Frequency Response/Contour EQ in full range systems.

Re: Frequency Response/Contour EQ in full range systems.

Ivan,
Concert PA systems almost always have a curve similar to the l'acoustics curve. This is not new information. It is a reality. I attribute this to "shape" of popular music. You disagree. What is your explanation?
I assume you have actual measurements to back up your claim? I have seen very few (as in only a couple) measurements of a system that is "in place" and ready for the show.

I would love for the people that are following this to post ACTUAL measurements of systems so we can look at them. NOT RTA screen shots of a system in use. I have posted many times measurements of all sorts of different sizes of installs that I have done. The common thing is flat from 100Hz up-as high as I can get it without straining the drivers. I have many hundreds of customers who are happy with that.

They just mix and use the individual tone controls as needed to get the result they want.

My guess is that the measurements by others will be all over the map. The only general thing they might have in common is that the low freq is louder than the higher freq.

But does that mean that is correct? If you really want to look at a common "item" that is used by most people but is actually WRONG, just look at the polarity switch on consoles. Only a few of them term it correctly-polarity. The rest call it "phase", because that is "common" but is still wrong. And all the "explanation" still does not it correct.

And I will STILL argue that the SHAPE is due more to the eq and instrument levels-RATHER than the actual eq/response of the PA. And on the top end-if the FOH guy likes lots of cymbals, then the "shape" of the actual output will be higher than if he doesn't turn up the cymbals as much. And yet the "shape" of the PA has stayed the same.

When you put a low freq boost on the kick drum, that is a channel EQ, NOT the system eq. But is reflected in the "shape" of the sound we hear.

Just because the sound we hear is generally bass heavy-does not mean that the "system" is bass heavy-just the mix.

The amount of this applied eq depends on the actual response of the system. Some systems need more bump than others. So therfore they are not all the same.

I really don't know how to explain it any better than that.

Of course others have different opinions on how to do it, but I get good results (and happy customers) with my way-so I will stick with that.

Feel free to do it whatever way gets you business.
 
Re: Frequency Response/Contour EQ in full range systems.

I agree that as long as they aren't breaking anything, they should be allowed to do as they wish-sometimes it is very much a part of the "style" of the music.

I just relate that a flat system (more on "flat" in a second) is like a white canvas to a painter. If the canvus is already "colored", then the artist has a harder time creating the image they want to.

I see the beauty of handing someone a white canvas. But you know how some people use grey projection screens to help make the black level of the image really black? Kinda the same thing, it's just an EQ to compensate for any number of phenomenons. If whoever mixing likes to start with an "off-white canvas" to precolor everything just a little bit -and chooses to mix in extra white in thoose sections that don't need the offwhite precoloring - who's to say they're wrong? In my opinion it's just another technique for an artist to use. But it needs to be the artist who chooses the canvas - not some random person.

I can think of an example in lighting also: Many light techs are so used to how the bulbs of PAR cans add color "contamination" to colored filter gels that when they use true white discharge lights to actually get them the color that the filters have - they don't like it and have to add color in other ways to get to the "inncorrect" color they usually use.
 
Re: Frequency Response/Contour EQ in full range systems.

The common thing is flat from 100Hz up-as high as I can get it without straining the drivers.

When I am very happy about my system tech-ing, this is how it measures. From as low as I can get it till 100Hz, it stays at +12 dB, then from 100Hz upwards, I try to keep it at 0dB, as high as I can get it.
A not so system tech savvy colleague of mine describes the result like this: "If you tune a system, and I turn my lead singer on, I hear a normal, natural voice, without any channel EQ"
Maybe this is the result of growing up as a sound dude with a lot of folk(rock) and world music, as opposed to heavy rock, metal or techno?
I really believe in a soundsystem being a means of amplifying a band, not distorting. I think that the sound of a band should be made by the talent on stage and behind the desk, not by the sound system.
But then again, if I would be touring with a band engineer who wants me to make the system sound a certain way, I could see how that could be part of a signature sound of a band.
Most of the time however, I don't have to please one BE, but a lot of them, or I only meet the BE right before soundcheck, for a one-off...

Just my 0.02€...

