EQ Question

Discussion in 'Audio Hardware' started by jtw, Jul 10, 2018.

  1. jtw

    jtw Forum Resident Thread Starter

    If I have a 9 band equalizer (either hardware or software), there is no difference in sound between setting the first 8 bands at 0 and the 9th at +8, vs having the first 8 bands at -8, and the 9th at 0, correct? After all, the scale is relative, and only represents the frequency range volumes relative to each other, right?
     
  2. Davey

    Davey very clever with maracas

    Location:
    SF Bay Area, USA
    I don't think the response would look the same, you would probably have some ripple between the bands, but it depends on the filter Q and slope. With the sliders at 0 there is no filtering so it should be a relatively flat response over the first 8 bands, but if you set two adjacent filters to -8, they have a lot of overlap and will not likely combine to a perfectly flat response throughout that bandwidth at all settings. You would also have a difference in the response outside the first and last band unless those sliders were implemented as shelving filters. And this ignores all the phase response changes being added to the signal that would probably lead to some big sound changes.
     
    Veni Vidi Vici, Kyhl, tmtomh and 2 others like this.
  3. timztunz

    timztunz Audioista

    Location:
    Texas and Brasil
    Wait, whaaaaat? Hahaha, makes me glad I don’t have an EQ and makes my head hurt.
     
    Lemon Curry and Dave like this.
  4. Uglyversal

    Uglyversal Forum Resident

    Location:
    Sydney
    That above plus you might also add component noise since you'll be effectively lowering the volume of the music while the electronic component noise will still be there at the same level, that at least would be valid for the latest stages as the early ones might be reduced when you go down in the scale of the pots. Either way I think what you are proposing is a bad idea and not the way to use an equaliser. You should aim for the center and if anything needs moving you should only move those pots.
     
    MusicNBeer, tmtomh and trd like this.
  5. jtw

    jtw Forum Resident Thread Starter

    Please understand that the user intends to listen at the same perceived volume, in both cases. In other words, the user will NOT be starting out at a certain volume, and in one case raising the volume of the 9th band, and in the other case, reducing the first 8 bands. Instead, the user will set the eq, then raise the volume to the desired level.
     
  6. Uglyversal

    Uglyversal Forum Resident

    Location:
    Sydney
    Yes I understood that, still all those notes apply regardless.
     
  7. jtw

    jtw Forum Resident Thread Starter

    Hmmm

     
  8. JohnO

    JohnO Forum Resident

    Location:
    Washington, DC
    What happens is, on a consumer level analog equalizer, if you move one slider it affects the adjacent bands also. For example if you have everything set flat then raise 1kHz by 8db, that will raise 500Hz by 4db and 2kHz by 4db too. Or 3db or 2db, depending on the quality of the design. And there is still some effect to the next band. Even if you move a most extreme slider it is affecting the adjacent frequency band somewhat. There is also a phase change - in every band affected - not absolute like "out of phase" speakers that are exactly reverse or "180 degrees" out of phase, but phase change measured in degrees out of 360 like 90 degrees, 75 degrees, 135 degrees. You can see and measure the phase change using a sine wave source on a scope, and you can hear it sometimes, but usually not all the time, but it is there.
     
  9. Uglyversal

    Uglyversal Forum Resident

    Location:
    Sydney
    You don't understand what I am talking about, I am referring to the signals travelling inside the components but Mmmm you could have requested clarification in a proper manner, so be it.
     
    trd likes this.
  10. jtw

    jtw Forum Resident Thread Starter

    This is the 'Q'? I think I get that. But if you remove the scales from the hardware equalizer or from the display on the screen, the relative position of the sliders is exactly the same in these 2 cases. So, in both cases, the lowest frequency ranges would be 'x', and at some point (the same point in both cases), the volume would ramp up so that it would be 'x+8dB'. No?
     
  11. tmtomh

    tmtomh Forum Resident

    What @Davey and @JohnO are saying is that the zero position on an EQ is unique in that it uses no filtering - it basically just lets the signal pass through. Any other position brings in a filter, which affects both phase and linearity. So by this way of thinking, having the most possible sliders at the zero point would be best. In your original example, therefore, it would be better to have 8 sliders at zero and one slider at +8, than to have 8 sliders at -8 and one slider at zero.

