time independently.  This approach is way way  beyond the world of simple crossovers that  Richard Small discussed in his paper 40 years ago. 

Having said that - it doesn’t mean someone hasn’t wired one  incorrectly during assembly or  there is something else wrong, but I’m sure the guys at QSC know what they are doing. Especially something as simple as get the polarity of the HF driver optimized.


Peter



Peter

[Jay Mitchell]

You missed the point… anyway ...  the K series uses Intrinsic Correction™ - that means they are combining both IIR and FIR functions.

That is essentially meaningless in this context. As the data I took makes abundantly evident, no signal processing - IIR, FIR, or analog - can ever correct directionally dependent speaker response problems caused by three-dimensional geometry.

They are able to directly weight the delay taps  that derive the FIR response. As such they can modify amplitude, phase and time independently.

The one potential benefit of FIR crossover filters is to narrow the range of interaction between low- and high- frequency elements and therefore to narrow the inevitable crossover notch that occurs  with all non-coincident loudspeaker designs. The relatively wide notch that my data shows is a pretty strong indicator that this benefit was not realized.

White papers are great marketing tools. Documented product performance is much more compelling, and that's what I've acquired here.

This approach is way way  beyond the world of simple crossovers that  Richard Small discussed in his paper 40 years ago.

Yet the practical result - assuming that "this approach" is even implemented in the K10 - reflects no immunity to the principles Small laid out in that paper. Ain't it cool how correctly-done mathematics never becomes obsolete?

[Peter Morris]

You missed the point… anyway ...  the K series uses Intrinsic Correction™ - that means they are combining both IIR and FIR functions.

That is essentially meaningless in this context. As the data I took makes abundantly evident, no signal processing - IIR, FIR, or analog - can ever correct directionally dependent speaker response problems caused by three-dimensional geometry.

They are able to directly weight the delay taps  that derive the FIR response. As such they can modify amplitude, phase and time independently.

The one potential benefit of FIR crossover filters is to narrow the range of interaction between low- and high- frequency elements and therefore to narrow the inevitable crossover notch that occurs  with all non-coincident loudspeaker designs. The relatively wide notch that my data shows is a pretty strong indicator that this benefit was not realized.

White papers are great marketing tools. Documented product performance is much more compelling, and that's what I've acquired here.

This approach is way way  beyond the world of simple crossovers that  Richard Small discussed in his paper 40 years ago.

Yet the practical result - assuming that "this approach" is even implemented in the K10 - reflects no immunity to the principles Small laid out in that paper. Ain't it cool how correctly-done mathematics never becomes obsolete?

This is from QS K10 brochure-

K Series features Intrinsic Correction™, a process found on many of our high-end concert sound products. Far more than just simple signal processing, Intrinsic Correction begins with between 60 and 75 extensive measurements of the loudspeaker's characteristics taken across a spatial field. The results of those measurements are then averaged.

Utilizing an average derived from this many sampling points produces results that are far more representative of the loudspeaker's performance than taking a typical, single on-axis measurement. Next, a combination of IIR and FIR filters is applied in the DSP engine to adjust time, frequency and amplitude response to a maximally flat band-pass target.

The end result is a loudspeaker with flat "power response", meaning that the overall energy radiated by the loudspeaker is even across its operating range. Those wishing to learn more may read the Intrinsic Correction White Paper.

Soooo …. the white paper on intrinsic correction is relevant even though the example used to explain the principles in the paper was based on their line array.

My point was very simple …. If you are smart enough to do this stuff, the issue of the HF drivers polarity and correct summation thought the crossover band is trivial, its 101 speaker design and the boys at QSC would not make that mistake, certainly at the design level.

If there is any problem it could be an assembly error, the HF or LF could be  installed with the polarity the wrong way around …shit does happen from time to time.  There is also another possibility; as I understand, all of the K series more or less uses the same amp/DSP module. In this case, the DSP could have a programming error e.g. the settings for the K8 or 12 could be used which have different time alignments and EQ’s and produce an error. 

