Do you have knowledge of felt diffraction reduction? Advice needed!

[Eigenmind]

Hello everybody, I'm newly registered here at Tech Talk even though I am a long time and lurker and PE customer. I'd like to first say hello and share that I've gained a lot of information and inspiration from all of the knowledgable posters here over the years. I hope that this is the start of more active role in this community!

I am posting here because I would like to hear from anyone who has any experience with high frequency diffraction control using felt. I would be remiss not to mention my gratitude to Dave's Speaker Pages for the study he has done regarding felt diffraction amelioration for tweeters. This page has been my motivation to experiment with the technique myself.

I have been working on (yet another) Reference series MTM system for music and HT usage. The design uses the familiar RS28F-4. I am at the crossover simulation / emulation stage with these towers. What has causing me grief is trying to understand the effect of edge diffraction on the high frequency response; I am trying to understand what frequency response features are characteristic to the system, and what features are position-dependent acoustical anomolies; of course I want to avoid attempting to 'solve' any acoustical problems through electrical network means.

I have done some simulation of the cabinets (The Edge) and gained a basic appreciation for the signature of the diffraction to be expected from the geometry of the design. From my measurements I can see the familiar series of peaks and valleys caused by diffraction, albeit at a lower frequency than was predicted. The top blue curve in the graph below is the on-axis raw response of the tweeter in the cabinet.

My next course of action was to try to "even out" these undulations using the prescribed pressed wool felt treatment around the tweeter. Unfortunately, my first attempt has not been as successful as I hoped for. In the picture, you can see I am using 5/8" F13 felt, leaving a 3"x3" window around the tweeter.

I have attached a graph of the frequency response at three angles in the anticipated listening region - the blue curve are without the felt, the red curves are with. From the graphs, I feel that using felt in this way suppressed the first "hump" of the diffraction effects; if all of the measured off-axis responses looked like the 10 degree case, I would be satisfied. However, a big dip at 3 kHz starts to develop at the listening angle increases.

This is the reason I am seeking your opinions and prior experience with this sort of technique. I would have just jumped in and started experimenting with different placements and whatnot myself, except that a) the felt is stuck on with an adhesive backing, and b) the felt is not the cheapest stuff in the world to experiment with. Thus, I would like to understand what is happening here and try to go about changing the felt in a informed and deliberate manner.

From the website I linked to above, I feel that I probably placed the felt too close to the tweeter dome and may be suffering from "horn effects" due to this. Perhaps the felt shouldn't be as symmetrical as well for best results? Maybe using slightly thicker felt is causing a problem? This is why I'm asking you for your input before I go cutting up more of this felt!

If anyone is interested in any sort of incremental measurements, I would be happy to contribute to the collective knowledge base here. Thanks for reading what I'm sure is the longest first post in history!

Regards,
Dan

[davepellegrene]

Re: Do you have knowledge of felt diffraction reduction? Advice needed!

I think the inside edge of the felt is creating the dip around 3kHz.The felt could be eliminating the boost you would normally get from the baffle since the baffle is really a 180 degree waveguide. What may help would be to cut the inside edge of the felt on a 45-60 degree angle. It would essentially look similar to a waveguide but should not give you the low end boost that a waveguide would give you since the sound would not be guided by the walls of the felt.

Another possibility is the dip is being created by cancellation from the wave bouncing off the 90degree edge of the felt. Which I believe the fix would still be to cut the felt on an angle.

Dave

[Eigenmind]

Re: Do you have knowledge of felt diffraction reduction? Advice needed!

I think the inside edge of the felt is creating the dip around 3kHz.The felt could be eliminating the boost you would normally get from the baffle since the baffle is really a 180 degree waveguide. What may help would be to cut the inside edge of the felt on a 45-60 degree angle. It would essentially look similar to a waveguide but should not give you the low end boost that a waveguide would give you since the sound would not be guided by the walls of the felt.

Another possibility is the dip is being created by cancellation from the wave bouncing off the 90degree edge of the felt. Which I believe the fix would still be to cut the felt on an angle.

Dave

Hi Dave, my first thought was that there was a destructive interference caused by the edge of the felt, but the geometry doesn't back it up. At the distance I was measuring (~30 in) the edge is too shallow to support a first bounce back to the microphone at these relatively shallow angles.

I also thought about beveling the edges to make a smoother impedance transition. Ironically, a 45 degree bevel would be perfect for bouncing high frequency sound back at the listener. With this level of absorbtion, though, I doubt it would be problematic.

Basically, the two easiest modifications I could think of would be beveling the edges (may look better, too) and/or reducing the height of the felt. The pressed felt is packed in such a way that I feel I could peel off horizontal layers.

I am trying to understand your first statement: are you thinking that the absorbtion of the felt is essentially making the driver 'see' a 4 pi space as opposed to the 2 pi loading of the baffle? It would make sense that this would affect lower frequencies because the effective baffle is acoustically smaller at low frequencies. Would this effect show up on-axis though? And I wonder why the 3 kHz dip is indeed a dip and not a shelf that I might expect from a phenomenon like that? (<- of course the absortion of the felt would decrease rapidly with frequency, reducing the effect...) If I remember properly, these graphs are valid to around 500 Hz.

