Post by ansat on Jan 2, 2015 15:39:12 GMT -5
So lets look at the charts.
NOTE: This is what I expect to see from this particular speaker - the weird hump at the end seems to be a trademark of Definitive's older silk dome tweeters.
The chart above shows the spectrum. The blue line is what Dirac saw from your speaker. The orange line is what Dirac want to do. And the green line shows what Dirac did. A note about the green line - This is not an actual measurement and if measured, this is not what you would see.
Lets start with the blue line. The bottom of the chart is your frequency response from 10hz to 20,000 hz in your room and the side is db from reference (which is not a specific number). In a perfect room (or better yet - outdoors) this response would be + or - 3 on this chart across its manufactured measured frequency response.
The Orange Line. This is the target that Dirac is going to make a filter to match to.
And last, the Green line. This line represents the filter that is being created.
So what information will this tell us?
The blue line shows that the natural roll off my back speaker in my room should is 50hz. Therefor, my Xover for this speaker needs to be 50hz or higher (regardless of what the manufacturer claims).
The blue line compared to the orange line shows me that dirac is adding 8db of gain at 12k, 3db of loss at 5k, 8db on one speaker at 250hz (3-4 on the other). 3db of loss at around 120, 12db of loss at 65hz and 10 db of gain from 30-45. Out of all these gains and losses in the filter, which ones are important? I would say that any gain over 6db should be looked into to see if it can be corrected by other means. Any loss that is being applied is usually caused by reflections in the room.
Now to the misunderstood green line. For the most part. It is not helpful or an accurate representation of much. The only major thing to look for which you can see between 30hz and 50hz on my graph is where the green line deviates from the orange. This indicates areas where dirac applied the full 10db gain and it was still not enough to fix the problem. In my case, its the passive radiator on my rears that gave the boost at 30hz and dirac is trying to correct for that. I will be crossing over at a higher point then that to be sure that I avoid that much gain.
From flak (Dirac) on the green line -
The green line is often misunderstood because it is the average of the nine FR responses in the nine positions after correction, as such IT IS an accurate representation, even if a computed one, and it can be validated by averaging the same nine measurements with REW when using the same smoothing and mic position.
The blue line is an average of nine measurements before correction and it can be usefully and meaningfully compared to the green after correction line.
Those average graphs are useful in real life... the before and after correction graphs for the single individual positions are also accurate when present as in the standalone full Dirac Live version.
To illustrate what Flavio is talking about. Here is a image that
I pulled from the MiniDSP manual.
As you can see, each measurement carries wild swings below 1k and this would be normal. The measurement that we are seeing in the Dirac LE is just the average line. I do not know if we will get this functionality in the full version.
This chart is your impulse response. Dirac uses this to set your delays and this is a representation of energy over time. This chart is a pretty good one. (but the worst one that I have) If you look at the blue impulse line for any speaker other then your subwoofer. You want to see only 1 point and a nice tight grouping into that point (like what you see on the green measurement). Looking at my blue measurement the peak is at 9.94ms and you can see another peak at 10.25ms. While Dirac did not capture this perfectly, it is cause for concern as this is a reflection in your room that is pretty close to your speaker and is almost as loud. The further your peaks are out the further the reflection is. If you are getting some really bad peaks then REW will be needed to identify what is going on further.
REW tutorials with the XMC (Beginner tutorials using REW)
Do I need a minidsp for multiple subs and how to use one for such
Subwoofer placement with REW
how to use the USB input to measure other speakers
Dirac / XMC-1 F.A.Q.
As always, your questions and comments are welcome.
Let me know if I missed anything important.