Charlie
 
Re: Frequency Response/Contour EQ in full range systems.

I agree with Tim - 'flat' has all sorts of (often negative) connotations, so the term I use is 'accurate'. If I'm tuning an installed system and they have a knowledgable tech, I'll give 5 minutes overview of dual channel FFT, and how we start with an accurate system from about 100Hz on up. Very similar to what Ivan does. After that, I don't mind some light tweaking, based on their preferences, usually how much sub boost and maybe some HF rolloff they prefer. Maybe even a couple presets in the DSP for 'low SPL' and 'rock!'.
If it's a generic system, IE a neutral canvas for many FOH mixers (med/large churches here), then the flat above 100 typically works well.

Someone I was talking to recently said a non-neutral PA sounds like a hold-over from old times. Front loaded 18's, 15's, and 2" horns - lots of 50-110, sloping down to 1kHz, some 1-6kHz, then rolling off above that.

Caleb
 
Re: Frequency Response/Contour EQ in full range systems.

I try to tune the sound system I am using for linearity. What goes in is what comes out. However when I do this I always add an inserted EQ on the vocal sub group or groups. For these vocal subs I group like mics together. I then EQ for the best gain before feedback. Usually by routing one of those vocal mics to my measurement system and using that as my measurement mic for the inserted EQ (while playing pink noise thru the mains only routed thru that EQ) and I tweak (cut) the frequencies that are shown to be accentuated with this form of measurement. I find I usually still have to make some cuts on the inserted EQ in the 200hz range. I find my method gives me the linearity that I need for miced instruments and playback and the best sound for the vocal mics. YMMV
 
Re: Frequency Response/Contour EQ in full range systems.

Here is a random group of traces that show the general trend. Not saying this looks ideal :)
I added a picture of the RTA with my favorite test track rolling. There are obvious similarity between the system response and the average "shape" of popular music. I understand that the curve of the PA will act like a filter and Ivan's Zeppelin through a marshall cab analogy, but I also don't think its a complete coincidence that many PA's have a general sloping trend from low to hi freqs.
 

Attachments

  • Picture 16.png
    Picture 16.png
    60.7 KB · Views: 2
  • Picture 15.png
    Picture 15.png
    88.1 KB · Views: 3
Last edited:
Re: Frequency Response/Contour EQ in full range systems.

Charlie Hughes' comments about getting different results when measuring in different acoustic environments, at different locations, or with different time windows are apposite.

Notwithstanding the point about direct field versus power response, I'm in the "flat response" camp when it comes to system voicing for general use (in general, indoors, measured far field but close to the source, with no time windowing), maybe allowing a few dB bump below 100Hz (if speech intelligibility matters and routing to the subs can't be controlled per channel, then no more than a few dB). Some styles of music have certainly come to be associated with exaggerated bass, but I'd rather add that emphasis to channels where it's appropriate than have it applied to everything reproduced by the system.

With regard to time windowing, some measurement systems enforce a time window by their nature, some don't support one at all, and some (like Smaart) make the facility available at the user's discretion. In my experience of the sorts of system-tweaking I think we're talking about, few people using Smaart employ a time window so the magnitude response ends up being the same as an "RTA" analysis (which is to say, the magnitude of the sound pressure field at the measurement location). The time window embodied by the FT process itself isn't the same thing.

I'm still quite attached to my handheld RTA. Its usefulness stems from the ability to quickly measure an average response whilst moving across a loudspeaker's area of coverage. Aberrations from specific reflections are averaged out, though one still needs a brain to spot a floor reflection.

I don't really understand the notion that "flat sounds bad", though it obviously depends entirely on what is being measured and in what way. "Flat" sounds like ... whatever the source sounds like. I'm yet to hear anyone complain about a mixer or power amplifier for exhibiting flat response.

Nick
 
Re: Frequency Response/Contour EQ in full range systems.

Here is a random group of traces that show the general trend. Not saying this looks ideal :)
I added a picture of the RTA with my favorite test track rolling. There are obvious similarity between the system response and the average "shape" of popular music. I understand that the curve of the PA will act like a filter and Ivan's Zeppelin through a marshall cab analogy, but I also don't think its a complete coincidence that many PA's have a general sloping trend from low to hi freqs.