Thank you for your insights! I'm interested in hearing others chime in on this...

Regards,
Dan

[[email protected]]

Re: Do you have knowledge of felt diffraction reduction? Advice needed!

I suspect the horizontal parts are not needed and could be problematic. Have you measured with just the short side piece installed?

[dlr]

Re: Do you have knowledge of felt diffraction reduction? Advice needed!

I'll throw in my $0.02. First, don't reduce the thickness. The placement will change what you need to some degree.

There is some validity to reflection. That's why I suggest that the felt not be placed too close to the tweeter dome. I've found that stepping can help, with shallower pieces closer to the tweeter. But there should not be much overlap of the faceplate. If the latter is nicely flush, overlap is not necessary.

The suggestion to remove the piece on the longer side is looking at the right area, I think, but not what I would do to start. The shape is rectangular and very close to the dome. I would spread those two out, make the opening rectangular. At least move the longer side out a bit, looks like the inner side has no space. With a square opening I found that an offset, not with the dome centered, always worked better. Helps to distribute the distances better. The closer the sides, the more of a "horn loading" effect occurs, but that is only a partial effect because the felt is largely absorptive. When non-centered and/or non-symmetric I suspect that horn loading is minimized. If you look at all Dunlavy speakers, he used large rectangles and stepping in almost all of his designs. My tests correlate well with this arrangement.

Keep in mind that the goal of felt is to absorb the wave passing through it and it must actually pass through it to be effective, but you want to minimize reflections. My thought is that it helps by making the wave that reaches the edges more "disruptive" in that it is not fully in phase with the main wave and a different magnitude. Since there is a very distinct change in the step (I've made 2db+ reduction at times), the wave reaching the edge is either greatly reduced or is partially destructive, probably more the latter in my guess.

You could try removing the piece facing the longer side to see what it does, but as long as the felt is not reflecting I suspect it will be better with it in place. You could also try having the longer side piece of felt a bit farther away than the short side. After repositioning, the one other suggestion I have is to add another piece on the long side. There is always a point of diminishing returns to thickness and width, but at lower frequencies, wider is better to a point. At higher frequencies the felt is more effective.

dlr

[timw]

Re: Do you have knowledge of felt diffraction reduction? Advice needed!

I cut a saw tooth pattern in the felt edge facing the tweeter, 1/2" deep from tip to root. A sharp serrated bread knife or electric knife works well to cut the felt.

[davepellegrene]

Re: Do you have knowledge of felt diffraction reduction? Advice needed!

I am trying to understand your first statement: are you thinking that the absorbtion of the felt is essentially making the driver 'see' a 4 pi space as opposed to the 2 pi loading of the baffle? It would make sense that this would affect lower frequencies because the effective baffle is acoustically smaller at low frequencies. Would this effect show up on-axis though? And I wonder why the 3 kHz dip is indeed a dip and not a shelf that I might expect from a phenomenon like that? (<- of course the absortion of the felt would decrease rapidly with frequency, reducing the effect...) If I remember properly, these graphs are valid to around 500 Hz.

Thank you for your insights! I'm interested in hearing others chime in on this...

Regards,
Dan

I believe the on axis hump you see at 1600Hz of your measurement without the felt is do to the baffle. That wave length is around 8.5" which is about the width of your baffle from what I can tell. Below that frequency it is wrapping around the baffle. I was saying the felt is either obsorbing the boost or scattering it. As you move off axis the boost is less.

The felt does seem to help around the 5Khz range similar as what a shallow elliptical waveguide would do.

Dave

[dlr]

Re: Do you have knowledge of felt diffraction reduction? Advice needed!

The felt does seem to help around the 5Khz range similar as what a shallow elliptical waveguide would do.

Dave

I think that what we see here is a similarity in results, but from entirely different mechanisms. The waveguide helps because it is controlling the dispersion directly without absorption as evidenced by the increase in sensitivity in the area of control whereas with the felt there may be a small bit (emphasis on small) of horn loading (pretty bad shape for that sort of thing), but is primarily altering the response due to damping of that portion of the wave whose path is in direct contact with the felt.

The single best device for diffraction control is IMO a well-designed waveguide (large enough to provide the needed benefit) since it simultaneously controls the dispersion while having the benefit of increasing sensitivity in its range of control. Baffle edge shaping, felt and driver placement are second.

The exception is for dipoles, of course, which make a waveguide moot.

dlr

[davepellegrene]

Re: Do you have knowledge of felt diffraction reduction? Advice needed!

I'm not so sure that dipole is an exception for a waveguide. I did a build with the AST2560 AMT with a matching waveguide on the front and rear. They had near identical FR graphs which created cancellation at 90 degrees. I would think this would help quite a bit to eliminate side wall reflections. I do not have a lot of experience with dipoles so maybe this is not a good thing but it does look really nice on a 3D polar plot.
Here is a link to it.
https://picasaweb.google.com/1016322...09461178254674

Sorry if I went a little off topic

Dave

[dlr]

Re: Do you have knowledge of felt diffraction reduction? Advice needed!

