TOPIC: Why use REW ahead of Dirac?

Hi Everyone,
There are a few YouTube videos out there which suggest using REW before running Dirac. Why? If Dirac corrects as much, as it does, what are you supposed to do first with REW?

Thanks

REW can give you insight before doing Dirac calibration. While Dirac can correct room response, it is not able to correct large holes in the frequency. This is more relevant when integrating a subwoofer. This will give you a change to play with the crossover and subwoofer placement. If you get this closer to acceptable Dirac will do a better job correcting any anomalies

Ah ok. So use REW for large corrections / initial set-up, then run Dirac for “fine tuning”.

Well yes, but we have to be precise about what we mean by “corrections”. You don’t want to use REW for creating filters only for insight.

Ahhhh OK. So measure with REW and use it to help find the best physical location for speakers and sub to minimise dips and peaks and use it to help identify where to set crossovers. But you're saying don't add filters.

Exactly

Dirac Live is far from perfect and doesn't always end up doing exactly what you'd like when you go back and measure again with REW after applying room correction. I've tried both ways and have gotten much better results by fully integrating the subwoofers using the PEQ on the Minidsp, letting Dirac do it's thing, and then measuring again and make final adjustments with the PEQ. The EQ section of REW has a great prediction window that can show you what any custom filters should do in real time. It's the best way I've found to actually make everything measure just like the predicted response that Dirac shows.

I'm hoping a future iteration of Dirac Live lets you take measurements, run room correction, and then remeasure to see the actual effects and let Dirac adjust it's room correction again accordingly. Until then, the best method is to let Dirac Live do the heavy lifting, and then come behind it and smooth out the response with very specific filters with the Minidsp PEQ.

Hi,
Thanks for your thoughts also. I’m sure everyone who runs Dirac can’t help but run REW afterwards to see what Dirac has done. You don’t see many before and after REW plots which is a little telling and I fully expect the curve after Dirac to not be ideal. Especially when Dirac themselves say that it won’t be. But they justify this by saying this is the best average that can be achieved for all measured positions and that their correction has fixed driver alignment issues and power response.

So my question now would be that if you do your own PEQ fine tuning after Dirac, might you be messing up some of the improvements Dirac made to the response and power?

Thanks

I'm sure it slightly counteracts some of the good that Dirac does by adding filters after correction, and the debate if anything like that is actually audible is still raging on I'll bet. I for sure cannot hear any negative artifacts from the added corrections. The PEQ filters don't seem to negatively interact with any of the filters already applied by Dirac either when I was experimenting with trying to cause a phase issue by using a narrow Q filter in an area I knew Dirac already adjusted. Stacking filters like that with just the PEQ would definitely have some negative consequences show in the measurements.

I'm totally fine with Dirac getting things 90% of the way there and adding the last 10% myself. It still sounds better than just 90%. When adding PEQ filters before Dirac, if there were any improvements I wanted to make afterwards, I would have to stack PEQ filters on top of each other. I've also seen Dirac over or undercorrect areas that I had PEQ filters already set at. By running Dirac first, I can see what it did right or wrong and go back and correct any anomalies myself with far fewer negative consequences, if any.

My normal workflow when setting up a new set of speakers is as follows:

1) Use REW to measure the driver responses individually, overlay them on the same REW screen, and make an initial guess at where the best crossover points will be
2) Set up the crossovers, and use REW to see what is happening at the crossover points.
3) Adjust the levels, measure again, ... etc. to get the best possible fit at crossover(s).
4) Check to see if inverting the phase of one (or more) of the drivers improves the behaviour at the crossover. If it does, repeat from step 3).
5) Set up PEQ filters to correct any gross problems with the overall FR curve. If you're feeling energetic, this is ideally done outside with the speaker mounted on a pole a few metres off the ground (and mic likewise), but the "gating" function in REW is a good substitute. Don't bother too much with the LF end, say below 200 Hz, because Dirac will actually do a much better job of dealing with this than you will. If you're not happy with the end result, save the REW data and the settings so far and repeat from step 2) with different crossover frequencies. (In some cases, getting a good overall result can be achieved with some overlap at the crossover point - in one of my 2-way systems I run the Mid/Hi using a 100 Hz high pass and the bass drier with a 300 Hz low pass).
6) Run Dirac using an appropriate seating arrangement, and of course apply your fave Target Curve.
7) Leave it alone, and enjoy!

I haven't done any rigorous testing between the above approach and simply leaving the levels and PEQ alone and letting Dirac figure it all out for itself, but intuitively it seems sensible to give Dirac the least possible amount of work to do so that it isn't wasting its filtering capacity (which must be finite) on problems that could be fixed by the tools available in the Outputs section.

Thanks both

@Tony_J, wrt step 4 inverting one of the drivers. Surely you can’t invert left or right as then the mains would be out of phase or can you?

Indeed, if you (for example) invert the tweeter on the left side, you'd better do the same on the right otherwise you will have a mess. Similarly, if you choose a crossover point, or define PEQ filters, you'd better define the same crossovers and filters for both left and right.

Ok thanks for clarifying. I can’t do just the tweeters as tweeters and woofers on same Dirac channel. Sub on ch 3.

In that case, might be worth switching the electrical connections on the tweeters (or the woofers, whichever is easier) to see which works better.

Slightly hasty reply there...sounds like you are integrating a sub with a pair of passive speakers? In which case, I would forget about messing with the relative phase between the tweeters and woofers, just concentrate on the crossover between the passive channels and the sub. You may well find that inverting the sub will make a difference (good or bad!)