Impulse response help in REW

Hi To All,

I'm trying to understand how to correctly apply delays in taking semi- dual channel measurements. Up to this point, I have only worked with Umik-1 using single channel mode. What I have now is a Dayton EMM-6 mic and Scarlett 2i2 interface using a loopback connection on the reference channel. Everything seems correct connection wise, but my lack of understanding has to do with applying the information presented. So I have captured a delay on reference channel of 2.78 ms for woofer and 2.65 ms for tweeter. My question is how to apply these values to the impulse response to accurately reflect the necessary delays?

Thanks In Advance,
Rich
 
Hello Rich,
I’m responding while a bit “burnt” so apologies in advance if I make no sense (or end up just being wrong).

Were these measurements made with the microphone about 1 m from the speaker? 2.78 ms corresponds to ~0.96 m if a speed of sound of 344 m/s is assumed.

That would put the tweeter (2.65 ms) 0.91 m away. There may be delays related to the crossover network that have to be considered, but does this sound physically reasonable?

It would helpful to nail that down before talking about next steps.

Few
 
Hi Few,
Thanks for your reply. My goal is to export frequency response measurements for passive crossover design in Xsim. There is mod delays within Xsim to account for the path length differences between drivers. I just need to know how to accurately arrive at the correct figures based on the measurement delays I originally posted. I took the measurements at 36 inches on the tweeter axis.

I understand about being 'burnt' because I'm there myself. Maybe we can continue the discussion tomorrow.😀

Best Regards,
Rich
 
Okay, 36” makes sense. Were there capacitors or inductors in the mix, or were the drivers connected directly to the amplifier? If connections were direct, then the mic wasn’t equidistant from the acoustic sources (woofer and tweeter). If one was above the other, then adjusting the mic height would affect the timing differences. It’s not trivial to identify the “true” acoustic origin for each driver; it might not align with an obvious physical feature.

Few
 
You’ll likely have to decide what you want to define as the “preferred” vertical listening position. Ideally, your speakers will not be sensitive to this choice, but some crossover designs make vertical position very picky. I’d call that a bug, rather than a feature, but everything is a compromise.

Few
 
Because you measure dual channel, the measured delay numbers are valid. The resulting phase difference due to the delay has to be accounted for in your crossover. Place the first window marker in both impulse graphs at (somewhat less than) 2.65ms and calculate the response for both units. Remove excess phase and import the data in Xsim.
 
@markbakk
Thanks so much. That is what I thought was the steps necessary. So, in Xsim I would delay the woofer the equivalent of .13 ms or 1.74 inches since to reflect differences between tweeter and woofer timing on reference channel. Is that correct?

@Few
To answer your question(s), measurements were taken without capacitor or inductors in the mix. The mic was not changed between measurements and woofer is below tweeter in typical two way bookshelf speaker on a flat baffle.

Best,
Rich
 
Montana, I use dual chanel measurement too, in REW. When done properly, measured impulses of the drivers already contains their phase relationships, so you do not need (and you should not) to insert delay to Xsim, or any Z coordinates in crossover modelling program. For dual chanel measurement, Remove excess phase = shift both impulses by 2.65 (be careful, start of the impulse is not the time of the peak. So 2.6ms might be safer) ms to the left, so they will be both very close to T0. As both are shifted by the same time 2.65ms, their phase relationship doesn't change. After doing that, phase showed in XSim will rotate significantly less and you will be able to align phase tracking of the drivers easily.
 
Last edited: