Friday, December 27, 2013

New Crossover and EQ for the Living Room System

I spent much time over the 2-day pre-Christmas weekend, and the 2-day Christmas Eve and Christmas Day holiday I also enjoyed, adjusting the Behringer DCX 2496 that serves as a digital crossover, eq, and time alignment device for the living room system.

When I started, I was thinking the sound was a bit thin.  The adjustments I made added solidity to the sound without adding too much bass heaviness, punch without bloat, impact without boom.  It takes skill and patience to do this.

In the process of adjusting, I used several tools.  I used my Kurzweil keyboard, which I temporarily moved in front of the listening position.  I created an instrument (Kurzweil calls any configured sound a "program") that's just a pure sine wave, extending as low as 16 Hz.  I started with the Default Program 199 and just changed "piano" to "sine wave", then I lowered the notes by two octaves.  Being able to play sine wave bass tones makes it easier to hear which ones need to be lowered and which ones needed to be raised.

I also used my B&K oscillator, which has digital readout.  Actually, this little oscillator is not as easy to use as it could be.  I choose a 100Hz range which actually gives me 10Hz to 200 Hz, with 100Hz in the middle.  But adjusting to specific frequencies below 50Hz gets harder and harder, as only tiny changes of the knob can scroll past a bunch of frequencies.  It works well enough to do slow sweeps to pick out problems.  But it might even have been better if I had used the Kurzweil and assigned a slider to do frequency adjustment.

I did not use any level measuring instruments.  I used my own ears, either at the listening position, or in other locations.

Let me run down the changes I made (and not elaborate all the history of what I tried to do, which would take too much time to describe).  I'll just give a few historical observations and/or rationalizations.

1.  I changed the crossover between the subwoofer and the Acoustat panels to 24 dB/octave Linkwitz-Riley at 80 Hz on both sides.  I am using this crossover with no polarity inversion on either side, and it is a nice property of the 24dB/octave linkwitz-riley that it does not require any polarity inversion.

Previously I had been using the crossover in a more ad hoc fashion.  I had been crossing over the bass side with 48dB/octave linkwitz riley, but left the highs with a gentler 24dB/octave slope.  My reasoning was that the subwoofer had best not be producing any mid bass since the panels do it so much better.  Well doing things this way there seemed to be a peak in the 82Hz region.  So I then also separated the crossover frequencies so that the panels were being crossed over at 88 Hz, and the bass around 80 Hz or maybe even lower.  This sort of worked.  But when I changed both sides to LR24, the peak at 82 Hz went away, and generally sound was smoother on both sides it seemed.  My ad hociness was requiring more ad hociness to fix.  Previously I didn't take the crossover settings very seriously, on the grounds that the speaker drivers are introducing so much additional frequency-related variation any crossover setting is really only an approximation.  It seems now that I should have been taking the actual crossover settings more seriously, especially when I was not seriously analyzing what the additional variation was that needed to be corrected anyway.  Given that you don't really know what else needs to be done, the nominal crossover is a good place to start.

2.  I eliminated the frequency contouring I was using for the super tweeters.  I removed the lowpass above 20kHz, which was rolling them back down again on the high side.  Now the super tweeters are crossed over at 15.5kHz, and that is it, they then continue on as high as their response (or more likely, the signal source) allows.   The effect of removing the lowpass was that the output level increased slightly, but I could not hear it as such, I just occasionally see a level indicator bar on the Behringer when previously I would see none.  I tried reducing the levels of the super tweeters, but ended up back where I had set them before, with the left channel only reduced 2db because it seemed to have about 2dB more output.  I confess the level settings make little sense to me and no measurements I've done have clarified the matter at all.  I would think the level settings should be much lower, but making it lower and the super tweeters lose the magic.  BTW, the magic is not at all brightness.  It's actually a kind of smoothness, where grain and grit go away.  As you increase the the level of the super tweeters (which actually have very little output that I can hear directly, as one would expect with crossover at 15.5kHz) the sound just gets smoother and more palpably real.  I just quit increasing the level because turning it up even more seems--to my mind--obscene.  I wish I was better able to do high frequency measurements to determine where the super tweeter level should be set.  But ultimately it comes down to what feels right anyway.

3.  I changed the bass EQ's.  I retained the huge notch at 45 Hz with -11dB.  I tried changing that big notch also, but whenever I flatten that notch, even to a very similar -9dB, the sound goes to boomy fast, without any improvement in bass tunefulness.  But many other bass eq's are changed, and many are new.  One thing that's also new is that I eq'd each channel differently, depending on several factors, both the pre-existing response, and the additional level that could be handled without causing distortion.

Notably, compared to before, I added various degrees of boost at or below 32 Hz.  That's because just just below 34 Hz or so the bass was tending to sound very weak overall, especially compared with the boom in the range 36-48Hz.  I specifically added a boost at 32Hz in both channels, since 32Hz is often found in music, and it needed boosting a lot (probably more than I actually boosted it).  Then I boosted tiny amounts at 27hz, 25Hz, and 20Hz.  The right channel starts to buzz when I boosted it for the lowest two frequencies, so i did that boosting on the left channel only.  Then, corresponding, I left out the 27Hz from the left channel, since it was already boosting 25 hz.  These are little 2db boosts, but they help.  The 20Hz and 32 Hz boosts are fairly narrow around 1/3 octave, but the 25Hz boost is a wider 1/2 octave or so.

I left the 45Hz notch alone in the end, but that was not without trying many other variations, such as moving it to 44Hz and then moving another notch at 57 Hz down to 48 Hz.  No matter what else I did, the 45Hz notch seems to be needed, so I ended up keeping it.  But the 57 Hz cut has been changed depending on the needs of that channel.  On the right side, there is more boominess above the 45 Hz notch, so it now gets notched out at 48 Hz.  On the left side, there is more boominess in the upper 30's, so it gets a notch at 39 Hz.  With the full LR24 crossover on both sides, there was no special boom that needed cutting in the upper 50's, so the 57 Hz cut is gone gone gone.

[I will be adding in here the last adjustments made in this time frame.]

No comments:

Post a Comment