Measuring Frequency Response

Discussion in 'Amps and Cabs [BG]' started by ceycey, Jan 2, 2005.

  1. ceycey


    Dec 22, 2004
    I saw that when measuring the frequency response of the cabinets all companies measure at different dB levels.
    Ampeg measures at -3dB, GK measures at +5dB,...

    What is the difference of measuring these intervals at different dB levels?
  2. unrealrocks


    Jan 8, 2004
    That isn't at what level its measured, that is at what point the response falls below that.

    So the response of a cab could be 30-300Hz -3db ... this means that before 30Hz and after 300Hz the effective sound output is half of the referance input.
  3. ceycey


    Dec 22, 2004
    I still didn't understand it.

    So what will be a better interval when considering to buy for standalone usage?
    42-18khz at -3dB
    36-19khz at +5dB

  4. jdagger


    Nov 21, 2004
    Hi, there is a thread in this forum where billfitzmaurice discusses frequency response. If not the thread that prompted your question, it will help you to understand a bit more what is going on. See replies #10, #14, and #17 for more information.

    Frequency Response

    Good Luck and write back if you need more explaination on what is said in that thread.
  5. IvanMike

    IvanMike Player Characters fear me... Staff Member Supporting Member

    Nov 10, 2002
    Middletown CT, USA
    yeah bill pretty much sums it up. seeing a frequency response spec without tolerances is useless. Generally you shoudl see something like blah Hz - Blah KHz +/-3dB. That lets you know that between the two extremes of frequecy quoted, the cabinet should only vary by 3 dB up or down from the "average" Spl of the cabinet. Still, seeing a chart is better, but you wont always find those. a 3dB tolerance either way is really good for a bass cabinet, but that still means that some frequencies will be 6 dB higher than others within that range.
  6. Jerrold Tiers

    Jerrold Tiers

    Nov 14, 2003
    St Louis
    It can be even more complicated, since there are ways of measuring the response, which differ significantly.

    You typically don't know how the measurement was done, or even IF a real measurement was done. You don't know if the spec is valid, or bogus.

    So you don't and often can't know if a +- 3 dB spec is REALLY better than a +-6 dB spec.

    Two method types:

    1) Frequency sweep, such as TEF, or other methods which involve measuring a sine wave at many points in the range.

    2) Band measurements, usually with a random noise signal. The response is typically measured by 1/3 or 1/10 octave bands, averaging all the signals over that band.

    The resolution of the sweep method is better. It theoretically should find any and every peak, dip, "wolf tone", etc, depending on exactly how the measurement is done.

    But, if anyone states a frequency response measured by sine wave sweep and given +- 2 dB, they are probably fooling themselves (and you). Speakers of normal quality are simply not that well behaved, as a rule. It can be done, but is expensive.

    Also, for low frequency measurements, the measurement area must be large and satisfy various acoustic requirements, even for the TEF or similar less restrictive methods.

    The band method is simpler to do. But, it fails to distinguish narrow peaks and dips that might sound awful. The narrower the bandwidth, the better, 1/10 octave is "better" than 1/3 octave for resolving narrow peaks etc.

    The band method also gives much nicer numbers, if you want to give impressive tight response specs. You can give a +-2 dB spec, and hit it, even if you have narrower peaks or dips that may be 6 or 10 dB high. So it can make a "bad" speaker look good in the specs.

    And, you can do the measurements in an area which does not necessarily meet the strict requirements for the sweep method. That is because some errors are averaged out, and the last low octave can be "carried" in the specs by its upper end, even if the lower portion drops off fast.

    In fact, with a "coarse" band measurement, a speaker that peaks and then drops like a stone below the peak can actually measure flatter than one with a smooth rolloff and actual response to lower frequencies. With narrower bands, that nonsense is reduced.

    Ain't specs fun? Too bad they often don't mean squat.

    It's best to play through the speaker and see if it does what you want. If you can gig-test it, so much the better. Then you "know".

    Youn can't hear's about sound.

    (hey, I like that, new sig time.....)
  7. metron

    metron Fluffy does not agree

    Sep 12, 2003
    Lakewood Colorado
    If they are measured in the same way then -3 dB is always a better spec to go by but it wont look better. -6 dB will make it look like your speaker has a wider response but there isnt much output at -6 dB.
  8. metron

    metron Fluffy does not agree

    Sep 12, 2003
    Lakewood Colorado
    This +5 dB number is probably wrong (its likely -5 dB) but its only serving to confuse you. Use the specs given at - 3 dB. Thats what most manufacturers use and is a standard for frequency response specs. That way you can roughly compare speaker FR of different brands if you want to. Dont worry so much about how the values are measured. Of course the best way is to test with your ears.
  9. It sucks that there are no standards adhered to. In stereo geekspeak, the low end limit of a loudspeaker is usually 3 db.