If you are using FIR filters there are some tricks you can do to correct for some directional properties at the crossover region that are not really possible with standard crossovers.  These guys have an example on their web site. (can’t remember where it is exactly) http://www.four-audio.de/en/technical-articles.html . QSC didn’t do any of these ticks as far as I know.

FIR filters may or may not have high slope rates. The Lake processor for example can emulates the amplitude response of a LR 24 or 48 dB / oct filter while maintaining zero phase shift. If the phase needs to be modified it can be done with digital all-pass filters.  More recently Lab Gruppen and Powersoft have allowed 3 party access to the FIR coefficients in their processors to allow EAW focusing (similar to Intrinsic Correction). The world has moved way beyond the simple world of correct summation at the crossover. People are now focusing on impulse & power responses, issues in the time domain, improved phase response...and...and ...  and this approach is now starting to appear in MI boxes such as the K series.

Peter

[Jay Mitchell]

The end result is a loudspeaker with flat "power response", meaning that the overall energy radiated by the loudspeaker is even across its operating range.

The response data taken both by myself and the "other" Jay in this thread make it abundantly clear that the K10's power response is not flat. There is clearly a substantial power deficit in the crossover region.
In order for an average to be flat, there must be some angles at which there are response peaks to offset those angles at which there are response notches. This is most definitely not true of the K10. The marketing fluff and actual measured product performance do not match.

Soooo …. the white paper on intrinsic correction is relevant even though the example used to explain the principles in the paper was based on their line array.

See above. It is only "relevant" in that it is clearly not descriptive of what the product actually does.

If there is any problem it could be an assembly error,

If there is an "assembly error," then there are at least two, since Jay's data show that the tweeter in his K10 is also reverse polarity. There is also an inconsistency in Bob Lee's statement that, if the tweeter is in polarity, there will be no notch in the on-axis response at the crossover frequency. The opposite is true: the on-axis notch is absent with the tweeter in reverse polarity.

If you are using FIR filters there are some tricks you can do to correct for some directional properties at the crossover region that are not really possible with standard crossovers.

I am very familiar with what FIR filters make possible. One thing they do not make possible is violation of the principles that govern the radiation of acoustic energy.

The world has moved way beyond the simple world of correct summation at the crossover.

The rhetoric has "moved beyond" that. The reality - as clearly demonstrated in the performance of the product - has not.

People are now focusing on impulse & power responses,

Apparently not in the case of the K10.

[Jay Barracato]

One more observation: your phase plots clearly indicate a reverse-polarity tweeter, as do mine. This establishes that at least two K10s came from the factory this way.

You can make it 3, I measured two different boxes. I am in perfect agreement that the stock tweeter has a negative initial impulse. This was also shown in the impulse response graphs that I measured.

I entered into this thread being an interested K10 owner and user. I took the time to read the threads on both the QSC board, and on your own board. I spent a pretty significant amount of time looking at your graphs and trying to match them with your posted conclusions.

Since then, I see that the tone has gone from "is definitely better" to "is arguably better" and I still invite you to make those arguments. Despite all the words you have used, you have not answered the two basic questions I have asked:

1. Please describe the physical setup of your measurement arrangement.

No where in this discussion have I questioned the quality of your measurements. All my initial statement meant was that not knowing the quality, all I could do is interpret them at face value. However, before I could even possibly consider acting on your measurements, I want to be able to replicate them myself. I think if you look carefully, you will see that my measurements confirm most of yours. All I could not do is replicate the magnitude of the largest dip. This could just as easily be a difference in resolution, sampling/smoothing, etc.

2. Please defend your conclusion that reversing the polarity to the positive position is "better" using specific examples from your data.

