very small room - mixes translate a loss in high frequencies

Well, I found the problem. Turns out the default input device was somehow set as.... my.... webam.... Grrrr... :lol: Oh, well.... Strangely, I'm now even more concerned about the results. Apparently - just going by graphics, I'm really not in any better shape... perhaps even worse. Base measurement is in blue, and post-treatment measurement in red. Any thoughts??!! :( Thanks! Chris

Can you describe the difference between the "before" and "after" situations? In other words, what is the room like, in terms of treatment, for the red curve, and what is it like for the blue curve? Maybe you could post photos, showing the differences in treatment between the two. Also, show the position of the measurement mic in one of the photos. - Stuart -

Okay.... quick summary then. Before: After: I got rid of the big desk and the big chair. The smaller desk allows me to have the speakers on either side of the desk on stands (which still need to be covered, but that's fine for now.... this way you can see the "design") Three panels across the back wall. 2x2, 1x2, and 2x2. A pair of panels on either side. 1x3 I have one 18" x 24" panel behind me on the wall, but the matching panel that will go on the back of the entry door is not up yet, as the door is not up yet. Panels are on 1x3 furring strip frames. Semi-rigid fiberglass (similar to OC703) 2" thick in two layers for a total of 4" of fiberglass. Wrapped one layer with 6 mil vapour barrier (sides and front, but not underneath), and wrapped with a breathable cloth (cotton/poly blend). They are stood out from the wall using 3.5" lengths of ABS drain pipe so that the fiberglass is out 4" from the wall. Pics of construction are here: https://picasaweb.google.com/1051514723 ... directlink I built a 4x3 cloud by adding 4" of semi-rigid fiberglass onto a sheet of 1/2" plywood, covered in vapour barrier and then covered in cloth. The front is lower than the back, so it slopes upwards and away from the mix position, ending directly above the mix position. Results: Blue = "before" and Red = "after" - taken at mix position only. By mix position, I mean this: Thanks! Chris

A variation of only a couple of inches in mic position, speaker orientation etc. can cause such skewed results. Since you made erroneous tests previously (webcam I believe) am I correct in assuming you went to the trouble of removing all the treatment again and redoing the tests? If so you probably had to move the mic and speakers. To ensure the mic gets positioned in the same spot, hang a plumbob or other pointed weight on a string from the bottom of the mic and mark an X on the floor with tape and marker. Also use tape to mark the position of speakers if you need to move them. You might also give the position markers names as the next step will be to take readings with different positions in order to find the optimum. In truth I wouldn't bother taking new readings with the treatments removed, just try and find the best listening position. You might also want to try adjusting the positioning of the absorbers - the ones to the sides look rather to far forward to me. Also maybe the angle of your cloud, it should be at least 12°. It also looks a bit too far forward.

Thanks for the reply! I took some measurements.... The right and left sides (as if facing the speakers while sitting in the mix position) are slightly out of symmetry, but not much. I tried to squeeze the right speaker inwards just a smidge to allow for the closet door to the right of it to open and close - and to even get at it - a little better. Right speaker - from wall to center of speaker enclosure = 16" Right panels (affixed to either side of the bifold door) = 23" across starting at 10" from the front wall to 33" from the front wall. Mix position = 39" from front wall. Left speaker - from wall to center of speaker enclosure = 18" Left panels = 23" across starting at 11" from the back wall to 34" from the front wall. Cloud = starts at 13" from front wall and extends 36" to 49" from front wall. Rise over run is 3" of slope over the 36" length, which as close as I can figure, is only 5 degrees. No, I didn't take all the panels down and re-run all the tests. I will if that would be necessary, but that cloud is a brute. The combination of dimension and weight made it *very* awkward and difficult to hold up and maneuver so it could be affixed to the ceiling. Does that confirm anything? What should I try first? Thanks! Chris

What's on the REAR wall, behind you, at the opposite end of the room? And I agree with Brian: I'd angle that cloud a bit more, and get the side absorption pales further back: they don't seem to be on your first reflection points, just from eye-balling the photos.

Cloud = starts at 13" from back wall and extends 36" to 49" from back wall.

