Check phase matching

[Dan Twomey]

After following the instructions on the MiniDSP website I came up with the following.

Two way active speaker with both crossovers being LR4 with a crossover at 2500Hz. It appears that inverting the polarity of the tweeter is the best option?

Regards,
Dan

 

[sm52]

It appears that inverting the polarity of the tweeter is the best option?

For the best frequency response - yes. But you may not like how it sounds. It is always better if the tweeter is in positive polarity. Another solution is to try different filter orders on the left and right. For example, Left LR3, or LR2, Right LR4.

 

[Dan Twomey]

For the best frequency response - yes. But you may not like how it sounds. It is always better if the tweeter is in positive polarity. Another solution is to try different filter orders on the left and right. For example, Left LR3, or LR2, Right LR4.

I have tried Woofer LR4/Tweeter LR2 with some 'success'. There's a paradox at play here. The more I learn, the less I know.

Regards,
Dan

 

[John Mulcahy]

Odd looking measurement with those regular dips. 160 dB SPL span for the plot also, 60 dB would be more typical.

 

[Dan Twomey]

Odd looking measurement with those regular dips. 160 dB SPL span for the plot also, 60 dB would be more typical.

I could add more smoothing?

 

[John Mulcahy]

Hiding problems doesn't tend to make them go away maybe worth investigating what is causing those dips. If you post an mdat I can check for anything obvious.

 

[Dan Twomey]

Attached here is an MDAT that tests two things. First is a basic sweep of left and right speakers. Second is the same basic sweep with the tweeters with inverted polarity. The final sweeps are the left and right with -5db on the tweeters.
I was trying the -5db to see if it reduced the number of filters I needed to achieve a good house curve.

Regards,
Dan

 

[John Mulcahy]

There are a lot of strong reflections in those measurements, hence the dips. They would likely be reduced by keeping the speakers and mic as far away from walls and other surfaces (desks, tables etc.) as you can manage and measuring at about 1 metre from the speaker, if you are not already doing that.

 

[DanDan]

Hi Dan, as John said, a lot of strong reflections. Take a look at your Impulse graphs, ETC, you have reflections up to -5dB FS. Or your RT60 which has rising HF Decay put to nearly half a second. A common requirement for critical listening is below -20dBFS for at least the first 20mS. RT60 200mS.
In an extremely lively environment such as yours I doubt if small details are audible.

 

[Dan Twomey]

There are a lot of strong reflections in those measurements, hence the dips. They would likely be reduced by keeping the speakers and mic as far away from walls and other surfaces (desks, tables etc.) as you can manage and measuring at about 1 metre from the speaker, if you are not already doing that.

After taking the summer off to do other things I'm back to tinkering with my active two-way speakers. The furniture and speaker positions are dictated by room layout so the measurements are 'reflected' in that. In the near future I'm going to post a basic set (Right, Left, Right+Left) of room measurements with a few small changes. Past experience has taught me that I may prefer Woofer LR4 and Tweeter LR2. Both crossed over at 2500. I also may try a -3DB on the tweeter to help minimize the number of filters needed for a good house curve. More soon.

Regards,
Dan

 

[sm52]

Make a measurement separately for the tweeter, separately for the midbass of one speaker. The distance to the microphone is 1 m - 1.5 m. The microphone must not be moved during these measurements. And don't do FDW. Normal measurement. All filters must be disabled. All delays have been reset. Post here.
Then I'll try to predict how best to match the tweeter and midbass.

 

[jtalden]

Dan,
Just FYI in case you overlooked it.
The apparent isolated deep dips and otherwise smooth appearance of your posted SPL traces and attached file are the result of an REW FDW setting of 15 cycles. Removing that FDW better shows the actual response differences. When that is done the room responses then look pretty typical.

 

[DanDan]

I wouldn't sweat the small stuff. But I would encourage you to think about treatment. If you have height an absorbent ceiling would work wonders forever, taking no space. And clouds or full ceiling lend themselves to great lighting.

 

[Dan Twomey]

Before I 'progress' further I'm going to take a brief side trip and check the time alignment of my drivers as explained on MiniDSP's website here. Time Alignment

Regards,
Dan