Flat Frequency Response

[Ken Pachkowsky]

I have created this new thread after reading posts from two recent threads where the off topic conversation wandered into a discussion of "flat frequency response" vs. "Room Curves" (personally derived curves that deviate from the theoretical ideal to suite one's needs or desires). What follows are copies of posts from those threads... since the copies are posted in chronological order, there is a bit of a sequence shift from the merged posts, but I hope it will still make sense.


Widget


Hello Ken

How much EQ do you use on them???

Rob:)
Hi Rob, nice to hear from you.

I have only had them setup in 2 rooms to date. After analyzing the room from several positions using the gains on the 4-way crossover to get the system as flat as possible, I then put the BSS 960 into the chain and did the final adjustments. I have always found the need for fairly radical (+- 6/10 db) of adjustment in some lower freq due to room anomalies.

Don't get me wrong, it sounds great after eq'ing but this little experiment proves the point that every active piece of equipment is in effect another preamp with its own set of characteristics/anomalies.

In short it seems to reinforce the straight wire theory.

Ken

[Charley Rummel]

Greetings, all:

I've had the profound pleasure of mind-f@#&ing a few individuals over the years who initially laughed at my eccentric system, composed of mostly home-brewed tube gear (now driven primarily by a Mackie CFX16 in conjunction with a variety of other pieces) with 30 and 50 year old JBL and Altec gear on 5 channels ("...Charley, you actually (ha! ha! hee! hee!) have (ho! ho!) turntables?!?!). Ain't nothing like the feeling of turning a self proclaimed audio expert-snob into a broken man, as I'm sure many of you have also done.

On the other hand, different individuals expect different things from their preception of what fine audio gear should deliver. I lean more towards the effect of having it feel like the stage is actually in the room (like recreating the Grande Ballroom in Detroit with an MC5 concert), or where I'm within the first few rows centrally located in front of the event or orchestra (like Eugene Ormandy conducting the Philidelphia Philharmonic Orchestra), whereas some acquantences of mine find the effect of recreating the ambience of the concert hall or venue more apealing (such as the Lyric Opera from the balcony, or, dare I say, Pink Floyd 30 rows back from the stage).

Therefore, it's more like deciding what color paint or style of wallpaper an individual will select, rather than striving towards alleged technology perfection. No color at all is still a color, relatively speaking - right?

Kind Regards,
Charley Rummel

[Ken Pachkowsky]

No color at all is still a color, relatively speaking - right?

Kind Regards,
Charley Rummel

True enough Charley.

Ken

[Steve Schell]

Charlie, you made many valid points. Good sound is where you find it, and many people's preconceptions set them up for a mind-blowing reappraisal upon hearing a good vintage system. Converts are made this way; I'm sure many of us have had such an experience, I know I have.

Just think of the fun the Bell Labs engineers had in 1933 when they conducted the Auditory Perspective experiments and successfully reproduced the sound of a symphony orchestra in a large hall using the Fletcher Horn systems. At that point in time most of the public had heard only small and squeaky reproduction, so it must have been a mind-roasting demo for them.

As to individual preferences in sound and some folks' intolerance, Joe Roberts once said in Sound Practices magazine something to the effect that if no one has yet invented a hi fi system that is a perfect ten, why shouldn't he be able to enjoy an 8 or 9 that he likes?

[Ian Mackenzie]

Charley,

I suppose that is one trait that will stick with the older..larger systems ..they were very sensitive and dynamic even with the lower power amps back then so you could get that big close up presentation.

Perhaps the debate is then are the older systems using vintage drivers more natural sounding over the more technically advanced drivers used in like systems today?

Is more detail better? What I find interesting is that I can vary the degree of detail by using different amps old and new without resorting to fancy driver upgrades or materials.

Ian

[pentictonklaus]

like recreating the Grande Ballroom in Detroit with an MC5 concert


While comparing old and new speakers. The ones in the back are still the No. 1 choice for bringing up the MC 5 when they were asking :" Do you want to be the problem or the solution? "

The ones in the front make Eugene Ormandy sound very nice too.

Both models are keepers for shure. Both models are equally painful to move. Also about the same price range. Both belong to my brother.

Klaus

[Ducatista47]

After analyzing the room from several positions using the gains on the 4-way crossover to get the system as flat as possible, I then put the BSS 960 into the chain and did the final adjustments. I have always found the need for fairly radical (+- 6/10 db) of adjustment in some lower freq due to room anomalies.

Don't get me wrong, it sounds great after eq'ing but this little experiment proves the point that every active piece of equipment is in effect another preamp with its own set of characteristics/anomalies.

In short it seems to reinforce the straight wire theory.

KenI think Ken does have something there, it seems to work out very well for me, and it is nice to see recognition that there is more than one way to audio nirvana. I didn't mean to denigrate any other approaches. Let's just say I am not a fan of ruler flat frequency response being nearly as important as other qualities of musical sound reproduction, for instance naturalness of tone quality. I feel flat response is seldom or never as pleasing as other schemes, at least while listening to music for pleasure. Which is what I do, not being an audio professional.