BACK wall???? Don't you mean FRONT wall??? The back wall isn't in the pictures, but it sure looks like way more than 13" from the edge of the cloud to the edge of the photo! I see your measurement mic in the second photo, but surely that isn't the position where you had it for the tests? It should be positioned where your head would normally be for a mix session, facing forwards, aimed midway between the speakers and angled maybe 30° to 45° upwards: Yeah, its supposed to be omni, but still you need a standard way of positioning it to, and that's the normal way. Like Brian said: Mark that position very carefully so you can repeat it accurately, to within a half inch or so, for consistent measurements as you change things around. - Stuart -

?Question? Are you freq graphs with a single speaker playing? or both at the same time ?

Soundman2020 wrote:

What's on the REAR wall, behind you, at the opposite end of the room?

I've been wondering if you're onto something there. I guess at the moment, I shouldn't say "after treatment" but more "80% of the way through treatment." On the rear wall, I have placed a single absorber (exactly same design as the others) that measures 18"x24" on the wall above the amplifier and beside the light switch. It is vertically directly in line with the speaker. The light switch itself represents pretty much dead center in the room. See: Now that entry door... is still not up. (been working on panels y'know... haha). I have another 18"x24" panel that I'm *going* to hang on the back of that door, once the door is installed. I have another very small panel (12"x12") to hang just below the light switch as well. Originally, I was thinking that this door wouldn't make much of a difference - and in a larger room, that might well be correct. (of course, I'm just guessing....) But then I started thinking that, if you're in the car with the stereo playing and someone opens a door, the sound changes considerably. Maybe I'm trying to assess strengths and weaknesses prematurely. But then I started thinking that the work I have done so far should at least count for *something*, yet it doesn't appear to. Maybe it's almost like an electrical circuit. You can spend hours hooking up lights, switches, buzzers, wires, timers, and more wires.... but none of it means anything until you complete the circuit. So, to that end, I installed the closet doors last night (see below, right side) and am going to install the entry door today (though time this weekend is going to be a little precious) and see what happens.

Soundman2020 wrote:

And I agree with Brian: I'd angle that cloud a bit more, and get the side absorption pales further back: they don't seem to be on your first reflection points, just from eye-balling the photos.

Even with the measurements provided, you don't think they're at first reflection points? It won't take much to move them, but I don't want to do that if they are actually *at* the first reflection points. I will angle the cloud a little more.

Soundman2020 wrote:

Cloud = starts at 13" from back wall and extends 36" to 49" from back wall.

BACK wall???? Don't you mean FRONT wall??? The back wall isn't in the pictures, but it sure looks like way more than 13" from the edge of the cloud to the edge of the photo!

Yes, front wall. :oops: Fixed in post.

Soundman2020 wrote:

I see your measurement mic in the second photo, but surely that isn't the position where you had it for the tests? It should be positioned where your head would normally be for a mix session,

Erm.... yes... I pull my chair up close to the desk and listen, and the height is about right to my ear. I pull my chair up that close because I'm trying to stay away from the exact center of the room, which is only about a foot past that point.

Soundman2020 wrote:

facing forwards, aimed midway between the speakers and angled maybe 30° to 45° upwards: Yeah, its supposed to be omni, but still you need a standard way of positioning it to, and that's the normal way. Like Brian said: Mark that position very carefully so you can repeat it accurately, to within a half inch or so, for consistent measurements as you change things around.

Okay. Thanks for your response! Chris

RJHollins wrote:

?Question? Are you freq graphs with a single speaker playing? or both at the same time ?

Interestingly, I tried with a single speaker playing just to experiment and found very little difference. Should there be a significant difference? CT

I've been wondering if you're onto something there.

[ Looking at photo of rear wall ] Yup, I sure am! :) Right there is a large part of your problem! That wall needs thick, deep, large, wide, tall absorption! Right now, there is practically nothing on it, and your plan doesn't seem to go anywhere hear far enough. Your room is probably too small to satisfy the 15/15 criteria, maybe possible to hit 10/10, but right now you aren't even getting 5/5. That refers to the amount of attenuation o the first reflections that get to your ear, and the time frame. 15/15 means that the first reflections are delayed at least 15 ms, and are at least 15 dB below the direct sound, which is a good goal for a studio, as the psycho-acoustic effects of shorter delays at higher levels really mess things up in your head. So I'd plan to get a couple of thick (4" at least) panels of absorption on that rear wall, hopefully spaced at least a few inches away from it, and covering as much surface area as you possibly can! Cut a hole on the middle for the light switch, but cover all of the rest!

