Starting an Active Open Baffle design.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Regarding delays for trimming crossover phases, the difference needed in ms is really small. My AINOs have 0,00ms for woofer, 0,13ms for Low mid, 0,06 for high mids and 0,00 for Tweeter. I have noticed that for M to T a change 0.02 ms makes a difference! At 3000Hz a full cycle of wave is 1,13ms. 1ms means 34,0cm distance of sound sources. That means that you'll easily set delay several cycles late! (correct me if comma is at wrong place!) Here is John Reekie's advice for setting delay (scroll down to "Time alignment of drivers").
There is another way to introduce delay using the MiniDSP that you might consider. What you mention is true - the minimum digital delay of 0.02ms is too large for offset compensation of the typical M-T crossover. But when you have a plug-in with the "advanced biquad" programming capability, you can implement one or more all-pass (analog delay) filters as a biquad. Unlike digital delay, these have frequency dependent delay but this can be easily calculated. For instance I have a spreadsheet on my web page to do it:
http://audio.claub.net/software/Analog_All_Pass_Designer/Analog All-Pass Filter Calculator v1.1.xls
Then it is a matter of calculating the biquad coefficients for the all-pass filters and this can be done using my Active Crossover Designer tools.
the Active Crossover Designer web page

-Charlie
 
Regarding delays for trimming crossover phases, the difference needed in ms is really small. My AINOs have 0,00ms for woofer, 0,13ms for Low mid, 0,06 for high mids and 0,00 for Tweeter. I have noticed that for M to T a change 0.02 ms makes a difference!
At 3000Hz a full cycle of wave is 1,13ms. 1ms means 34,0cm distance of sound sources. That means that you'll easily set delay several cycles late! (correct me if comma is at wrong place!).
For a full cycle time of 3 kHz you have to divide 1 sec by 3000. The result should be 0.333 ms. For a deep dipole attenuation at 90° you want the phase difference between the drivers at the crossover frequency to be less than 10° in a 360°circle. This results in 0.333ms/36 = ~0.01ms.

If DrDynas tweeter is delayed by 0.02 ms with respect to the mid, that is not tragic, but it compromises the 90° attenuation already.
 
It's a beautiful morning, nice and cool! It won't be for long though!

Now testing no delay, instead trying goofy crossover positions. Woofer @ 350hz 12db/oct Mid @ 250hz 12db/oct individual and sum.

Edit: Mid is still peaking off-axis around 1.5k..I can't cross the Neo3 that low, I might have to just live with it, or use a smaller mid.
 

Attachments

  • woofer-mid-sum.jpg
    woofer-mid-sum.jpg
    83.5 KB · Views: 140
Last edited:
Previous post was with no LPF on the mid, hence the breakup was still present.

Here's the mid and tweeter sum with filters engaged, both @2800, mid 24 db/oct tweeter 48db/oct.

Edit: Added complete with overlay, individuals, plus with mid reversed.
 

Attachments

  • mid-tweeter-sum.jpg
    mid-tweeter-sum.jpg
    81.8 KB · Views: 136
  • complete-overlay.jpg
    complete-overlay.jpg
    97.4 KB · Views: 133
Last edited:
Aaaaand now...a very small amount of tinkering with the tweeter's response, on axis, 45 off and 90 off.

Edit: And, just because I'm curious (sometimes way more than I should be!) Here's an additional, 180 degrees measurement.

Edit2: I'm thinking I might leave the woofer as it is, with the roll off below 200 cycles. Once indoors, I can cross over to the subs at ~150 and adjust the slope so that they sum while running the mains full.

Playing music through them outside right now and walking around in a circle doesn't sound as peaky as the measurements would suggest.
 

Attachments

  • complete-overlay-off axis compare.jpg
    complete-overlay-off axis compare.jpg
    90.5 KB · Views: 129
  • 180 deg.jpg
    180 deg.jpg
    62.7 KB · Views: 133
Last edited:
Here's the mid and tweeter sum with filters engaged, both @2800, mid 24 db/oct tweeter 48db/oct.
Mid-tweeter is almost there. If you can try a 36 dB/oct high pass for the tweeter (instead of 48 dB), the acoustic crossover might become really symmetrical.
The woofer-mid transition is still a (sort of) mess. Can you show the response of both drivers in that area without any corrections? Possibly you need to apply some EQ before applying the high-and lowpass filters.
 
Sure thing, here's the tweeter changed to 36db/oct BW filter, no other changes yet.

Edit: It might get a little peaky now, Cicadas are starting to wake up...they're loud!

Edit 2: Mid and Woofer..Woofer is basically uncorrected, the mid has a little EQ to soften it up.
 

Attachments

  • tweeter-slope-change.jpg
    tweeter-slope-change.jpg
    82.3 KB · Views: 49
  • woofer-no corrections.jpg
    woofer-no corrections.jpg
    64.1 KB · Views: 42
  • mid-a little eq.jpg
    mid-a little eq.jpg
    65.9 KB · Views: 33
Last edited:
Thanks, but still not symmetrical enough. How about lowering to 24 dB/oct? And probably raise the frequency of the high pass 200-400 Hz on that way. Just try to achieve the deepest notch for the inverse phase case.

Thanks, I will try that. I have to stop outdoor measurements for today, the sun is starting to shine a little too much, so I have the cicadas making wayyyyy too much noise plus the heat on all the equipment :p

I'll try moving the crossovers around a bit more in the next couple of days and see what happens.

<3 the advice!
 
Sorry about wrong figures in my post, but I know I am poor at maths! It's sort of afasia, not noticing when I make an obvious/simple mistake. This poor mathematics makes me reluctant to try biquads and Charlie's calculator too, sorry.

Minidsp can set delay in 0.01ms steps. I think it is enough, I have reached very deep nulls with that. The difficulty is more in getting slopes symmetrical acoustically, to get symmetrical inverted nulls.

One problem is, which IR right window setting to believe???

Here is a collashe of just one outdoor measurement from my archives, different IR windows, no smoothing.
 

Attachments

  • ainogice v61 out o¤ hminv 2 4 8 12 20ms.png
    ainogice v61 out o¤ hminv 2 4 8 12 20ms.png
    386.8 KB · Views: 162
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.