    If you are like me, you have a band mate play your bass while you are setting up and stand in different places in the room. I use the "wall shake test" and bar shake test and booty shake test, the low end frequecy response is acceptable. It should be a pass/fail test in the real world. Of course, your intonation, definition and articulation needs to be heard too.
  10. Jerrold Tiers

    Jerrold Tiers

    Nov 14, 2003
    St Louis

    Of course........but it's often a BIG "if".

    The method is generally not given. We are as guilty as anyone else, unfortunately.

    BTW, somewhere between 6 and 10 dB is the difference in SPL that SOUNDS "half" as loud, or "twice" as loud. So it isn't quite SO silly to quote -6 specs for speakers.

    For amps it would be another thing altogether, the measurement process is pretty trivial.
  11. metron

    metron Fluffy does not agree

    Sep 12, 2003
    Lakewood Colorado
    True however relative to the input -3 dB is half of the signal, -6 dB is half of that and -9 dB half again. Running out of signal here... The main reason why its not so useful is because most quote a -3 dB point. Its not really useful to compare one with -6 and another with -3 or +5. :D
  12. brianrost

    brianrost Gold Supporting Member

    Apr 26, 2000
    Boston, Taxachusetts
    Nope, it means that it's already 3 dB less at 30Hz than it is at some point in the midrange (like 1KHz).

    After all, speakers are great at pumping out mids but not super lows, so you have to expect that response at the lowest frequency will be weaker than in the midband.

    It doesn't tell you anything about response BELOW 30 Hz. The missing info is what the lowest usable frequency is (i.e. if you use an EQ boost and some extra wattage maybe the speaker can cleanly reproduce stuff down to 20 Hz or lower without damage from overexcursion...or maybe NOT).

    I'll say it again: people worry TOO MUCH about these specs. Use your ears. For most situations reproducing the full fundamental is a waste of time and cabinets that have weak fundamentals but hype the 100-250 Hz band will sound "fatter" to most people anyway. Unless you only play your open B string all the time, response at 30 Hz aint of much interest.

    Waiting for the heated replies from all the fundamental lovers :p
  13. I agree with the above. The Bill Fitzmaurice stuff about the harmonic energy being concentrated in the midbass ocatves makes sense when you think about it. I still want to know the low end limit of any ported cab because I want to know what it is tuned to. I really wouldn't want to waste 70% of my precious wattage trying to reproduce a 30 hz fundemental anyway. Hopefully, my amp has a low frequency cutoff filter somewhere.

    I only have one minor critique of the Bill Fitzmaurice posts. They all have such a grumpy and disparaging tone. I mean, its just acoustics, electronics and physics stuff, nothing to get bummed out about, ya know? But whatever, man. That's just his style.
  14. ceycey


    Dec 22, 2004
    Thats what i understand from i read;

    We can't trust any value given in the specifications chart. SPL values are calculated according to different measurement standarts. We can see how it changes in the link: SPL Calculator .

    Also frequency response is important on calculating the SPL value. So the link is useless. I saw billfitzmaurice calculate it by using that responses;
    "If the f3 of the 93dB box was 40 Hz while the f3 of the 97dB box was 80 Hz, for instance, then the sensitivity of the 93 dB box at 40 Hz would be 90dB, while the sensitivity of the 97dB box at 40 Hz would be 82dB if it was a sealed cabinet, and 70dB if it was a vented box."
    But how can he calculate this without knowing the -dB written in the frequency response. Thats what i don't understand.

    The dB's written in frequency response means when trying to get any other frequencies that are not on the interval, it's how much dB the speaker will lose. For example: If it's rated at 36-18khz at -5db when we try to get 30hz the cab will lose -5dB of loudness. It's same if we try to get 20khz. Again it will lose -5dB. As brainrost says companies should write a usefull lowest and highest frequency responses when giving the specifications. I just saw this info on Ampeg.

    Please correct me if i'm wrong anywhere.
  15. IvanMike

    IvanMike Player Characters fear me... Staff Member Supporting Member

    Nov 10, 2002
    Middletown CT, USA
    nope, the dB indicated in frequency response indicate the frequencies within the specified frequency range. Lookin gat the specs above, the cab could be -10 dB @ 30HZ.

    But as mentioned, who cares. ears rock. your ears will tell you if a cab is a peice of crap pretty fast.
  16. Tim__x


    Aug 13, 2002
    Alberta, Canada
    For a given alignment the response of a cab is fairly predictable, ported lose 24db per an octave below tuning (tuning is usually the same as F3), sealed lose 12db per an octave below F3.

    No, the response drops off lower and lower as you get farther and farther away from from the passband. Passband being the range of frequencys the cab will produce well.
  17. ceycey


    Dec 22, 2004
    Now it makes sense.
    Thanks everybody.