Even with the measurements provided, you don't think they're at first reflection points? It won't take much to move them, but I don't want to do that if they are actually *at* the first reflection points.

Hard to see for sure in the photos (2D representing 3D, perspective, etc.). But the best way to find out is with the "mirror trick". No, it ain't a course in magic or sleight of hand! :) Just get someone to help you here: get a decent size mirror, you sit in the chair at your normal mix position, and get your helper to move the mirror around all over the front and side walls, and the ceiling. Any place you can see either of the speakers reflected in the mirror is a first reflection point. Mark that spot on the wall, and treat it. Make sure you cover ALL of the surfaces in front of your mix position: front wall, both side walls, ceiling, floor, desk, etc. You might be getting reflections from unexpected places, but the mirror will reveal the truth. Simple and effective! :) - Stuart -

Hey! I feel like I'm getting somewhere! First, I installed the entry door and hung my last two panels, as shown in the pic below. The panels are the same design as the others - two layers of 2" semi-rigid fiberglass mounted on top of a 1x3 frame, covered with 6 mil vapour barrier and covered with a breathable fabric. They are mounted so that the fiberglass panels are 4" from the wall by using a 3.5" length of ABS plumbing pipe as stand-outs. I also hooked up my sub and put it under the desk. Here are the results so far.... I still have to change the angle of the cloud, but I do feel like I'm getting somewhere. I had my wife move a mirror in front of the width of the side absorbers, and there wasn't anywhere that I could see either speaker on either side that was not within the range of the absorbers, so I guess I'm good there. I don't really have a lot of wall behind me on the rear wall to add more absorbers. I suppose I have *some*, but am interested in whatever suggestions. Might I be better taking the middle panel between my monitors (1'x2') and putting it horizontally above the light switch? I'm a bit reluctant to hang anything else on the back of the door, as it is already limited to opening to about a 70 degree angle, at which point, the absorber hits the wall behind the door preventing it from opening further. I doubt it makes much difference, but I still have to add the trim around the door, so there are some gaps around either side of the door where it was shimmed out to fit the opening. - roughly 1/4" on either side. Thanks again! Chris

Bit of an update here.... I guess I'm making some progress.... I still have to get some chain so I can change the angle of the cloud, but I have made a couple of changes... 1. I moved my side absorbers and moved them closer to me by about a foot. In my small room, that's pretty significant. It also allows me to open my closet door to get at my guitars without having to take down the panels, because one of them is now on the wall, rather than being pushed backwards by the bifold into the absorber on the wall behind the speaker. In and of itself, this change really didn't make a difference in my graph. 2. I removed the 1 x 2' panel from between the speakers and put it on the wall behind me. Between the two changes, this seems to have flattened things out somewhat. The blue graph represents the room before any treatment. The green is what I measured today *before* making those changes above. (essentially, reflecting the same environment as in my last post a couple of weeks ago) The red is what the room "sounds" like now with those changes. A couple of observations: 1. The blue graph, before any room treatment, shows the graph almost flat from 3200hz to 17khz. (within +/-1db) Now, I have a dip and a peak representing +/-4db 2. The extreme dip and peak between 1k and 4k has been tamed. In fact, from 117hz to 4.7khz, there are only two wee spots that is not within +/-3db. Mind you, the green graph, from before I moved that one panel behind me from between the speakers was within +/-3db from 113hz to 12khz. 3. That huge dip around 100hz just won't go anywhere. 4. Moving that panel from between/behind the monitors seems to have helped tame some of the mids between 1.5k-4k and taken down a bit of the peak around 15k, but has messed up the area between 4k and 7k. Maybe I should build another panel to put back between the monitors on the wall in front of the desk? I have done some basic recording in the room before I made the changes today (basically representing the green graph), and it sounds a lot better. It's not boxy sounding anymore. :D However, mixing yielded very similar results to what I had before I did any treatment. And by very similar, I mean almost imperceptible. :cry: Any other suggestions? Thanks! Chris

