Baffle diffraction study for new project

[ergo]

Hi,

I have been messing with LspCad and the new KHF tool. As I'm planning a new high quality, higher cost speaker project I want to do better than my FatMargaret project from few years ago.

http://www.hifiklubi.ee/fat_margaret_project/

Back then I assumed that if I round the edges so much then it will not leave me too many diffraction problems - I was wrong....

***************

Why you ask - did it measure that bad. Well yes, it seems it can be done much better.

I started by creating a response of "ideal tweeter" and used the new KHF Tool for study



Then I simulated the tweeter as it is in my Fat Margaret.


Now the comparison of the simulated response againt the real measured response of the tweeter in above box

brownish = simulated
grey = measured 60cm on tweeter axis with 6ms window.

So it can be pretty much said that the rounded edges do not make too much difference

Now I know better for my next project

To be continued....

Ergo.

[ergo]

The next project will be a floorstander with tweeter, 6'' midbass and a side firing woofer.

Now I was thinking that for the new project I would make it different than the usual - meaning I would place the midbass higher than the tweeter on the front baffle.

This in itself gives a much better situation

Now if the tweeter is additionally offset in horisontal line the response looks even better

Can enyone give feedback of experiences with the tweeter horisontally in center of the front baffle versus the tweeter offset by few cm ?

There are loudspeaker designers who claim that offsetting the tweeter creates a smoother response but the off axis behaviour will be worse and thus the centered version would work better - any comments?

Regards,
Ergo

[Paul W]

My brain is a little whacked from a 12 hour plane ride back from Japan, but here are a couple of thoughts:

The side radius on the baffle in the picture appears to be more than 12.5mm.

The top edge of the cabinet (and woofer) also impacts diffraction, especially when the tweeter is so close to the top edge. In addition to offset, the mid-on-top configuration on-axis simulation looks better because it spaces the tweeter further from the top edge.

Narrow baffles are more subject to diffraction in the treble region than wider baffles, so offset is more effective with narrow baffles. Offset tends to average diffraction on-axis, but off-axis response does become asymetric (not necessarily worse).

Now I need more coffee.

[cjd]

For starters, set your window to be much smaller vertically. BDS uses like +/- 7.5dB IIRC.

Second, rounded (or chamfered) edges can indeed help the situation. While it never removes diffraction issues, it absolutely can help spread them out so that you get smoother response despite everything.

Diffraction is at its best (on an untreated baffle) when you have the tweeter a different distance from every edge because it is a relationship between frequency wavelength and that distance. With most modern tweeters, you can not eliminate the actual tweeter face-plate, which means you're dealing with diffraction of some kind - the goal then is to minimize it.

You can improve things a huge amount by using wool felt on the baffle though. I have a couple measurements in the stickied RS 3-way thread (think they're on p16) that show a little of the differences felt can make. Earlier on there are also some discussions of diffraction. In fact, if you find it is an issue on your current design, stick a slab on either side of the tweeter (vertical, ~1/4 to 1/2" thick). For a quick and dirty stab, you can get a sheet of wool felt at the store and "roll" it up to a 1/2" thick by 6" and tape it to your baffle. :P

FWIW, design is always about picking your compromise. Most audiophiles go on and on about off-axis performance and then deal very little with it by refusing to get out of the "sweet spot". YOu can also balance between centered and not centered - the issue is more a relationship between all your drivers to each-other, not the relationship of the drivers on the baffle. So you could simply offset ALL your drivers and keep them centered relative to each-other.

C

[JonMarsh]

As Chris points out, panel layout and drive positioning is a complex topic, which is why I model any speaker extensively for baffle diffraction/loading effects prior to planning and cutting a test box for the front panel. Has a big effect on both the midwoofer and tweeter, and asymmetric is virtually always better, even though it doesn't "look" as nice.

In my article on the M8a, I documented changes in edge diffraction as a result of heavy felt treatments. It wouldn't work as well as it does if the driver layout hadn't already been carefully optimized for midrange smoothness for both the tweeter and woofer.

The same concerns apply to dipole systems.

IMO, this is one fo the nicest add-ons Ingemar has done for LspCAD6; now I don't have to wrestle with BDS any more.

~Jon

[dlr]

More to consider

The others' posts covered most of what I would have mentioned, with one exception. There is a source of diffraction from the tweeter response that is almost always an issue and is too often overlooked. That is the diffraction from any driver(s) mounted close to the tweeter.

