BFD and boosting dips --> successful (1 Viewer)

[WarnerL]

Hi all,

I had posted my attempts at BFD equalization of my SVS sub last year, fooling around with house curves and such. I decided I wanted to try things again with a flat response in mind with my sub and front mains which I hadn't tried before (previously had tried a flat response with sub alone). My "before" response this time and also last time had what I assumed was a null at 111 Hz and this is what Wayne A. Pflughaupt had also mentioned to me but I realized that I hadn't even tried to fix this dip because I assumed it was a null.

I have looked at graphs of various other forum members here and other forums and noticed that some of them had an exact same dip at 111 Hz. I thought that maybe instead of it being a room null that maybe it was just the effect of the receiver's crossover and that the integration of the mains and subwoofer just wasn't working out as perfectly as the receiver wanted to.

It is my understanding that there are numerous variables such as the frequency crossover point itself as well the different slopes of the high pass and low pass filters around this crossover point (which are different with different receivers) as well as the variable of the frequency response of the sub and mains themselves. Since these numerous variables are present, I assume the manufacturers of these receivers could not even hope to get a perfect ideal integration of the sub and mains at the crossover point with all the different combinations of sub and mains out there.

Anyway, again since I noticed some other members' graphs having a similar dip at 111 Hz, maybe it was just an effect of the crossover by the receiver and not really a null. Even Wayne A. Pflughaupt had mentioned before that true nulls are usually of very narrow bandwidth (1/6 octave or less) and that dips present in the response may not be nulls but just what the frequency response happens to be or what I am suggesting, an effect from the crossover of the receiver.

My Yamaha receiver has a fixed crossover at 90 Hz and I am using HTD Level 3 bookshelf speakers as my mains (and surrounds) along with an SVS PCI 20-39 sub. Here are my readings and settings on the BFD and the resulting before after response of the sub and front mains together:







What do you all think? The so-called null at 111 Hz is gone so I assume it wasn't really a null or else I wouldn't be able to boost it. Isn't a true null resistant to all boosting?

For watching movies, I am running the bass about 4-5 dB hotter than the mains (using AVIA test tones) and I think it sounds pretty good even though I had equalized for a flat response and I know there is a debate about how a house curve may be needed for things to sound right. I was wondering though, since I am boosting my bass the 4-5 dB, isn't that somewhat similar to the effect the way a house curve is made, ie. a rising response in the bass frequencies? I know it is not as exact as the frequency response that someone may be trying to obtain with a certain house curve (where exactly the response starts to ramp up and where it levels off, etc.) but is this logic wrong in aiming for a flat response and just increasing the level of the bass to get a good sound?

 

[David Judah]

Well, you gotta go with what sounds best to you, but I don't cut anything below 30 Hz. That is interesting you got it to work out at 111 Hz. I would expect the dip to be lower a little in frequency if it was a crossover induced effect, but it is suspiciously close, so I guess it's possible, or it could be as you suggested--a FR anomaly with one of the speakers.

It looks good to me, but I wouldn't cut it as severely, actually not at all, at 20 Hz, especially since you got the woofage to handle it.

DJ

 

[WarnerL]

Why would you not cut anything at 20 Hz? Without that cut, wouldn't that area be out of whack compared with the rest of the frequency response? Also you said personally you don't cut anything below 30 Hz. Any particular reason?

 

[David Judah]

On a related note, I was just reading an interview of Tom Bohlender, president of Wisdom Audio(a line source speaker company), in Widescreen Review.

When they dial their speakers into a room they use what's called the Gundrey effect, named after a professor who did some research on frequency response preferences in the 1920's.

Basically, it calls for a 6-8 dB boost from 20-30 Hz sloping down ever so slightly to 100 Hz. It is flat from 500-1 kHz, down a dB or so from 1k-3k, down another dB from 3-6k, and sloping gently down until you are about 4 dB down at 20k.

They use a different type of equalization built in to their speakers, so I don't know if a traditional EQ would be sensitive enough for such small increments to matter on the high end, or for it to even matter because most speaker's FR is +/- 2.5 dB but I thought it was interesting and maybe would be something fun to experiment with.

Certainly on the low end from 20-30 Hz, I've found that it being a bit hot is preferrable, as has been discussed in the many EQ threads here, and it makes sense since we are less sensitive to that region based on how our hearing works.

DJ

 

[David Judah]

Why would you not cut anything at 20 Hz? Without that cut, wouldn't that area be out of whack compared with the rest of the frequency response?

Like I said above and has been discussed in other threads, what is flat on the meter isn't flat to our ears, and to many people not as subjectively pleasing. Now if it was really out of whack around 20 Hz, then some adjustment might be necessary, but I've never found that to be the case in any of my rooms or one's I've set-up.

DJ

 

[Bill Polley]

Congrats Warner. I EQed mine flat also and once I got used to the sound, I realized how much more realistic it sounded. I'm sure you are pleased with the resulting sound!

 

[dave alan]

One man's opinion:

If you take a blank graph and draw the crossover point/slope scenario, you'll see why the dip may have indeed been influenced by your particular setup.

All digital BM crossovers are based on a Linkwitz/Riley 24 dB/octave constant voltage network (also known as a dual cascaded Butterworth network).

Since you are applying a low pass filter to the sub, it's natural order of roll off is not effected, so the filter is 4th order, or 24 dB/octave.

Thanks to THX, the high pass filter is 12 dB/octave. This is because the high pass filter of the BM crossover DOES effect the natural order of the mains by cutting it short at the set crossover point.

TH based the 12 dB/octave HP slope assuming that all mains would be of the sealed variety, which have a 2nd order natural roll off, which, when affected by a 2nd order HP filter slope, would result in a 4th order (24 dB/octave) slope that would then match the 4th order LP slope affecting the sub and, voila... a L/R 4th order crossover.