To be honest, by moving treatment around from one location to another you are not fixing issues; you are just moving them. You stop treating one mode in order to start treating another mode, and you change the reflections, comb filtering, phasing issues, and such like. It's a different room after you do that, so I'm not surprised that the graphs are different! I'm also not surprised that the 100 Hz problem did not change: I strongly suspect that one is associated with the vertical axis of the room. Moving things around on the front back and side walls won't help at all with a vertical mode. I'm hoping that changing the cloud angle will help a bit, and I'd at least expect to see SOME change on the graphs from doing that. I also get the impression that the mic may have been in a different position for the blue graph, and then the other two: It seems to me that the response changed too much to account for the amount of treatment that you put in. Also, waterfall plots are far more use for judging where your modal issues are and seeing what other problems you have. A simple frequency response curve is just a snapshot at one specific millisecond in time, whereas what you REALLY need to know is how the response changes over a period of several hundred milliseconds. That's the key. So it would help if you can post the waterfall plots for each of those readings. Your 100 Hz problem might need a panel trap, but first I'd get the absorption sorted out as much as possible. - Stuart -

Soundman2020 wrote:

To be honest, by moving treatment around from one location to another you are not fixing issues; you are just moving them. You stop treating one mode in order to start treating another mode, and you change the reflections, comb filtering, phasing issues, and such like. It's a different room after you do that, so I'm not surprised that the graphs are different!

Okay, fair enough. I guess I was just hoping that better placement would mean better results.

Soundman2020 wrote:

I'm also not surprised that the 100 Hz problem did not change: I strongly suspect that one is associated with the vertical axis of the room. Moving things around on the front back and side walls won't help at all with a vertical mode.

Good point.

Soundman2020 wrote:

I'm hoping that changing the cloud angle will help a bit, and I'd at least expect to see SOME change on the graphs from doing that.

Okay, I made two changes, though. I changed the angle of the cloud and it is now 11 degrees. I also built another 1x2 panel and replaced the one between the monitors against the front wall. (the one that had been there got moved the other day to behind the mix position) Here are the results: The green graph is the measurement from the other day. The red is today. The 100hz issue hasn't changed in the slightest, really, but throughout the rest of the graph, the curve seems to have been moderated towards the middle by about 1db throughout much of the frequency range. It doesn't look like much, but I guess it's good. Now, comparing today's results to the pre-treatment, the results are considerably more significant. The blue represents pre-treatment, and the red is today's measurement after re-angling the cloud and putting up the new panel on the front wall between the monitors. I'm still unimpressed by the big sweep above 4k, but below that really looks considerably better.

Soundman2020 wrote:

I also get the impression that the mic may have been in a different position for the blue graph, and then the other two:

Yeah, surely it moved a little, as I had to move the desk a little further from the wall. It was only a couple of inches-ish, but I realize that makes a difference. Lately, though, my measurements have been pretty consistently placed.

Soundman2020 wrote:

It seems to me that the response changed too much to account for the amount of treatment that you put in.

Really? Okay, this surprises me. I guess maybe I was expecting more than what I have gotten from this whole process, but it seems that, from your expectations, that this has been more successful than I should have expected? I mean, I'm delighted that I can record something in here and have it not sound boxy anymore, but with all the effort I have put in so far, when I did another trial mix the other day, the results were almost indiscernible from the pre-treatment mix. That was really disheartening.

Soundman2020 wrote:

Also, waterfall plots are far more use for judging where your modal issues are and seeing what other problems you have. A simple frequency response curve is just a snapshot at one specific millisecond in time, whereas what you REALLY need to know is how the response changes over a period of several hundred milliseconds. That's the key. So it would help if you can post the waterfall plots for each of those readings.

Ah! Awesome. Here they are... though I'm a little sketchy on how to interpret them....

Soundman2020 wrote:

Your 100 Hz problem might need a panel trap, but first I'd get the absorption sorted out as much as possible. - Stuart -

Great, thanks again for your help. It really is appreciated. CT

Doing a little reading here... found out some close-ups might be helpful. ^ Pre-treatment ^ two days ago ^today Chris

FWIW, here is a closeup of the SPL graph with no smoothing from 21hz to 312hz. The blue is pre-treatment; the green is two days ago; and the red is today. That big dip in the green suggests that, at that particular frequency, that the treatment made things worse by 12db and moved the trouble frequency from 125hz to 104hz. Changing the angle of the cloud and putting in the extra panel improved the 104hz dip by 14db (thus, now 2db better than it was pre-treatment), but still at least another 14db from the range of many of the other frequencies. I don't know if that's helpful or not, but it serves as an additional point of reference for the waterfall plots above. Chris

If you're treatments are having little or no effect, and you're sure the listening position is not in a null, then it could be one or both of your speakers may be in a null! Adjust the speaker - LP positioning and check again.