Am I prejudiced toward analog signal chains and their results? You bet! Do I hold "less is more" in audio electronics as an article of faith? Until converted to another approach by my personal experience, yes. If I had time and money, I would surely experiment with this idea Ian's posts turned me on to:http://www.diyaudio.com/forums/showthread.php?threadid=42259&perpage=10&highlight=&pagenumber=1 (http://www.diyaudio.com/forums/showthread.php?threadid=42259&perpage=10&highlight=&pagenumber=1)
Looks great on paper, but what would it sound like driving a big monitor's biamped top?

In the end, when a beautiful, simple audio chain plays, it sounds sweet. Change and/or add elements and the sound is different. But if it is still sweet, if you like the way your rig sounds, you can't lose! Whatever works is good...

Clark in Peoria

[Ken Pachkowsky]

I think your opinion is logical and probably valid.

Ken

[boputnam]

Let's just say I am not a fan of ruler flat frequency response being nearly as important as other qualities of musical sound reproduction, for instance naturalness of tone quality. I feel flat response is seldom or never as pleasing as other schemes, at least while listening to music for pleasure. Which is what I do, not being an audio professional.Cool, and as Ken put, a very valid viewpoint.

I too had a preference for a certain EQ - it was not "flat" and was bumped here, and notched there (not on any EQ, but in the response curve). Over the past 30yrs doing live sound I did best I could, by ear. The recent 10yrs have provided affordable tools to measure the acoustic response of the room. Being quizzical, I decided to embark on trying to understanding this information - and even more, to "go flat" in my home system for a long enough period and see if I can adjust my hearing preference. That is, unlearn my preference for my personal EQ. I succeeded.

I've found that flat is the best opportunity to recover the tonality and "sound" originally intended, be it live or recorded. Exactly what you describe, Clark. Why should the speaker cabinet and room characteristics be allowed to wreak random acoustic havoc over what the artists intend(ed)? As an example, I've found that LF is a band of frequent abuse - "if it thumps you it is good" - and tonality is so often lacking (but can be "recovered" but some focussed and subtle EQ'ing). I strive to get every string on the bass to have it's acapella (and preamp) character. Too many bass players end-up being merely an extension of the kick drum - shame. This is not necessary if you get the response as flat ("honest" I call it) as possible, removing room resonance as-is-possible.

Flat saves greatly in required amp power - amps work hard to sustain EQ humps. Preamps can clip in the humped bands, which leads to subtle distortion - you many not see the outputs clipping, but there is clipping upstream in the signal path.

If you have the means to make quality measurement and adjustment, give flat a trial for a month. It takes the ears a while to re-groove. You might find the results more overall pleasing, discover subtle tonality you might have been masking and the overall sound less tiring. Just my 2¢

But, this is an unending point of discussion...

[Robh3606]

Flat response??

Now there's a topic for discussion if ever there was one. I have always been in the "Flat Response" camp. Not to the +/_ 1 extremes just reasonably flat so if you are using EQ you don't go to far with it. I don't get too upset if the curve has a wobble here of there, I think matching the stereo pair is really just as important and goes a long way to stabilize the imaging at the listeners seat as well. It also ensures they each are the same tonally so pans don't sound odd or certain notes don't pull to one side from irregularities between the pair. Even if it's not smooth as long as the response contours match between the stereo pair your on the right road. I like it flat from say 100Hz up to say 10K. I like a bit of boost below 40Hz and some roll off above 10K.

Rob:)

[boputnam]

Ha! This thread will now never die...

I thought of another point: in-addition to all the other attributes of flat EQ, the most important is that it is non-personal. This is particularly important in live applications - it will please almost everybody all of the time, and will be least tiring. Pay attention to the A-weighted SPL curve and beware of humping that part of the curve...

Speaker manufacturers go to great lengths to produce a flat response. It only makes sense that we users work on the acoustic response to try and achieve that.

[alskinner]

When tailoring the sound for live audiences, I agree that most people will be pleased with the sound of a flat response. On the other hand when designing a system for a paticular person, other factors such as hearing losses come in to play. Most of us (especially us older folks) don't have ruler flat hearing. Also, most recording companies don't record to a ruler flat standard. As far as I know the only way to insure speaker response is truly flat is to sweep the entire audio frequency spectrum using a standardized tone generator insuring a flat response in all drivers of the speaker system. Even doing this may not be the best sounding system depending on the anamolies of the drivers.


AL:blink:

[boputnam]

As far as I know the only way to insure speaker response is truly flat is to sweep the entire audio frequency spectrum using a standardized tone generator insuring a flat response in all drivers of the speaker system. Good points, and yes, Sir.

I've done that for my mains (and monitors/wedges) and fitted parametric settings to their characteristics (essentially done same-same for the home system). I also run Pink Noise at every event, and run an FFT function during live shows with real-time data (provides the same sort of acoustic feedback loop as measuring Pink Noise). It is critical for me to do so, to ensure that I am not "biasing" the response to personal preference especially as my hearing fatigues during a show, or due to age. Being careful with SPL and paying attention to the realities of the A-weighted SPL curve are paying me large bonuses.

