TOPIC:

Hi all.

I have recently replaced my satellites+sub with full range speakers, and I am having some challenges getting Dirac to work as good as I had come to expect with my previous setup.

Both speakers are too close to the front wall, and one (the left) is too close to a corner. I think the corner one is causing some sort of cancellation that Dirac is unable to compensate for.

Measuring each speaker separately they have 10-15 db peaks around 60hz. Dirac compensates for this, but this results in a pretty bad dip at 90hz.

I am able to remedy most of the dip by adding a 5db filter at 90hz on the right speaker (PEQ, Gain 5, Q5).

So, my question is, am I doing something "wrong" by overriding Dirac in this way, and is there some other way I can force/instruct Dirac to handle this better?

(sorry if my terminology etc. above is way off, I am a fairly basic user).

Thank you for reading!

Red is before PEQ, purple after.

With the Dirac Live measurements, you are measuring in nine locations, whereas with the measurement you have posted you are (I assume, yes?) measuriing in just one location.

Correct. That measurement is equivalent to the Dirac initial measurement.

OK, so Dirac is making a correction based on nine measurements, and then you are overriding its correction based on just one measurement...

I see your point (I think), but how can I avoid the dip in sweetspot?

Edit: I tried using Dirac with only one measurement as well, but the same dip was produced.

Dirac Live limits boost to 10 dB, so one thing you can do to "trick" it is to lower the target curve. In some listening situations, this may be fine.

However, it would be better to understand what is going on. If you have a true null then you can't do anything about it. (Except move the speakers/LP.) A null is like 1-1=0. Mutliply by 10: 10-10=0. And so on on - no matter how much power you put in, the result is always zero.

It looks like your measurement have smoothing? (It's looks like it but I can't be sure.) This will disguise if it's really a null. I would suggest looking at your measurements without smoothing.

Thanks.

Could it be some sort of cancellation? In the "better" curve above, I have set the 5db gain to the right speaker only, if I set it to both, the dip is the same.

I also got better measurements by adjusting delay on the left speaker, but I had to skew this so much that it was obvious when listening to music, so it could not be used.

No smoothing:


Edit: Below is the measurement from Dirac Live. It does not seem like there is any null before Dirac filters are applied?

Sorry for the delayed response.

Right, so yes, it's a cancellation. You have one at 90 and about 160. DSP can't do anything about a complete cancellation.

You haven't said whether the Dirac curve you have shown is one location or the average of nine locations. You have to compare like with like.

That curve is for nine positions.

After my last test, I set a house curve, and chose not to measure too much. The sound is really quite good, and I will be happy with the result as long as I manage not to obsess too much about the curves. I did test your suggestion, lowering the target, but the result was pretty much the same.

Thanks a lot for all your replies.

Yes very good thing to not obsess on measurements! Especially only one point, it will drive to bad solutions.

So your previous REW graph is 1 point, and your Dirac graph 9?
If so it's normal that the dip disappeared in the averaged response from Dirac, you can probably get it back by showing the « all before » tab.
If it appears on only 1 or 2 of the blue traces nothing to worry about. If there's more maybe you can optimize the sub crossover, timing or placement, so pre-Dirac.

If you want to compare both REW and Dirac measurements you should try to always measure the big picture, so the same 9 positions every time. And even with this way you'll get some minor discrepancies, different windowing/smoothing? I don't know.
(so 8 sweeps averaged into one during one measurement, or 9 independent sweeps averaged by hand).

Thanks grosso, and john, I see now what you mean about the averaged responses. I have looked at the individual graphs in Dirac, and it seems, as you say, that the dip is much more present in some of the measurements.

I guess "sweetspot" is the exact place where the dip is the worst, something that makes sense taking the (ill suited) room into consideration.

Thanks again.

How many and what speakers you got?
What are their crossovers?
Best to draw the locations in your room.
Also, Dirac "after filter measurements" are not really so, they are simulations.
Only REW with Dirac Processor on shows what is the true effect of Dirac filters. Do REW in the same 9 spots you did Dirac, then average them and in the end smoothe 1/6. Dirac does 1/8, so pretty close.

Hi.

The setup is very simple, just two fullrange speakers. The problem is my room. It is a small room in a cellar, low under the ceiling. And with extremely limited possibilities for acoustic treatmets. One speaker placed close to a corner, the second with more space. Both speakers close to the front wall.

So, I understand that this is probably the worst possible setup to work from, but Dirac has an extremely positive effect right now. I was just hoping that there was some options/tweaks/tricks that could help with the cancellations.

Thanks.

"Digital Room Correction" is a misnomer. No electronics/software can "correct" the room. What it does it tries to influence the speaker-room interaction through changing the sound source behavior. What is achievable has severe limits. Especially under Schroeder frequency (typically 200 Hz) where the physical room takes over the control of sound. The only way to really "correct" the room there, is to rebuild it. Best start byincreasing its size above the wave sound's length, for bass it is over 30 feet/10 meters. Now, that's not very practical. The next best thing is to break standing waves by inserting multiple (3-4) low frequency sources (subs). Google Earl Geddes, Todd Welti, Duke LeJeune (Swarm subs).

Hi, increasing the size of the room won't solve the problem, it will move the fundamental mode below the audible range but all the hundreds of other (higher in frequency) modes will still be there.

Multiple bass sources can be used although it requires careful tuning - just plonking a bunch of subs down won't necessarily be an improvement.

The simple(st) solution is to change locations of either bass source(s) or LP. I got the impression that this is not feasible for lintoft. However now that you mention multiple sources he could (if he wanted) reintroduce the sub that he was using before - not for the purpose of more bass for for the purpose of filling the notch. It would mean a lot of measuring.... MSO (multiple sub optimizer) may also help.

The other question I suppose is exactly how audible the notch is. It may also be interesting to see measurements of both main speakers at the same time.