    Yes, the center frequency of each slider will be in the same position relative to the center frequency of each other slider either way - but as you move up and down from each slider's center frequency, you get deviations in amplitude and possibly phase - except if the slider is set at zero, in which case you don't get those deviations. That's the difference, and why it's generally best to arrange one's EQ curve in such a way as to have the most sliders as possible at or near the zero point.
     
    rtrt, G B Kuipers, Kyhl and 1 other person like this.
  12. VU Master

    VU Master Forum Resident

    Right, and the key word here is basically. There are a lot of things that will happen even with all the sliders set to zero. For starters, the zero points will never be 100% accurate. I would bet that there's at least 1/2 or 1 db of cut or boost going on at each band with those sliders (which are probably very cheap parts and may have a 20% tolerance spec) set to zero. There's probably phase shift going on too, and I bet that transient (square wave) response at the output of the EQ if fairly compromised compared to the input. I think a 9 band equalizer is much too coarse because the bands are so far apart and I wouldn't recommend using one for serious listening but if you do have one, it's much better to bypass it entirely when not in use. Less is more.
     
    trd likes this.
  13. head_unit

    head_unit Forum Resident

    Location:
    Los Angeles CA USA
    No. To re-phrase what was said above, let's compare 3 scenarios:
    1) All sliders to 0: response should be flat, if the circuit is ideal and the zero points correct blah blah
    2) All sliders set to maximum: the response will NOT be flat
    3) All sliders set to minimum: the response will NOT be flat
    Why? See this Equalizer sample for CSharp, Visual Basic and look at the figure below where it says "Physics background": at max or min, each filter looks like a cone shape, and they do not add flat, they will add looking like ocean waves or something. Or if you can load this massive thing All About Audio Equalization: Solutions and Frontiers look for "Figure 8. Magnitude responses of an octave graphic equalizer" where you can see that if all bands were at max, not only would be the response be more like +12 or something (and likely clipping/distorting) it is very uneven.

    Also, a system should never really need "+8" boost or cut, that means something is wrong. What are you trying to achieve? Or were you just curious?
     
    Veni Vidi Vici, Kyhl and Grant like this.
  14. tmtomh

    tmtomh Forum Resident

    I don't disagree with what you're saying, but what you're saying here is an argument for (as your final sentence notes) bypassing EQ entirely. No one is arguing that one should have an EQ in line and active when set to all zeros. The point of the OP's question is about when a curve is needed, which curve would be preferable. But of course yes, an EQ is not going to be purely neutral even with all sliders set at zero. There's still going to be active circuitry there unless the EQ is in bypass mode (and even in bypass, if the EQ is still turned on, the signal is going to be running through its passive components and in close proximity to its power supply, which could have a small impact on the sonic performance of one's system).

    Great explanation! I would respectfully disagree with your final statement, though. Room resonances can sometimes create large frequency response spikes or dips. And even with more modest frequency nonlinearities - based on speaker voicings, or speaker placement, or standing waves, or some combination of those - it can require a large EQ adjustment to effect a complete or near-complete adjustment of the actual frequencies in the room. In other words, I've seen 3-4dB peaks or dips as measured by an SPL meter, which took 6-9dB EQ slider adjustments to flatten out.

    I do agree, however, that room treatments and speaker placement options should of course be tried and exhausted before resorting to EQ (and that for one or two isolated frequency problems, a parametric EQ probably would be better than a regular graphic EQ). But sometimes people don't have living situations where they have a dedicated listening room without irregularities or problematic dimensions; and people can't always do all the necessary room corrections and so on. In those cases, an EQ can be valuable, even one where some large-ish adjustments are made.
     
  15. KT88

    KT88 Forum Resident

    I guess then, by your example, what you are trying to achieve is a high frequency boost? Whether the sliders go up or down a bit depends upon the EQ design and so its performance. I have tried cheaper, but decent EQ's and found them to add noise, even when flat, and surprisingly, even when in-line but bypassed using that button. Just going through the added connections and switches, maybe a buffer stage as well, added noise. If all you want is a HF boost, less frequency deviation should be experienced by raising just the band needed, especially if it's at the end of the scale as in your example. Some EQ have pretty crappy gain structures and can add noise more so when boosted than cut, so it has also sounded better to these ears many times to reduce bass rather than increase highs when needed. That's more of a gain symptom of their preamps than EQ though. Both can have consequences. As others are suggesting, there are likely alternatives to arrive at the results that you want rather than manipulation of the signal electronically. Acoustic solutions and speaker selection can be effective and not cause such side effects.
    -Bill
     
    Mike-48 and tmtomh like this.
  16. harby

    harby Forum Resident

    Location:
    Portland, OR, USA
    Here is a paper (pdf) from Rane, who patented the Constant-Q equalizer, and without doing deep reading or math, it shows some examples of how not only do several filters adjusted away from 0 not give a constant response, they have optimized filtering in a way that most EQs don't, so your performance will likely be worse.

    Start at 0, and apply the corrections only as needed. If a change in level is needed to avoid clipping, lower the level before the EQ and not by moving all the faders in that direction instead.

    Some digital EQs (notably ones with hundreds of bands) use FFT separation of the bands, which can be more linear, but also depend on windowing functions implemented properly to not affect adjacent bands with ripple.

