getting Cara'd away

And now for something completely different, or at least, something I'd meant to for a while - playing with a shiny new SPL meter, the Cara acoustic simulation suite and a test-tone CD.

to the graph (23KB) Method
  • Acquire SPL meter, test cd, tape measure, masking tape.
  • Mark a grid on the floor
  • Spend a lot of time shuffling the speakers about and entering the measurements in Excel
  • Profit!
  • The session started with the speakers with their backs against the wall (Impulse H5s, 400mm deep) and the meter on an old camera tripod on the couch. Keeping Nait volume setting constant (about 9am mark), I measured level with broadband pink noise as a reference, then run through the test tone trcks (1/8th octave intervals from 20Hz up to 200Hz). Then we pulled the speakers forward 100mm and repeated the exercise. And repeated... it took about 2hrs in total. Finally I compared what I measured with the Cara model I've put a few hours into, plonking the speakers where it suggested, and measured again.

    This method gives quite a good view of how the system output varies solely due to the speakers position, but it is NOT entirely accurate. The biggest flaw in this scheme is that, actually, an SPL meter is not suited to this task - it is designed to integrate over a broadband, not to measure at spot frequencies, for which the design and polar response of the microphone becomes rather significant. So treat the results as a relative rather than absolute indicator. Still, the results are rather interesting to listen to - you'll all just have to rely on my measurements for this ;-)

    The long and the short of it - the turquoise trace is where the speakers have lived for the last six months, about the best 'by ear+ rhythym method' compromise so far. Blue trace is the 'smoothest' alternative found during the afternoon of speaker shuffling. Now the red trace is the result of running the Cara model for positional optimisation over the range 20-800Hz (remember that location is pretty much irrelevant once into the midrange). The results of which were uncanny - speaker distance to sidewalls, and listener position all came with 50mm of where they were already! Cara predicted smoothest response in chosen range with the Impulse's front baffles 104cm from rear wall, so I plonked them there and measured again - and that gave the red trace.

    Quite a difference, no?

    Basically the system now appears to measure 'flat'(i.e. within 6dB) down to 32Hz, below which it falls off at 24dB/octave (as it should for what is essentially a horn-loaded box). To me it sounds like the bass falls off a bit higher up, which only goes to show how insensitive the ear really is at LF. NB the average of 83dB is pretty darn loud...

    The transition from bass to low midrange both measures (and sounds) very, very smooth now; compare with some of the alternative locations. Note the hump in the middle of the plot is due to low-order room modes boosting levels - 74hz is 2nd order mode across room, 59Hz is 2nd order along room and I can't do much about them at all. Otherwise the curve above 93hz would neatly meet the curve below 50Hz. Oh well, have to settle for some excess 'slam' while investigating what space I can give up to bass traps ;-)

    Most importantly: the sound is much better; notably bass is much tighter in quality, though still fairly generous in quantity(!). Differentiation of pitch down low is much improved, and percussion a real treat. There remains a verymild colouration to voices that seems to go with the cabinet design, but I willingly overlook that for the Impulses' sheer vitality of presentation and dynamics. I can also live with the monoliths that far out from the wall; my only 'problem' is that the soundstage 'image' is alarmingly good, positively distracting infact. Nevermind, so is the groove ;-)

    Things learned:

  • You can - and should - trust your ears :-)
  • Cara is a verifiable tool that does have a place in the Flat Earth pantheon & I mean to do an extended write-up on this in the near-future.
  • Proximity to wall affects the low midrange more than the deep bass - and ripples here have a fundamental impact on how 'right' things sound - to me, anyway.
  • Listening to music is a damn sight more interesting than test tones.
  • It's possible to peak in excess of 108dB at 2m with a Nait2...

    Please let us know how things turn out if you try any or all of the above.

    © the twisted pair 2004

  • 25.04.04 Page added