Brandon,

I think your conflating the idea that pop music has a specific shape with system tuning shape. I would submit that the natural spectral shape of music content is much more fundamental than for "pop" music, and that music's shape, as well as the general "softness" in the top octave is built in to human sound perception. The early frequency response RTA curves suggested in the mid 1920s look not unlike you RTA curve above.

-Phil

P.S. Even though I am generally supporting Ivan's position, please don't think I am blindly shilling for him. One of my most recent projects was a complete re-tune of a system that Ivan's company designed and installed, and Ivan tuned. My tuning approach, dsp configuration, and the end result, are noticeably different. The client is happier than with Ivan's tuning, and have retained me to re-work another one of their spaces installed by yet another AV integrator. We certainly approach things differently.
 
Last edited:
Re: Frequency Response/Contour EQ in full range systems.

With regard to time windowing, some measurement systems enforce a time window by their nature, some don't support one at all, and some (like Smaart) make the facility available at the user's discretion. In my experience of the sorts of system-tweaking I think we're talking about, few people using Smaart employ a time window so the magnitude response ends up being the same as an "RTA" analysis (which is to say, the magnitude of the sound pressure field at the measurement location). The time window embodied by the FT process itself isn't the same thing.

Nick

Nick,

As a point of clarification, transfer function measurements in SMAART most certainly do have time windowing applied. One can choose a fixed window by choosing a fixed length FFT, or one can use the MTW feature, which is turned on by default. MTW applies windows of different durations to different areas of the frequency spectrum (longer at low frequencies), but the response is most certainly time-windowed.

In Systune, one has a similar choice, but with yet another level of flexibilty. You choose the length of the window at 1kHz, and the software automatically picks different window lengths for the frequencies above and below 1kHz. The default window at 1kHz, if I recall correctly, is 1ms.
 
Last edited:
Re: Frequency Response/Contour EQ in full range systems.

Hi Phil,

As a point of clarification, transfer function measurements in SMAART most certainly do have time windowing applied, and there is some flexibility on the choice of windowing function. One can choose a fixed window by choosing a fixed length FFT, or one can use the MTW feature, which is turned on by default. MTW applies windows of different durations to different areas of the frequency spectrum (longer at low frequencies), but the response is most certainly time-windowed.

I think there are three different meanings of "window" in play here. First, if I have this straight, to do the DFT, the block of samples is multiplied by a window function (Hann, Hamming, or what have you). Second, the length of the DFT implicitly imposes a time window and, as you say, there are modes in Smaart (FPPO or MTW?) where different length DFTs are performed for different frequency ranges, thus giving different time windows.

Third, at least in Smaart 5 (which is what I use routinely), there's an optional "Time window" (specified in milliseconds) that can be applied after the TF has been derived by dividing the measured signal DFT by the reference signal DFT and which is achieved by doing an IDFT to get the IR, truncating it, and DFT'ing back to give the windowed magnitude response. It's this last windowing option that I was talking about.

The question of the practical usefulness of the implicit windowing involved in the DFT itself is fascinating. It's undoubtedly a good means of decimating data for display but, beyond that, I've always been sceptical of the claimed benefits. You may be able to help convince me! It certainly doesn't achieve the effect that IDFT'ing, truncating the IR, and DFT'ing back achieves.

Nick
 
Re: Frequency Response/Contour EQ in full range systems.

Hi Phil,



I think there are three different meanings of "window" in play here. First, if I have this straight, to do the DFT, the block of samples is multiplied by a window function (Hann, Hamming, or what have you). Second, the length of the DFT implicitly imposes a time window and, as you say, there are modes in Smaart (FPPO or MTW?) where different length DFTs are performed for different frequency ranges, thus giving different time windows.

Right.

Third, at least in Smaart 5 (which is what I use routinely), there's an optional "Time window" (specified in milliseconds) that can be applied after the TF has been derived by dividing the measured signal DFT by the reference signal DFT and which is achieved by doing an IDFT to get the IR, truncating it, and DFT'ing back to give the windowed magnitude response. It's this last windowing option that I was talking about.