    I have an AudioSource EQ-One ten-band equalizer with spectrum analyzer and pink noise generator, good technology for 1985. After coming up with a set of four RCA->XLR cables, I can demonstrate its sweep-tested frequency response with all sliders away from 0.
     
    tmtomh likes this.
  17. head_unit

    head_unit Forum Resident

    Location:
    Los Angeles CA USA
    All the things you said are true, though standing wave problems would be time domain related and not well solvable by analog EQ as The Grateful Dead found out in their early days (you can make the "average" measured steady state sound "better" but actually you have simply worsened the direct sound and then the reflected sounds as well). Hmm, I kind of feel like saying that any sharp notches-even from speaker driver/crossover-would be time related but that's just instinct off the top of my head. Audyssey and Dirac et al *should* be much better for fixing those type of problems...eh, well, maybe not the reflective cancellation...gotta think about that some more.
    Saying nothing should need a huge boost or cut was really directed more at novices, who will tend towards such because it sounds better. It does! Certainly it sounds more exciting which is better for a while, until maybe maybe they realize it doesn't really sound too good after all. +8 dB boost at the top band sounds like either the tweeter is shot, or not very good, or quite possibly it is time for a hearing test.
     
    tmtomh likes this.
  18. tmtomh

    tmtomh Forum Resident

    I could be wrong/naive, but yes - to me the noise added by many cheap(er) EQs is more of an issue than nonlinearity of the frequency-band filters. I'm guessing there are very expensive, super-low-noise EQ units out there somewhere, but in my experience the best units that are commonly available and affordable are older ones from a couple of decades ago, made by the somewhat better-quality brands. In particular I've found Yamaha units to add little if any noise to the signal. I'd also guess that Soundcraftsmen units would be very good in this regard, though likely quite expensive on the used market.
     
  19. jtw

    jtw Forum Resident Thread Starter

    Reading and digesting. In the mean time, I'll be using eq as a loudness button to try to make low level listening more enjoyable. Also, I'll be trying to compensate for diminished high frequency hearing.

    Very interesting that everything set flat at -6dB will sound different than everything flat at 0dB. I would have never guessed.

    I'll be using software eq, most likely.

    This discussion makes me wonder about the numerous times I've read that it is better to cut than to boost when using eq. It looks like that is wrong, correct? If someone wants to add bass, according to this discussion, they should boost bass instead of cutting the rest of the 'knobs'.
     
    The Pinhead and tmtomh like this.
  20. Davey

    Davey very clever with maracas

    Location:
    SF Bay Area, USA
    I think the concern with bass boost is over-driving the amp and speakers, but you should be safe at lower volumes.
     
  21. VU Master

    VU Master Forum Resident

    Yes, my reply was a little off topic in view of the OP's question, but I wanted to point out that having knobs and sliders set to zero doesn't usually mean there is no cut or boost going on -- and I don't think it's a small thing, I think it's a big thing. I service pro audio gear and even new, high quality equalizers going for $5000 or more can have significant discrepancies between what the front panel says and what the unit is actually doing. When we service units with setscrew type knobs we can sometimes reposition the knobs for true zero based on actual measurement but with sliders, that's not an option. The more cut and boost that is available on each band, the more error one can expect at the supposed neutral position. Sliders and knobs with mechanical detents at the center position are no better and sometimes they're actually worse, because the spring action of the detent position can make the control pop out of the true zero position, even if one finds it by measurement.

    To anyone with an equalizer and the right measurement equipment -- try measuring each band at zero. The results might surprise (and disappoint) you.
     
    tmtomh likes this.
  22. Hipper

    Hipper Forum Resident

    Location:
    Herts., England
    I would suggest the OP experiments to find if he can hear any noise or other anomalies that equalisers are
    said to produce.

    I use a Behringer DEQ2496 digital equaliser and can't hear any artefact at all. It's difficult to compare with the Behringer completely out of the system, as opposed to just in bypass mode, but I still can't hear any difference. Further, the benefits of careful EQ, especially in the bass region, are substantial.
     
    The Pinhead likes this.
  23. The Pinhead

    The Pinhead SLEAZY SOUTHAMERICAN CAVEMAN

    Yes; and if you need to alter any specific frequency, try the least boost or cut you can do with or you'll end up with a sonic mess. And this is coming from a guy who owns an uses a 10-band equalizer.
     
    tmtomh likes this.
  24. harby

    harby Forum Resident

    Location:
    Portland, OR, USA
    Here's the result of testing my own AudioSource EQ-One, using test tone sweep, and pushing the sliders to +10dB and -10dB with a ruler:
    [​IMG]

    Yellow is with the EQ in bypass mode, red is with the sliders all set to 0. Low frequency wiggles are from using a short sweep instead of testing for 10 minutes...

    The -10dB is not quite so flat, even though the sliders were set flat...
     
    Last edited: Jul 12, 2018
  25. The Pinhead

    The Pinhead SLEAZY SOUTHAMERICAN CAVEMAN

    Pioneer GR-777. Making me happy since 1993 :

    [​IMG]
     
    fogalu likes this.

Share This Page