In both this thread and the thread on your board, your definition of "better" seems to be centered around "transient response". Earlier I asked you to define this. What in the data shows better "transient response"?. I think we are in agreement that the polarity reversal moves the largest response notch from off axis to on axis. What do we gain be introducing a 1000 hz wide, 20 db deep notch into the on axis response?

Each time I asked this basic questions (which are exactly what I would expect to accompany any public display of data), instead of answering you have responded with a personal vitae, a condensending discussion of some basics which have nothing to do with interpreting the data, additional totally off topic items, and ample jargon. If I received that type of response during a professional presentation of data, I would have raised a "attempt to baffle with bullshit" flag quite quickly.

After all of this I have to fall back on my original statement. I am unconvinced. Because you have failed to support your position with evidence directly from the data, I still must rely on my own conclusions.

For the record, my conclusions based on both your and my own measurements are:

1. The stock position of the HF has a negative initial impulse (negative polarity).

2. In the stock position, the response is relatively flat through the crossover region on axis. Based on my measurements, as you move off axis in the horizontal plane, there is very little change in the response until you start to approach the published limits of the coverage pattern. Based on my measurements and yours, as you move off axis in the vertical plane, there is a dip in response that gets larger towards the published limits of the coverage pattern. My measurement of the size of the dip was about half of yours.

3. Reversing the polarity of the HF moves the largest dip in response directly on axis. In the 10-20 degree off axis measurements (using both your and my data) there MAY be a slight improvement in response, however, because all the positions are so close in response and similar to the stock position, I am not sure if this is a signifigant (I mean that statistically) change or the natural random variation of multiple measurements of the same thing. Reversing the polarity (based on your measurement) greatly improves the response at the 30 degree off axis position.

4. Based on your measurements changing the polarity of the HF moves the phase from 0 to -15 degrees at the crossover to -60 to -90 at the crossover. In both positions, the slope of the phase is negative through the entire crossover region, but changing the HF polarity steepens the phase response.

5. Based on your impulse response curve, the amount of time for the initial impulse is the same for both positions, just opposite in direction. In both positions the amount of time to return to the steady state (my definition of transient response) is the same. The positive polarity position may show a slight (fraction of a mv) more damping.

Since I deploy my speakers with more of the listeners closer to on axis than off axis, I see the stock polarity as the "better" position. If anyone sees anything different in this data set, my mind is still open to alternative interpretations.

P.S. just because I said early in the thread that I wasn't a speaker designer, doesn't mean that I don't have a technical background, lots of experience with measurement and data, and the process of peer review.

[Rich Grisier]

In both this thread and the thread on your board, your definition of "better" seems to be centered around "transient response". Earlier I asked you to define this. What in the data shows better "transient response"?. I think we are in agreement that the polarity reversal moves the largest response notch from off axis to on axis. What do we gain be introducing a 1000 hz wide, 20 db deep notch into the on axis response?

I still plan on at least trying this when I get a chance.  I'll set it up with some different types of program music to hear how it sounds.  Maybe hook up a drum machine to it and program some basic beat to hear how that sounds.  If it sounds dead and lifeless in one polarity vs. the other then I'll go with the better sounding one.  If it's too close to tell then I'll stick with the stock setting.

[Peter Morris]

http://srforums.prosoundweb.com/index.php/m/579762/0/#msg_579762

It seems not everyone got the same results as you did !!!! ….Hmmmm

These results mirror what I measured with the K8
 
- bass boost off 
 -vocal boost on (see message #579779)
- on a stand
-about +/- 3 dB 70Hz to 16Khz (which is inline with what QSC are claiming)

Considering where the box is priced - great results.
 While I didn’t like the  MI sales EQ- ing - bass boost and or no vocal boost settings etc.  I suspect the marketing guys got it  right making it easier for the salesman in the shop to sell their product over boxes with a flatter sound.