I've found recently that hard diaphragm drivers have significantly more influence. The Accuton C95-T6 had more of a diffraction signature than any other driver I've tested. I used a double thickness (1", 2" wide, 6" long)) of felt between the tweeter and the Accuton. It still had some influence, but it was greatly reduced. The Seas mid/woofer will likely be similar, much worse than a treated paper cone diaphragm that has some self-damping.

So roundover the top (as was suggested), offset if there's room on the baffle after roundover, and put some felt between the drivers.

dlr

[JonMarsh]

I completely agree with you Dave, have seen that with the MTM's especially. A little baffle spacing with an MT and use of heavy felt helps a lot with that.

BTW, do you have the C95-T6 design posted? I couldn't find it readily. What do you think of the C95-T6?

~Jon

[oneoldude]

Could somone please post some pictures of what proper placement of felt looks like?

I would like to see examples between drivers (MTM?) and to the sides of Tweeters. Perhaps only one pic is necessary?

[ergo]

Ok, thnx for replies - seems I have good guidelines for the tweeter.

Now I have a next point to solve. As my speakers will have a side firing woofer (SS w8565-01) the question is how high I should mount the woofer on side panel relative to floor?

I simulated with LspCad 5.25 checking only baffle diffraction + floor bounce



Looking at this simulation it's clear that mounting the woofer closer to floor gains few dB of efficiency and has less floor bounce effect.

On the other hand the lower placement would make a bigger cap between midbass and woofer and also exites the room mode in room height direction with almost maximum efficiency.

Does anyone have experiences with similar speakers (side firing woofer)?

Ergo

[dlr]

One example of felt placement

Could somone please post some pictures of what proper placement of felt looks like?

I would like to see examples between drivers (MTM?) and to the sides of Tweeters. Perhaps only one pic is necessary?

It's not very pretty, but it was when I was experimenting early on.

http://www.speakerdesign.net/felt_amelioration/3way_module_closeup.gif

This was a more complete application to help with both the edge and driver diffraction.

Here's the full page with corresponding measurements.

http://www.speakerdesign.net/felt_amelioration/feltssenter.html

dlr

[dlr]

Accuton C95 impressed me, a surprise

I completely agree with you Dave, have seen that with the MTM's especially. A little baffle spacing with an MT and use of heavy felt helps a lot with that.

BTW, do you have the C95-T6 design posted? I couldn't find it readily. What do you think of the C95-T6?

~Jon

I was leary of this, but received the pair for doing some measurements for a friend moving and clearing out his stash (talk about a luck day!). I'd been very disappointed in the Seas W17 earlier, probably more must my crossover not working for it.

Anyway, I was surprised at just how easy it was to work with. The crossover (I think I did 2nd at 2K, have to check) work was fairly simple. I also just stuck them in a box I had sitting around unused, maybe in the 30L range, I need to check. I still haven't even modelled it to see what it looks like on paper.

The box was ported (a cheap yard sale buy, I tossed the drivers, but it has a 1" MDF baffle), but I had no idea what the tuning would be and intended to set it on a woofer for a 3-way test. So I highly overstuffed the box up to the port.

In the 3-way (Peerless 10" with an old Dahlquist active crossover) it performed fairly well, since I could actively adjust the Peerless by ear. But as free-standing 2-way it's surprisingly good. At DIY 2005 it was considered by many there to sound better than the Ellis 1801-b that was brought. The bass wasn't as extended due to stuffing the port, but the midbass and midrange was what impressed, I think.

I mated it with one of my hybrids,a MDT-23 diaphragm assembly in a PE 275-070 motor, essentially a Morel MDT-32.

I keep thinking that I should put it in a real 3-way, but I'm happy with it as it is for now.

I have only posted my current 3-way design, no others. I've been thinking of putting up the C95 design, but I never tweaked it. I kept the first crossover I built for it. I did end up designing on the c95 axis with an inverted tweeter, not my normal axis. There was just too much offset at 1m (flat bafflfe). The C95 axis was nearly ideal for the offset, but since I figured I'd use it where I was up and down moving around, the tilted up design axis seemed appropriate.

dlr

[oneoldude]

dlr - Very interesting page.

In fact, the entire site is interesting.

You answered my questions.