It could also be that the speakers themselves do not have flat response. Have you tested them outside, in the open air, far from any reflective surfaces? If not, maybe you could do that, just to eliminate one possibility from the equation. - Stuart -

An interesting development.... I tried moving the speakers around. I was able to get some improvement by moving them a little further apart (they were about 46" apart, now about 58" apart), and moving them back closer to the front wall. They were 13" from the front wall, now they are about 10" (and only about 2" from the absorbent panels). They are both now closer to the side walls too... much closer to the corners now, which flies against conventional wisdom, but the results are somehow better. Also, whereas the speakers were before about 32" from the listening position (cone to ear), they are now more like 36". I also moved the "listening position" back about 5 inches or so, to the mathematical 38% mark. Here are the results: ^ waterfall from two days ago ^waterfall today. ^SPL today - purple is two days ago, red is today. Some marginal improvements, but mostly just tradeoffs. But I moved the mic back to the former mix position of the last number of tests, and it is even more interesting, but I'm still not sure what to make of it. ^waterfall today with mic moved back closer ^red = mix position moved back to 38%, green = mix position moved back up to where it was for the last few tests. It seems that moving the speakers and the mix position around is making some differences, yes. Maybe now it's a matter of playing with these to find the best balance of trade-offs? Those significant dips and swells aren't really going far, for the most part. Given that my mixes lately tend to sound lacking in the high end, maybe I should be happier with a graph that shows the top end a little more understated so that I'm encouraged to push the highs a bit more. I don't know.... As far as trying the speakers outside... I suppose I could link a couple of long mic cables together to do a measurement, but I'd have to buy some more speaker wire to get the speakers around the corner, up the stairs, and out the door. There is a window in the room, but the previous owner nailed and siliconed it shut. CT

FWIW, I've been providing SPL graphs smoothed at 1/3 octave. Smoothed to a 1/2 octave, the difference my treatment has made is a lot more encouraging, but is smoothing to that level "cheating?" CT

Hey there! FINALLY! SOMETHING THAT FEELS LIKE SIGNIFICANT SUCCESS!! Okay, here's what I have done. I figure that, recording in the room sounds good. That means that my treatment is very much in order. Mixes coming out wonky must (mostly) be (as was suggested) speaker placement. I moved them this way and that and back again and to the middle and this way and that again, using various combinations of speakers standing and speakers on their sides and ultimately settled on what I wasn't expecting: Woofer domes about 45" apart. Distance from dome to ears = about 27". Basically, getting them as far as possible out of the corners of the room without sending me back to the middle of the room, and with me being close enough to the speakers to minimize as much as possible the actual "sound" of the room, with that big nasty node around 100hz. (speakers standing) I ran a few tests and had some significant improvements. There were still some trouble spots, so I adjusted the volume on my sub, and for the first time tweaked the EQ on my power amp - 100hz up just a hair; 1khz up about 1/3 of the way; 10khz up about 1/3 of the way. ^ green = two days ago; red = today ^ blue = before ANY treatment; red = today Far from perfect, but a damn site better. There are still a couple of crazy dips, but not nearly as bad as they were (the worst one covers a distance of about 7db instead of 14db!!). The big drop off at the top end actually makes me happy (I think....) because it's the top end of my mix that tends to be dull. Having it drop off (yeah, even with some EQ on my power amp) will encourage me to add a little more. Perfect. (I hope I'm right about that....) The waterfalls look encouraging too. The RT60 time looks better, as does the distribution across the bottom end from 20-260hz. ^ waterfall two days ago ^ waterfall today Now that I finally seem to be getting somewhere, what could my next step be? Am I right about my thoughts about where I am right now? Any other guidance? Appreciated, as always; Chris

waterfall two days ago [SPAM SIGNATURE DELETED BY MODERATOR]

Update: I'm considering the next step towards improving sound in this space. I'm looking at getting a sub to power my Yorkville YSM-1p monitors. This thread contains the size, shape, construction, and frequency response measurements of where I am at now. Would a sub help matters in my room, or is my room too small for a sub? Thanks! Chris

Chris, you should probably post a link here to your other thread, and vice versa, so they are tied together. Otherwise you might lose track of what is where! And people who are following your thread might get confused too.... - Stuart -

Good point.... See "use of a sub in a small room" here: viewtopic.php?f=12&t=18682 CT