PDA

View Full Version : New Adventures In Hi Fi



Ed Zeppeli
09-14-2017, 05:37 PM
Just to show the process a bit as I'm sorting things out here with DSP. Here are a few screencaps of what REW is showing my Left transducers to be doing. Horn Module and Woofer are separated with individual sweeps. Before I touch the EQ I'm aiming for smoothest phase response over the XOver region of 750Hz. I can tweak this by delaying the horn or the woofer in the driverack. Mic position is ear level, middle of the couch. When I get to the EQing I'll take 3 or 4 averages from within the listening window. Also with this data and the impulse response I will be able to gauge the positioning and effectiveness of some sound panels I will be installing.

78286
78287

In the top capture, you can see the initial rise of the horn (steep spike) is aligned with the initial rise of the woofer. The peaks however are a little under half a millisecond apart. The lower cap shows phase relationship in a range an octave below and above the target crossover frequency.

Here are Left Woofer, Left Horn and a Full Range Sweep overlaid.
78288

I've moved the woofer .25ms forward in the time window to see what would happen at phase and crossover. REW will sum or subtract the traces as I make changes to simulate the effect. It correlates well with real-world results and so saves the time of running new sweeps for every new tweak. We can just test things in the box until we get a good result. Light green is a summation of the woofer and horn with the time shifted woofer and Olive colour is the subtraction of the same. The subtraction is useful, the deeper the notch the better. Here you can see the notch is down 28dB. This is the virtual equivalent of reversing polarity of one of the drivers and trying to get the best null with a tone playing at the crossover frequency.

78289

Here you can see the phase tracking a little better. The olive green is the inverted version. I wonder if I can do better in the upper region of the driver overlap.

78290

Ed Zeppeli
09-14-2017, 05:43 PM
Room's a bit messy but you get the idea. There's about 20 more feet of room behind the lense.

78293

Ed Zeppeli
09-14-2017, 09:27 PM
Flipping the polarity on the horn module and changing the woofer delay to 0.4ms has given a much closer alignment over a broader range. The notch may not be as deep as there is some divergence in phase at the specific crossover frequency. We'll see how this approach sounds.

78294

Ed Zeppeli
09-15-2017, 02:23 PM
I tried the same technique on the subs to woofer XOver today. I got the phase to look pretty but am also enjoying a major suckout at around 60Hz.

This is the phase with Woofer crossed at 50Hz LR12dB and Subs crossed at 45Hz LR24dB for a combined acoustic meeting point of 75Hz.

78298

Ed Zeppeli
09-15-2017, 02:27 PM
The developer of REW has been using my .mdata uploads to preview an upcoming software enhancement in the next beta of REW.

Check it out.

78299

Ed Zeppeli
09-21-2017, 07:05 AM
Both Subs and Both Woofs swept - phase alignment


78355

1audiohack
09-22-2017, 10:26 PM
You are on the way to great sound. So many DIY folks totally miss what you are digging into.

The phase traces of you ULF and LF is great. Do you have a high pass on the ULF? If so is it in the output channel or in the input?

What DR?

I like the color display top view, very intuitive. To what do you attribute the break about 400Hz?

The lack of hysteria over the LF group delay is refreshing.

Thanks for sharing and please continue.

All the best,
Barry.

Ed Zeppeli
09-22-2017, 10:53 PM
You are on the way to great sound. So many DIY folks totally miss what you are digging into.

Thanks. There's a lot more involvement than just delaying to align the impulse peaks if you're looking at best alignment over a range. I knew it was important but didn't know how to make sense of it in REW until I figured out Frequency Dependent Windowing. Then it became a little less jumbled and the picture became clear both literally and figuratively. The tough part was figuring out how to align them with delay. one tenth of a millisecond shifts things more than you'd think on these graphs.

The phase traces of you ULF and LF is great. Do you have a high pass on the ULF? If so is it in the output channel or in the input?

What DR?

The Subs have a high pass at 20Hz. The crossover is at the beginning of the output channels in the Venu360

I like the color display top view, very intuitive. To what do you attribute the break about 400Hz?

Yeah I like it too. I can't wait to play around with it in the next beta release. That break - which is glaring in that graph - seems to coincide with a notch in the frequency response data I had submitted to the REW forums. The notch it no longer as prevalent with all the changes I've been making so it'd be interesting to have a fresh look. It could be a room mode or group delay given the lag in time there.

The lack of hysteria over the LF group delay is refreshing.

Thanks for sharing and please continue.

Cheers. I'll keep posting as I make progress.

1audiohack
09-22-2017, 11:45 PM
That little blue lake between two and three ms and about 425Hz coincides pefectly with an eight foot ceiling as your RSD, Room Smallest Dimension. I am uncertain as to how REW derives phase but if you have a ton of stored energy there it might blow out the calculation.

If you had a cardioid mic you might be able to "look" at your estimated reflection points and measure the spectral content of the bounce and see what your panel placements do for you. TEF PET does this

Barry.