Home Theater Forum and Systems banner

1 - 5 of 5 Posts

·
Registered
Joined
·
261 Posts
Discussion Starter #1
I have been following jtalden's method for time aligning sub to mains in this thread. One point of discussion has been the relatively "messy" shape of the impulse response from my subwoofer and I am curious as to why this is and whether it is meaningful with respect to sound quality (aka is it a bad thing). Ultimately this is used as a precursor to phase alignment checks so I'm not sure it's meaningful at all but I am curious to learn more.

This shows the completely un eq'ed response of the sub in room (which is extremely uneven to say the least!), post eq it run broadly flat in room to ~10Hz.

raw_sub_fr.png
raw_sub_ir.png

I've done some limited research around the subject and came across this article from Bob McCarthy's blog who talks about why he just uses phase alignment. Of note appears to be this paragraph

The IR we see comes from a single LINEAR transform. The importance is this: linear data favors the HF response. If you have 1000 data points, 500 of them are the top octave, 250 the next one down and so on. This means that our IR peak – where the “official” time will be found, is weighted in favor of the highest octave. If you have a leading tweeter, The IR will find it ahead of the pack (in time and level). The mids and lows will appear as lumpy foothills behind (to the right) of the Matterhorn peak. If you have a lagging tweeter, the IR will show the lumpy foothills ahead of the peak (to the left), but the peak will still be the highest point.Our peak-finding function will still be drawn to the same point – the peak.
this suggests that the low frequencies played by my subwoofer are lagging the higher frequencies & perhaps this is driven by the room interaction? AIUI excess group delay is a measure of the drivers ability to reproduce frequencies without smearing and that graph shows an extreme, presumably room null based, spike at the XO (80Hz).

raw_sub_excess_gd.png

If this line of thinking is accurate then it suggests I would be able to verify empirically via some narrower sweeps (e.g. in 20Hz bands from 10 to 90).

Ultimately what I am interested in is how to "read" the impulse response & whether my mildly informed speculation above is accurate. It would also be interesting to know whether I should be changing any settings to get better quality data out of these sweeps.

Thanks
Matt
 

·
Registered
Joined
·
261 Posts
Discussion Starter #2
I've attached some limited bandwidth measurements & this appears to back up the thought process in my previous post. However I don't actually understand the maths behind how the IR is generated so this could be a completely spurious (there really is smoke without a fire) result :help:

ir_comparison.png

Cheers
Matt
 

Attachments

·
Banned
Joined
·
4,838 Posts
Not an expert in this area... I would start out by trying to get the cleanest - most even - unequalized frequency response plot possible for your sub, working with sub placement in the room, then your impulse response will be easier to interpret. Maybe that is not an option. The sub response as is looks kinda hard to work with anyway, so your end result might benefit from a better starting point frequency-response-wise. The phase alignment process might proceed easier from there. Just a thought.
 

·
Registered
Joined
·
261 Posts
Discussion Starter #4
Not an expert in this area... I would start out by trying to get the cleanest - most even - unequalized frequency response plot possible for your sub, working with sub placement in the room, then your impulse response will be easier to interpret. Maybe that is not an option. The sub response as is looks kinda hard to work with anyway, so your end result might benefit from a better starting point frequency-response-wise. The phase alignment process might proceed easier from there. Just a thought.
yes this is what I'm in the middle of, my current effort does look somewhat cleaner

sub_only_shaped_fr.png
sub_only_shaped_ir.png

Cheers
Matt
 
1 - 5 of 5 Posts
Top