[Kevin P]

I've been designing an acoustically transparent screen for my "forever in progress" HT room.    I thought I'd post some of my measurements of the various screen materials.  

Note, that I'm not going to comment on the visual performance of the screen materials.   I'd need a full size sample of each to give them a proper subjective evaluation.   Also.. I'm not a video projector guru.   There are other people more qualified to measure and talk about the video performance.   I'll give my simple subjective opinion of my screen when the project is done and some antidotal comments on the screen materials as I talk about each.

The measurement set-up consists of Praxis with my calibrated mic.   It's good for about -+0.5db and is more than accurate enough for speaker design and these measurements.    The mic is situated on-axis 1.5M from my reference speaker.   I disconnected the woofer network since it wasn't meaningful data for the screens.   All of the screens become completely transparent to sub 500Hz output.   Each screen sample was cut to identical size and hung in front of the reference speaker in the same manner with the same standoff of about 2 inches.

Here is the reference speaker.  It's our LCR center channel, which I use behind the screen.   Normally it has massive amounts of output to 25hz in-room but we have the woofers disconnected for this test.     You can see from roughly 2K-20K (the area we are going to focus on) it is -+1.5db so it's flat within the range of accuracy of our mic, production tolerances between drivers, crossover components and cabinet variability.   It was designed with a little bit of high frequency lift in anticipation of using it behind the screen.





As another frame of reference, I took a measurement of standard grill cloth.   You will notice about 0.5db attenuation at 2K and 1db at 15K, which isn't much.   You will also notice it doesn't cause any other problems with the FR of the speaker.   This is a bare piece of cloth hanging in front of the drivers.   Virtually all speaker grills have a frame, which will probably cause more diffraction problems than the grill fabric alone.    That is a discussion beyond the scope of this write-up and the grill cloth is given purely as a reference.




Here is our first screen sample.   This is a Da-lite High Contrast Cinema Perf.    It is a rubbery vinyl material with the screen coating on one side and the other is black.   It reminds me of a thin mouse pad.   The screen is made acoustically transparent by perforations.   The base material is thick and Da-lite says it has a gain of 1.1with a 45 degree viewing cone.

It should be noted that my sample of Da-lite was physically smaller than the other samples.   It still covered the two drivers I was measuring but this was the one sample that I had that wasn't identical in size with the others.  

Notice we are about -+5db, which is kind of ugly.   There is also a downward trend of rolling off the top-end.   I'd say we are down about 7-8db by the time we reach 20K.   I'd not call this inspiring performance from an acoustical standpoint.




Here is the Vutec sample sent to me from Jason at AVS.    The Vutec is woven sample that looks like a standard white unity gain screen.  I'd guess it's a gain of about 1.0 and the perforations are small enough not to be visible from anywhere past 5-6 feet.    It is very close in color to the Screen Research and Danzian fabric.  

I measured it with and without the Screen Research black backing material.   Screen Research sells their product with a black backing material that is designed to cut-down on reflected light, which penetrates the screen.   With the more transparent screens (Danzian) it's helpful to have a black material behind the screen to prevent light reflection and insure speakers, braces and other objects are not visible during movies.

Here is the measurement without the Screen Research backer.



Here is the same Vutec material with the SR backing material.




Some comments are in order.   The Vutec alone is an improvement over the Da-lite.    I'd say we are now -+3.5db and we don't see the same degree of high frequency roll-off.   There is maybe 2-3db of total roll-off by the time we reach 20K.    

From these measurements the Vutec with the SR backing is actually smoother than without the backing.   We have an extra 0.5db or so of total attenuation but we are now about -+3.0db.

Next is the Screen Research CP2 material.   John Caldwell @ St. John Group who represents the Screen Research in North America was kind enough to provide my sample.

The SR CP2 is woven like the Vutec.   It differs in physical appearance by having a more complex weave pattern.   The Vutec is a straight up-down weave on the X-Y axis.   The SR sample has a horizontal weave with fewer apparent holes in the material.   The two samples along with the Danzian are similar in color with the Danzian being the closest to pure white.    

