Home Theater Forum and Systems banner
1 - 6 of 6 Posts

·
Registered
Joined
·
3 Posts
Discussion Starter · #1 ·
I had measured the woofer and mid driver response separately and combined them using trace arithmetic (A+B) and am puzzled by the results. see attached screen shot. at freq = 212Hz, amplitudes from the graph are:
A=125.1, B=125.5 A+B=98.7db
From my calculation:
A+B=10log(10^(125.1/10)+10^(125.1/10))=128.3db
I had checked severl dips, they all came out to be way too low. Areas without dips seem OK.
what gives?
does anyone know how trace arthmetics is done in REW?
thanks.
 

Attachments

·
Premium Member
Joined
·
1,562 Posts
The math is correct as it also takes into account the phase of the signal.

If you look at the relative phase of the 2 measurements you summed you will probably see that at the large dips the phase difference is 180 deg.

Also note that, if you are measuring at some distance from the speaker (as at the LP), the phase at these higher frequencies may sometimes be incorrect. One or both of the phase measurement may have been influenced by a reflection rather than the direct wavefront. This will then indicate a dip that is either not there or greatly overstated.
 

·
Registered
Joined
·
3 Posts
Discussion Starter · #4 ·
did the vector arithmetic, there is a large dip. but I still could not get numerical agreement.
at 212 Hz
from REW 100db at 16 degree
from my calc 115db at 47 degree.
 

·
Premium Member
Joined
·
1,562 Posts
JohnM or other experts would need to clarify further.

I think the math manipulation is actually done on the calculated IR and then back calculated to the magnitude and phase and is thus not simply vector addition, but I do not understand the methodology and resulting differences in detail.
 

·
Premium Member
Joined
·
6,711 Posts
A+B in trace arithmetic is a vector summation of the FFTs of the individual impulse responses, with appropriate handling of the mic/meter and soundcard cal corrections of the individual measurements. If closely examining small parts of a trace it is best to zoom in and tick the box to show the trace points, then check values at measurement points - at sharp changes in values the interpolation function that draws the curve that connects the points can exhibit overshoots/undershoots.

It is important that the traces being summed are properly time aligned, e.g. by using the option to use a loopback connection as a timing reference and making sure that any t=0 adjustments are applied to both traces.
 
1 - 6 of 6 Posts
Top