I'm not so sure that dipole is an exception for a waveguide. I did a build with the AST2560 AMT with a matching waveguide on the front and rear. They had near identical FR graphs which created cancellation at 90 degrees. I would think this would help quite a bit to eliminate side wall reflections. I do not have a lot of experience with dipoles so maybe this is not a good thing but it does look really nice on a 3D polar plot.

Dave

I can't see your pics for some reason.

Were the two waveguides connected in normal phase or inverted? Consider that for a dipole, the rear is out-of-phase and the interaction at the baffle edge is what creates the dipole peak. At and below that point is the dipole operation region.

dlr

[davepellegrene]

Re: Do you have knowledge of felt diffraction reduction? Advice needed!

I can't see your pics for some reason.

Were the two waveguides connected in normal phase or inverted? Consider that for a dipole, the rear is out-of-phase and the interaction at the baffle edge is what creates the dipole peak. At and below that point is the dipole operation region.

dlr

The link should work now.
The sound comes out of the rear of the pleated diaphragm same as the front so I would think they are in phase. There is a perfect cancellation at 90 degrees so wouldn't they have to be in phase in order to cancel?
Looking at the graph the right side is the front the left side is the back and the purple color in the center is 90 degrees. I took measurements from on axis to 180 degrees in 10 degree increments to create the graph.

Dave

[dlr]

Re: Do you have knowledge of felt diffraction reduction? Advice needed!

I see now, it sounded like you had back-to-back horns. Interesting setup. A single diaphragm will have the rear out-of-phase, normal for a dipole. The issue with what the actual dipole operation is. As wide as that is, there won't be much operation as an actual dipole due to what will be a long moment (being a tweeter), though the vertical will have a bit more with the shorter distance.

I see that you're using a passive XO, so there's no dipole correction, which must mean that there isn't any actual dipole operation, tweeter nor woofer. If I'm looking at the response correctly, there is a waveguide type response above about 700K, none below. Above what would a dipole peak, you have waveguide response, but I think it doesn't look like a pseudo-dipole, it looks like a waveguide response.

I understand your thought on trying to get a dipole-like response on the front or rear, I matched the DXT monopole with a true midrange dipole to achieve a smooth off-axis polar response, but it required a very low crossover and still placed the tweeter firmly in the 2-pi area. At 1200Hz LR8, there is zero dipole operation of the tweeter. It does integrate well with the midrange that is transitioning from dipole to monopole at that point, but it's still just monopole. A tweeter in the back out-of-phase with the front would make it appear to be a dipole, but in that case there is no waveguide loading sensitivity peaking on-axis with waveguide drop off-axis. The off-axis axis is almost all simply tweeter diaphragm geometry-based off-axis response.

But I don't see a dipole response. The problem is that most waveguides are going to be too large to operate them in a dipole range. Above the dipole peak dictated by the separation of front-to-back, the response is simply that of the driver above that point. For a waveguide, it's what you show, a waveguide response. It does look like it will be a benefit in near side-wall placements, at least for the tweeter range.

dlr

[davepellegrene]

Re: Do you have knowledge of felt diffraction reduction? Advice needed!

This was more of a satellite set up with a dipole sub. We used a I-nuke amp which has an active crossover to cross to the satellites I believe around 250Hz. It's been a while but I think the dipole peak was around 300 which we was able to use the active crossover to nock the peak down.
Here is a link that shows the build a little better. Post #106 explains the set up a little bit. He was the designer I was just the builder.Post #107 shows some pics.
http://techtalk.parts-express.com/sh...er-Build/page6

Sorry again Dan for hi jacking your thread.

Dave

[Eigenmind]

Re: Do you have knowledge of felt diffraction reduction? Advice needed!

Some really good information and suggestions here guys (about all kinds of stuff)! Hopefully today I will be able to experiment with changing the geometry and aspect ratio of the felt. I will move the side pieces away as well as trying with and without the top and bottom strips. I'll probably use the top and bottom strips as a second stepped layer for the vertical pieces as well.

If anyone has any other ideas, I'm all ears. Of course I will post the results; hopefully later tonight.

Cheers,
Dan

[dlr]

Re: Do you have knowledge of felt diffraction reduction? Advice needed!

Some really good information and suggestions here guys (about all kinds of stuff)! Hopefully today I will be able to experiment with changing the geometry and aspect ratio of the felt. I will move the side pieces away as well as trying with and without the top and bottom strips. I'll probably use the top and bottom strips as a second stepped layer for the vertical pieces as well.

If anyone has any other ideas, I'm all ears. Of course I will post the results; hopefully later tonight.

Cheers,
Dan

The one thing I am sure of is that you need to keep the felt in place between the tweeter and and adjacent drivers. There is almost always more diffraction from a nearby driver than a baffle edge. That piece is the first piece and on occasion the only piece that I place felt. If wide enough, it does help a bit with the portion of baffle edges that are occluded by them.

dlr