Once again I took measurements of the PC2 with and without the black backer.   Here is the first measurement with the black backer.



I'd say we are about -+3db and there is about 3db of attenuation from 2K to 20K.

Here is the measurement without the black backer for reference.



We have less total attenuation and we are still -+3db or there about.

Just for kicks here is the black backer alone with no screen material.



This is about -+2db with 1.5-2db of attenuation sloping from 2K-20K.

The final measurement is the Danzian CCC, which is popular among the DIY screen building crowd.    It is lighter than any of the other samples.   It is a cloth weave that has less stiffness than the other samples.   Stretching on a frame is required.    The weave is less noticeable than either the Vutec or the Screen Research.   I'm using black speaker cloth behind this screen to help with light transmission and reflection.




This is about -+1.75db across our measured response.   It has little total roll-off from 2K-20K.   I'd guess about 1db of loss.    

The Danzian is a lighter material so I'm guessing that is the largest reason for the better performance.   I'm also guessing that it will allow more light to pass through decreasing total gain and increasing the need for controlling light behind the screen.   I don’t have the equipment to measure light so this is just speculation on my part.    I have a room with total light control but a relatively low lumen output DLP projector.   I’ll update how the Danzian CCC is working for me once my room is done.

[Vtech2000]

Excellent report, let me encourage you to keep up the great work.

[Kevin P]

Excellent report, let me encourage you to keep up the great work.

Thank you.   I don't know of anywhere else on the web where someone has actually measured the screens.

[Vtech2000]

It is exactly this sort of inovative, original work that makes Audio Circle so appealing and interesting.

[bald]

Thank you.   I don't know of anywhere else on the web where someone has actually measured the screens.

I've followed the front projection crowd at AVSforum quite closely for the past 1.5 years or so, and I am not familiar with any information of this kind. I think it's because the audiophiles and videophiles over there are fairly distinct crowds, and projects that attempt to excel in both areas are few and far between.

I have to say your measurements are interesting, although without having personally viewed any of the materials, I imagine the manufacturers of the professional perforated screens will claim that their materials will yield a superior picture to your more acoustically transparent DIY material.

In any event a very interesting project. The effects you describe with your Danzian material, you think those changes on the FR will be audible? I don't have a good grasp on how much change is significant, and what can be absorbed without much detriment.

Be interested to what you think of the subjective video performance of this material, maybe in comparison to something like blackout cloth?

bald

[Kevin P]

In any event a very interesting project. The effects you describe with your Danzian material, you think those changes on the FR will be audible? I don't have a good grasp on how much change is significant, and what can be absorbed without much detriment.

I don't think they are very significant.   I've exagerated the scale of the FR graphs so that we could analyze the differences.    If you measure from a listening position in a room the response is pretty ugly.   You get comb filtering effects off reflections from anything in proximity to the mic.    Much of this is more acedemic in nature than anything.    

Speaker designers strive perfectly smooth response but its good practice to ask these types of questions and keep some perspective.    I don't have an answer for you though.   I think it's obvious your loosing some high end in a couple of the samples.    My wife wouldn't notice though.  

My perspective is that the frequency response should be corrected for if possible.   For me it only takes a few measurements and some adjustments on the crossover to compensate so why not do it?   If my speaker response looked like one of these samples I'd not be happy with it.   I'd rework the crossover, change the cabinet or choose different drivers until it was correct.    Why  should we expect anything different from a screen application?  

[Vtech2000]

My perspective is that the frequency response should be corrected for if possible.

That was my immediate response upon reading your report.  If one has it within their means they should desire to do this as much or more as any tuning, tweaking or other system improvement effort.

[Kevin P]

My perspective is that the frequency response should be corrected for if possible.

That was my immediate response upon reading your report.  If one has it within their means they should desire to do this as much or more as any tuning, tweaking or other system improvement effort.

I've never seen a tweak that is measureable like this.   They don't have this kind of effect on the frequency response so I'd list it above most tweaks I've been exposed to in order of importance.