Trouble is, your HTD mains are ported (and so are most all of our mains ported) which roll off naturally at a 4th order rate. When the 2nd order HP filter is applied, the result is a 6th order slope overlapping the 4th order slope of the sub.

What happens then is that your selected crossover point is shifted up approx 1/12th octave. And, since the HP slope is steeper than the LP slope, there is no longer unity gain in the summing of both across the crossover region, but rather, a 'hole' is created above the selected crossover point.

I tested this theory by cascading a 2nd order LP filter with the 4th order filter applied by my pre/pro, at the same LP point and ran sweeps. First, I shut off the sub and graphed the mains response. Then, I shut off the mains and graphed the sub's response. Last, I turned both on and graphed the final resulting FR curve. It looked almost textbook, other than some minor room gain influences.

It indeed flattened the hole I was previously seeing at 90ish Hz (my crossover point is 80 Hz), this cascaded LP filter being the only difference. No EQ, etc.

Maybe it was coincidence, but I think not.

If you calibrate your sub 5 dB hot, then you are listening to an average end result of up to 20 dB distortion of a flat curve below 30 Hz. Though you can certainly get used to that curve and like the sound of it, it is not accurate listening by any stretch.

If sound mixers/engineers wore earplugs and mixed by looking at analyzers and meters, then the distorted curve might be a good argument, but they don't. They mix it by listening. The meters are only a reference to parameters set by the format's limitations. Thus, the best producers being referred to as having 'golden ears'. The 'house curve' is already there, more or less (as many producers and mastering producers have bronze, steel, aluminum, or even tin ears).

For years, I had to fight with a mixing engineer, then the mastering engineer to be sure my bass sounded right, only to have a friend play it on his system, grossly distorting my hard work. Sort of like adding some color to a painting you bought because you think it will look better with more green.

One day, discs will come with an on screen graph of the music as it plays, as it was mixed originally, and a second graph of how you are listening to it picked up by a mic at the LP, in real time. Then, aftermarket EQ/X dB boost will make sense to me.

To each his own, I guess, but if you like your bass very 'hot', just don't say how 'seamlessly' it blends, or how 'accurate' it is (my pet peeve, sorry).

Very cool post Warner, and thanks very much for the info.('')

 

[Paul_Ptaaty]

Thanks Warner, appreciate the time to put together that post!!

I think there are cases to cut the output under 30Hz. For some reason, my PB2+ in my room continues to rise, and I am up about 10 dB from where it was on the meter from 40 to 20 Hz. What this does is rather cool and geewhiz but not accurate. I have a BFD as well and can do what I want to my curve. I guess I will be taking the house curve mostly out. When my house literally is shaking (the floor feels like it is dropping down...) on too many scenes it is too much.

 

[David Judah]

Good post, Dave. I never really considered a total 6th order slope due to ported speakers, but it does make sense when you break it down like that with the resulting notch above the crossover point.

IMO, and it's just that, accuracy to the original is such an elusive goal that I don't feel it's necessary to be constrained by it, especially if we are doing it just because it's supposed to be "right" on a graph and goes against what could be our natural(by way of our hearing system)preference.

When I was fiddling with my BFD and doing comparisons, I couldn't get around the fact that I preferred the rising response below 30 Hz, despite my original goal for a flat curve similiar to Warner's. After reading about the Gundrey effect, it makes more sense, because I don't think it was coincidence that over 90% of the people in his tests preferred it that way also. It makes a good case for it being tied to our nature of hearing.

DJ

 

[Brad Russell]

I have been wanting to post my graph and settings to get peoples opinion about a peak I have but haven't fiqured out how. How are you guys getting your graphs into your post?

Thanks!

Brad

 

[ChuckRG]

The dip at 111HZ and the bump you put into your BFD to offset it (15db), may look OK on paper, but I would do some serious listening to vocals before I ever settled with that. Narrow band peaks in EQ of that amplitude (15db) can be VERY audible if you move your head even slightly. At that particular frequency, you will hear GROSS anomalies in voices, that will make the voice very husky—and very incorrect.

On the other hand, if you are OK with it, enjoy.

 

[MingL]

When I do my BFD'ing, I tend to finish the tune-up with a simple hearing test. I find that measuring can sometimes be flawed if the technique is wrong; fast sweep, slow sweep and noise can all measure differently. So thus I prefer using my ears to be the final judge on a simple test sweep tone.

I would run a amplitude fixed slow sweep across the suspicious frequency ranges.

I should hear this sweep sweeping downwards at relatively the constant amplitude to my ears. I avoid using the meter for this test as our ears can hear things differently from what the meter registers.

If I hear something is wrong, then its worth revisting that suspicious freq range. If I can't hear anything abnormal (ie sharp changes in amplitude as the sweep progresses)), then its not worth making a mountain outta a molehill.

 

[WarnerL]

MingL,

I followed your advice with using a fixed amplitude slow frequency sweep (the low frequency sweep in AVIA, actually) with the 111 Hz and 124 Hz filters turned off. I could easily hear a dip in the frequency response at around the 111 Hz area. When I added the two filters back in, and repeated the frequency sweep it sounded much more even so I believe I was correct in adding those filters. I also repositioned myself at one of the ends of the sofa instead of in the middle where I normally sit and redid the sweep with all my original filters intact and it still sounded relatively smooth and not boosted at all in that same frequency range.

This leads me to believe in what Dave Alan was talking about in his reply about the resulting 6th order slope overlapping 4th order slope of the sub thereby creating a "hole" in the frequency response above the selected crossover frequency.