Graphic from: Henderson, PD, Spectrum-mode Measurements with SmaartLive: Concepts and Applications (cf http://www.siasoft.com/pdf/SmaartLive_Spectrum_Mode_Measurements.pdf)

[Ken Pachkowsky]

Flat response??

Now there's a topic for discussion if ever there was one. I have always been in the "Flat Response" camp.

Rob:)

Yes Rob, you and I both. I try to get as close to flat as possible with the limitation's of my hardware combined with room anomolies. It just sounds better. An interesting thing to do would be to run some pink noise measurements with the dac plugged directly into the crossover.

Ken

[Mr. Widget]

On the subject of "flat frequency response" there are many ideas of what that means. For the most part loudspeakers are designed with the goal of achieving some sort of approximation of a flat frequency response in an anechoic chamber. Here is a plot from JBL of an actual anechoic measurement of the 4345. It is obviously not ruler flat... no speaker is, and horn based systems tend to depart from "flat" more than most. Anyway, the design goal is to arrive at a flat curve in the anechoic chamber... this is a standard. Once the "perfect loudspeaker with ruler flat response" is placed in a normal room there will be some degree of room gain on the low end and HF fall off with distance from the speaker that is expected. The actual in room response will not be as flat as the anechoic response was and measuring it isn't really very easily achieved... the closest approximation would be to take a dozen or so measurements at different locations and average them. In any event if you take your "perfect" loudspeaker and place it in your room you may or may not have a sound that is to your liking. At this point the room and your associated equipment are your equalizer...

Of course real speakers are all over the map as far as their frequency response curves go... some have bass bumps and zinggy highs like the L65 and others have a smaller bass bump with midrange presence bump like the L100 and some of the Altecs... the problem with deviating from flat is that a speaker with a strong sonic signature will make some music sound fantastic and other types sound exaggerated in some way... I suppose if the vast majority of your music has a certain sound and you have found a speaker that showcases that sound then your job is done.

Personally I like to hear a rather neutral sound that doesn't exaggerate... I admit at times it can sound less spectacular than say an A-7 playing Herb Albert, but I can set up my digital EQ with several distinct curves so I can have different "house curves" for different recordings... to date I really haven't played around with this much. As Bo had mentioned... listening to a flatter (less flattering... sorry about the pun:D) curve allows you to retrain your ears.

My "flat" speakers are not "ruler flat" They are as close to being a flat line as I could get without excessive external EQ... I am only using 3 bands of parametric EQ with no more than + or - 5dB on any of the bands... I can do this because I have a relatively dry room with very linear individual transducers. The curve I have has been subjectively tilted as a relatively straight line tipping down about 10dB from a high at 25Hz and a low at 20KHz. As I said this was derived at subjectively... I could adjust my system to be truly flat... and have tried it, it sounds very forward and bright. Having a flat curve that tips down sounds more natural to me and most of those that have auditioned the system... Having a nearly straight curve albeit a tilted one provides for a more natural presentation of overtones and harmonics... Zilch was over the other day and noticed how "realistic the piano's harmonics sounded" I attribute this to not having any significant peaks or dips along the curve.

As I said my curve tips down a bit and that was derived at subjectively... is it a coloration? I suppose. Every system has some coloration... however most mixdown studios use loudspeakers and have their own "house curve"... I think a survey of major mixing studios would show a downward tilting curve is the norm.


Widget

[Zilch]

Slope ~1 dB/octave like Stereophile's "exemplary" K2 (below). Full range is 10 octaves, so ~10 dB "tilt," total.

Flat HF's too "hot" for me; like Rob, I let the UHF roll off a bit, too (bottom).

[Zilch]

Of interest, in the June, 2006 issue of AES Journal:




Loudspeakers and Rooms for Sound

Reproduction—A Scientific Review*




FLOYD E. TOOLE

Harman International Industries, Inc., Northridge, CA 91329, USA






Much of our understanding of how listeners hear spatial acoustics originates from studies with large spaces, notably concert halls. While the resulting insights are frequently extended to small spaces, other phenomena weaken the validity of conclusions. Listeners may be adapting to the complexities of reflections and resonances in small spaces, thereby reducing or eliminating the perception of measured degradation. Informal evidence from the professional recording industry suggests that humans compensate and adapt to the acoustical anarchy of complex sound fields.

See, particularly, P. 468, Section 7.1.1 "Correlations between Subjective and Objective Domains," where:

"As subjectively interpreted, a smooth, flat, wide-band axial frequency response, combined with similarly well-behaved off-axis responses, up to and including sound power, appeared to be the desirable pattern."

Models based upon 1/3 octave data demonstrate a coefficient of 0.86, and newer models using 1/20 octave data achieve 0.995 - "near perfection."

"Single measures, such as the on-axis frequency response, sound-power response or steady-state in-room curves, are less reliable."

"So it seems that we truly are measuring quantities that are important to our subjective tastes. It is not an accident...."