Its been a while since I've used SMAART 5, but this order also seems right. Regardless of you get there, it still serves the same purpose, at least as long as you want to include the direct response in the window. I have not yet played with the IR enough in SMAART 7 to know if you can apply a secondary window to the IR away from the direct arrival. You can apply a window at any point along the IR in Systune, so if I have had the need to do that, I have used Systune.

The question of the practical usefulness of the implicit windowing involved in the DFT itself is fascinating. It's undoubtedly a good means of decimating data for display but, beyond that, I've always been sceptical of the claimed benefits. You may be able to help convince me! It certainly doesn't achieve the effect that IDFT'ing, truncating the IR, and DFT'ing back achieves.

Nick

Not sure what you mean about skeptical of the "claimed benefits." Windowing is at the core of using the FT kernel function for real signals.

At any rate, in real world equalization circumstance, I find that the direct response matters most for perception at high frequencies, especially in pro audio, and in the lower frequencies, where the array Q is substantially lower, a good sounding tuning also considers the magnitude of the power response of the array.
 
Last edited:
Re: Frequency Response/Contour EQ in full range systems.

As a point of clarification, transfer function measurements in SMAART most certainly do have time windowing applied, and there is some flexibility on the choice of windowing function. One can choose a fixed window by choosing a fixed length FFT, or one can use the MTW feature, which is turned on by default. MTW applies windows of different durations to different areas of the frequency spectrum (longer at low frequencies), but the response is most certainly time-windowed.

Correct. Both fixed length and MTW transfer functions are using a Hann window.
 
Re: Frequency Response/Contour EQ in full range systems.

Brandon,

I think your conflating the idea that pop music has a specific shape with system tuning shape. I would submit that the natural spectral shape of music content is much more fundamental than for "pop" music, and that music's shape, as well as the general "softness" in the top octave is built in to human sound perception. The early frequency response RTA curves suggested in the mid 1920s look not unlike you RTA curve above.

-Phil

P.S. Even though I am generally supporting Ivan's position, please don't think I am blindly shilling for him. One of my most recent projects was a complete re-tune of a system that Ivan's company designed and installed, and Ivan tuned. My tuning approach, dsp configuration, and the end result, are noticeably different. The client is substantially happier than with Ivan's tuning, and have retained me to re-work another one of their spaces installed by yet another AV integrator. We certainly approach things differently.
Phil,

Regarding the "natural spectral shape of music content", a girl with guitar, a classical orchestra, or a brass band will have a different spectral shape than Brandon's pop music, and Motorhead's music will look more like pink noise than any of the above ;^).

How does your tuning approach, dsp configuration, and the end result, differ "noticeably" from Ivan's?
I have heard some rather "noticeable" speaker changes over time, are you sure the system was still performing (and set) the same as when installed and tuned ?

Some would call a difference of 1 to 3 dB "noticeably different", others have to be hit with a 10 dB hammer before hearing a difference.

Do you have any before and after magnitude response pictures you could share to help quantify the "difference in end results"?

Art
 
Re: Frequency Response/Contour EQ in full range systems.

Phil,

Regarding the "natural spectral shape of music content", a girl with guitar, a classical orchestra, or a brass band will have a different spectral shape than Brandon's pop music, and Motorhead's music will look more like pink noise than any of the above ;^).

Art, the RMS content of essentially all recorded music follows the IEC curve. I suggest Dave Gunness' excellent powerpoint on this. It may not be intuitive, but it is measurable.

How does your tuning approach, dsp configuration, and the end result, differ "noticeably" from Ivan's?

I cannot account for Ivan's tuning approach, as I was not there for the original tuning. The only short answer I can provide to this question is that I changed/replaced/removed virtually every filter from the original tuning, and my final filter count, type, and behavior was entirely different from the original tuning.

More to the point, my tuning methods are my value-added proposition for clients, and it is imprudent on a number of fronts for me to exposit on those methods in this thread.

I have heard some rather "noticeable" speaker changes over time, are you sure the system was still performing (and set) the same as when installed and tuned ?

The system was less than three years old, and the existing tuning was, with certainty, the original tuning, as verified by the lead technical people who were present to observe and listen to the tuning process.