My personal priority list when engineering a system would be to spend money on getting the speaker/room interface correct first.    That is where you have the largest errors in playback and get the most profound improvements.   After you get the speaker-room interface licked then it makes sense to start chasing other improvements.

Some things make noticable changes but are not measureable in terms of FR, or their measurements are not well understood.    Examples of this would be amplification.   I have two amps which when used for FR measurements give me the same results yet they sound significantly different.    I think it's probably due to their distortion spectrum and output impedance.     Just a guess though.

Another area of voodoo is vinyl.   I don't understand why vinyl sounds as good as it does.    I work with Bill Firebaugh of Well Tempered Labs fame and he has some explanation about 2nd harmontic distortion created by the way the sylus moves through the groove.    It may be the 2nd harmonic introduced by the mechanical system that gives vinyl such a nice subjective sound.

Another area of improvement without respect to FR is some (well designed ones) preamps and phono stages.    Once again I think its the distortion spectrum.   I like a dose of 2nd harmonic served up in my music.  

[Kevin P]

Ok... round two of measurements.  

From the last round of measurements I had a couple of general questions.  

#1.  How is the small sample size effecting my measurements?  

You get reflections off any object close to the speaker.   This causes comb-filtering effects that cause bumps in the FR.    The question I had was to what degree are the small sample sizes responsible for the frequency responses measured?    As you will see from the measurements I think that the sample size was a large part of the reason for at least some of the FR problems.

#2.   How does speaker distance from the screen affect FR?

Some of the commercial screen companies state that you need to stand the speaker off by a certain distance.   Once again, you get reflections from all objects and those close to the screen or mic can cause various degrees of comb filtering.


Let's tackle question #1 first.

I set up the entire screen today with a layer of speaker cloth behind the Danzian fabric.   I took several measurements with the same speaker.   The mic for these measurements was 44" from the front of the screen.   The speaker for this measurement was 3" behind the screen and the graph is 1/3 octave smoothed.     You cannot directly compare these with the other measurements because I took them at different locations and at a different time.   It would have been nice to do all the measurements at the same time with the same mic location to eliminate set-up issues but from looking at these measurements of the speaker behind the full size screen there are absolutely no issues with this FR.   It is about -+1.5db, which is about as good as it gets.




I have to believe that my small sample sizes where giving me exaggerated FR problems.   Diffraction from the material edges was probably the main problem.    The smaller sample size Da-lite material had a rougher response so that would correlate with this conclusion.   I'd need full size samples of the other materials to see exactly how they perform.    My intuition says that they would smooth out considerably with the full-sized screen.

#2.   How does speaker distance from the screen affect FR?

For this I set up my mic at 44" and started with the speaker spaced 3" from the rear of the screen.    

These measurements are unsmoothed raw data so we can see trends.



I moved the speaker back to seven inches for this measurement.



And finally here is a measurement from twelve inches.




The trend is obvious.   With greater distance we get less comb filtering effects and they are smaller in magnitude.   The peaks/dips become more tightly packed as you space the speaker back further.

The natural question is how audible is this?    Well, the ear has an effective 1/3 octave smoothing behavior.    Looking at raw data is great for the trends but looking at a 1/3 octave smoothed graph gives us a better idea of how significant are the small FR aberrations.

Here is a one-inch spacing smoothed 1/3 octave.



Here is a three inch spacing smoothed.



Here is a six inch spacing smoothed.




I'd say the spacing isn't an issue with the Danzian material.   I'd feel comfortable running this speaker 1.5" from the rear of the screen.    The improvements you get from running it at a greater distance are not likely to be audible.    

In conclusion to question #1 we have shown that my small sample sizes caused FR problems that where related to sample size, not the screen material.    I'd need larger samples to get meaningful measurements.

Our conclusion for question #2 is that speaker distance does effect FR but not much with the Danzian fabric.   Run it as close to the screen as is practical for your application.

If anyone has any more questions about these measurements or if you see any problems I may have overlooked please feel free to ask or comment.    I'll take some more measurements if there are any unanswered questions.

[Vtech2000]

No further questions here, the follow up work is done well and is pretty conclusive.

Great job on a very interesting project!  Answered a lot of questions.