Peter

[Mac Kerr]

http://srforums.prosoundweb.com/index.php/m/579762/0/#msg_579762

It seems not everyone got the same results as you did !!!! ….Hmmmm

These results mirror what I measured with the K8
 
- bass boost off 
 -vocal boost on (see message #579779)
- on a stand
-about +/- 3 dB 70Hz to 16Khz (which is inline with what QSC are claiming)

Considering where the box is priced - great results.
 While I didn’t like the  MI sales EQ- ing - bass boost and or no vocal boost settings etc.  I suspect the marketing guys got it  right making it easier for the salesman in the shop to sell their product over boxes with a flatter sound.


Peter

Who are you responding to? There are at least 2 different people who posted test results you could be referring to. The quote is there for a reason, to eliminate this confusion. Why did you go to extra effort to make your post less clear?

Mac

[Peter Morris]

Who are you responding to? There are at least 2 different people who posted test results you could be referring to. The quote is there for a reason, to eliminate this confusion. Why did you go to extra effort to make your post less clear?

Mac

Thanks Mac

I was responding to Jay Mitchell .... when I posted every thing  locked up when I was trying to edit the quote. Needed to go and do some work and I didn't have time to post again...

Thanks

Peter

[Jay Mitchell]

You can make it 3, I measured two different boxes. I am in perfect agreement that the stock tweeter has a negative initial impulse.

That is the definition of "reverse polarity." Case closed. This is the point that Bob Lee and others argued incorrectly, and it is the only reason I bothered registering to post here. I posted the same polarity information on the QSC board, but the post was deleted.

Since then, I see that the tone has gone from "is definitely better" to "is arguably better"

No. That was my original "tone." You apparently did not read the first post in the sticky on the FAS forum. Here's a quote: "First, the speaker comes from the factory with the HF polarity reversed. Although this is a common "feature" of loudspeakers, it trades one set of problems for another, arguably worse, one." That was posted well before this thread was begun.

1. Please describe the physical setup of your measurement arrangement.

Asked and answered. In these particular measurements, the speaker and mic were elevated above the floor about 2 meters, but the results would be identical with a ground-plane measurement. To make it as clear as possible, there are no environmental reflections in my data.

2. Please defend your conclusion that reversing the polarity to the positive position is "better" using specific examples from your data.

It is audibly - to me, and to a significant number of other listeners - superior to have all elements in a multiway system respond with the same polarity. Furthermore, it is best when that polarity is positive, IOW a positive-going voltage at the input produces positive-going pressure at the output. Sounds containing transient information - i.e., drums, cymbals, other percussion instruments - are better-defined. It is easier to hear how the drummer has tuned his snare drum and snares, for instance. Additionally, the attack portion of the sounds produced by percussively-excited instruments - acoustic piano and guitar, for example - are better-defined to my ears.

What in the data shows better "transient response"?.

Positive polarity of both woofer and tweeter are necessary - but not sufficient - conditions for the best transient response. When reversed polarity is corrected, transient response is improved by definition.

For the record, my conclusions based on both your and my own measurements are:

1. The stock position of the HF has a negative initial impulse (negative polarity).

Again, case closed. That was and remains my point in responding here. You agree. Why argue further?

changing the HF polarity steepens the phase response.

Having both devices in correct polarity makes the group delay error caused by the crossover transition apparent. That error is there, regardless. Any argument that the speaker's behavior has been "optimized" via strategic use of FIR filters is clearly shown to be false by this fact alone. Edit: To clarify: the amount of group delay error in the crossover transition in the QSC is quite small and does not IMO constitute an audible issue. It is there, however, regardless of the choice of tweeter polarity.

5. Based on your impulse response curve, the amount of time for the initial impulse is the same for both positions, just opposite in direction.

"Direction" matters.

Since I deploy my speakers with more of the listeners closer to on axis than off axis,

Describe the typical dimensions and geometries of your "deployments," and we'll do the trig. My money says you're mistaken on this one.

[ProSoundWeb Community]