Some would call a difference of 1 to 3 dB "noticeably different", others have to be hit with a 10 dB hammer before hearing a difference.

The measure of success for me in this circumstance is satisfied clients and repeat business. I have both from this tuning, so draw whatever conclusions you will.

Do you have any before and after magnitude response pictures you could share to help quantify the "difference in end results"?

Art

In this particular case, I did not bother to take before responses. Even if I did, they would not be something posted lightly on the internet. The after responses are the clients, as a part of services rendered. In this particular case the measured curves are superfluous relative to the ready verification of the improvement in coverage from walking the space, and an overall tonality that was more in keeping with the clients' needs. I have no way to express either of these improvements meaningfully in a forum post, and no interest in trying to develop one.

That is all I have to say on this topic.
 
Re: Frequency Response/Contour EQ in full range systems.

Phil,

Was hoping for a bit more than a subjective response , “a picture is worth a thousand words :^)”.

You wrote “the RMS content of essentially all recorded music follows the IEC curve”, what IEC curve do you refer to?

I really don’t understand that statement, compressed music has more RMS content than more dynamic recordings, and the spectral content of music varies greatly with genre.

If you don’t care to say anything more, could you post a link explaining what you mean by “the RMS content of essentially all recorded music follows the IEC curve” ?
 
Re: Frequency Response/Contour EQ in full range systems.

For standards see IEC268-5 or EIA RS-426-B.

The relative compression of music, as recorded, has essentially no impact on the final RMS level of the music, as the phase shifts inherent in playback through a bandpass system that is even a perfect loudspeaker results in the reshuffling of the peaks such that the peak to average ratio tend towards that of gaussian noise.
 
Re: Frequency Response/Contour EQ in full range systems.

Hi Phil,

Regardless of you get there, it still serves the same purpose, at least as long as you want to include the direct response in the window.

But aren't they different beasts?

Not sure what you mean about skeptical of the "claimed benefits." Windowing is at the core of using the FT kernel function for real signals.

I think my point is that if the objective is to exclude late-arriving energy from the measurement, the time window inherent in the DFT won't achieve it. I created a little demo in relation to this years ago so I'll trot it out for your comment.

I used Smaart to measure some particular response. The 1/24th octave magnitude view below shows the result in orange from about 1kHz to 8kHz: it's gradually rolling off.

windowrta.png

I then created a copy of the signal that was delayed by 50ms and very heavy in 4kHz and added this in. This is the (exaggerated!) late-arriving reflection that I want to exclude from the measurement. The "RTA" view, of course, includes the late energy and this results in the magenta curve above.

I then looked at a variety of transfer functions below.

windowtf.png

The orange curve is the original signal in FPPO mode. It's the same as the orange RTA curve.

The yellow curve has the 50ms late-arrival added in FPPO mode. It's the same as the magenta RTA curve.

The green and magenta curves are the same thing using a 16K-sample and 512-sample (12ms TC) FT (@ 44.1ksamp/s) respectively.

The blue curve is the same thing using a 16K-sample FT but with a 30ms time window (i.e. IDFT, truncate, DFT). This is doing a pretty good job of chopping out the late energy.

All the examples have smoothing and a lot of averaging.

Interested in your wisdom!

At any rate, in real world equalization circumstance, I find that the direct response matters most for perception at high frequencies, especially in pro audio, and in the lower frequencies, where the array Q is substantially lower, a good sounding tuning also considers the magnitude of the power response of the array.

Absolutely.

Nick
 
Re: Frequency Response/Contour EQ in full range systems.

Interested in your wisdom!

Nick

Nick, what you describe certainly fits your assertion that the FPPO windowing is different, and insufficient. Unfortunately, I didn't program smaart V5, so I don't know the gory details about what you are seeing with respect to decimation and windowing. I can call Jamie later this week and ask him to elaborate, if you are interested.
 
Re: Frequency Response/Contour EQ in full range systems.

Phil,
I used the word "popular" very liberally to avoid examples of penny whistle solos etc..... The David Gunness presentation is what I was thinking of when I began posting on this thread, although I am not echoing the point of his presentation or implying he is of the same opinions as I am...... Also do you know of a copy of IEC268-5 on the web ?