Baffle diffraction simulation questions

[Saurav]

Hi,

My new speakers will look something like this:



I'm playing with edge.exe and the baffle diffraction spreadsheet to try and decide where to put the midrange driver. I don't think the tweeter's position on the baffle will matter much w.r.t. diffraction issues (XO will be at 5kHz or higher), that'll probably depend more on dispersion and XO and acoustic centers and related issues.

Anyway - first, the midrange. This will obviously be an open baffle, since the driver isn't inside the box even though there's a box under it. Current dimensions - the box is 24Hx14Wx15D, and the cardboard baffle is 14Wx10H. Questions:

* When modelling in edge.exe, do I put the baffle size as 14x10, or 14x34 (total height of cabinet + OB)? How does the height affect things? I didn't see the FR change much when moving the driver up and down at the same horizontal position.

* Speaker source density - what's a good number to use for a 6.5" driver? The ripples in the response get smoother as I use higher numbers, which is why I'm worried that I might be using too high of a number and masking off some unevenness that I'd have in real life. Is 6 or 8 or something in that range reasonable?

* My results are similar to what is shown on the help website. I get a large peak (+9dB) at around 1kHz, which settles down to about a +6dB response above 3kHz or so. This will be an active XO to the woofer, so that basically means that I have a +3dB peak at 1kHz I'll need to take care of. Interestingly enough, that peak is the smallest (~1dB) if I put the midrange right at the edge of the baffle. Is that realistic, or is it a simulation error because the driver is too close to the baffle edge? I've never seen any speakers where the driver is shifted all the way over to the edge of the baffle.

That's all I have for now.

Thanks,
Saurav

[Svante]

Saurav:

First, I think that there are "bad" positions for the tweeter as well. You are probably right in that the tweeter is less sensitive than the midrange, but still there are placements that affect the FR above 5k still. And, also, remember that the response somewhat below 5k can also have an effect since the filter will not be infinitely steep. I'd give the tweeter a try in Edge too, but prioritize a good placement for the mid.

Should the baffle be 10x14 or 34x14? Beats me. Strictly speaking Edge does not model your setup. The tweeter only radiated on the fromt side, so I'd go for 34x14 there, but the mid, I don't know.

Speaker source density. The higher number, the better, but slower. If the number is too low, you will se the ripple increase toward higher frequencies. This is an artifact of the simulation and could either be ignored or removed by increasing the number.

I have also noted that a placement near the edge seems preferable according to the simulations. I have some theoretical hints that makes this likely to be true, but I have no measurement data to support it.

Keep us updated on how it goes!

[Saurav]

First, I think that there are "bad" positions for the tweeter as well.

Agreed. It'll be offset a bit from dead center, and I'll probably try and keep it as close to the midrange as my woodworking skills will allow. But that will be the second step.

The higher number, the better, but slower.

So that's all it is then, an accuracy/performance tradeoff. That's good to know. I entered 81 in there by mistake, and that took a while. Which reminds me - the editing doesn't seem to work right. I had 16 or something in there, selected the whole number, typed in 8, and it became 81, and started the calculation immediately. So this is an informal bug report . Hitting backspace to clear the entry and then entering 8 worked fine.

I have also noted that a placement near the edge seems preferable according to the simulations. I have some theoretical hints that makes this likely to be true, but I have no measurement data to support it.

Thanks, I was curious about that. I'll measure my setup, but the tools I have at my disposal are pretty crude, so it is quite possible that your simulation will be more accurate than what I can measure.

Thanks for responding.

Saurav

[Svante]

Saurav:
Yes, it is just such a tradeoff.

And the bug; I know about it, it is one of the Delphi components. I'll see if I can find a way around it. If you want a poor excuse, it has to do with the limitation of the numbers in the spin boxes. For example I don't want fewer than 3 corners on the baffe. So, just erasing the number in the box will make it put a "3" there. Now, selecting the text and typing eg a "5" makes the box think that I first erase the number, it then happily adds a 3 there, and then it senses that I have typed a "5", so it adds it and the end result is that the number becomes "53". Arrrghhh...

[Saurav]

Ah, so it's not your fault. Spoken like a true software engineer (BTW, I'm one too, so I know how it goes )

Thanks for a wonderful piece of software, just in case I'm sounding unappreciative of all the work you've put into it.

[Svante]

Saurav:
No worry, you using it is positive enough for me .