DIY Sound-absorber panels

A while ago I built new crossovers for my floorstanders and while performing the measurements I noticed some nasty behaviour between 100-200Hz.

Frequency response with the old crossovers, 1/12 smoothing

I was suggested that it was probably because of something called SBIR (Speaker Boundary Interference Response). In short that means that the lower frequencies start wrapping around the speaker hitting the rear wall and then being reflected with a 180 degree phase shift and then interfering with the signals coming directly from the driver. The linked article from GIK Acoustics mentions treating the wall behind the speaker to try to absorb the sound that wraps around the speaker so you get less reflected sound coming back to interfere with the orginal signals from the speaker.

I had a look at buying absorber panels from different companies but they were all so expensive so I figured I could probably do something myself for much cheaper (and it's fun to know you did it yourself). I looked into what is important when chosing the right material and what thickness I would need. Of course, the thicker the better, but there's only so much space between the speakers and the wall and since this is not a dedicated listening room but a living room I do need to think about aesthectics too. For low frequency absorption and thin panels (less than ~100mm) a dense material with high flow resistivity is prefered. I had decided to go with a 70mm panel since it would still look OK and it's easy to find panels and wood for the frames in 70mm thickness.

Flow resistivity is not something that a lot of manufacturers list in the spec sheet for their products. I had a look at different brands of rockwool insulation and couldn't find anything helpful. After some further googling I found this thread with some helpful data on flow resistivity for insulation products here in Scandinavia. Regular fluffy rockwool insulation seems to have values of around 10 kPa*s/m2 while the denser ground insulation slabs seemed to have values of around 40 kPa*s/m2 and up.

There's a great porous absorber calculator here at Acousticmodelling which turned out to be super helpful. I compared two absorbers, both 70mm thick and with 20mm to the back wall. 20mm is the thickness of the baseboard in my living room.

Model of two different absorber panels with different flow resistivity.

The green line is with the denser material and the blue is with the fluffier material. If my goal was to treat reflections anywhere in front of the loudspeakers I would have gone with the fluffier material since it is way better from around 250Hz and up. However, now I'm trying to treat reflections behind the speakers and for that I need good absorption in lower frequencies because those are the ones that wrap around the speaker and my main problem area seems to be between 100-200hz and in that range the green line (dense material) wins.

I ended up buying a pack of three Rockwool 70mm thick 600x1200mm ground insulation slabs. Then for the frames I bought two 4.2m long 28x70mm bits of wood and cut four 1.2m pieces and four 0.656m pieces. I hadn't decided on what fabric to wrap them in yet and since I wasn't too confident about my skills in wrapping frames with fabric I bought the cheapest stuff I could find at Ikea. It was a non-bleached cotton fabric and I started with buying just enough for wrapping one frame once.

One frame being wrapped with fabric.

Result of first attempt at wrapping a frame. Not pretty but the best I could do.

Wrapping the frame wasn't as difficult as I thought it would be. No problems at all getted the fabric stretched evenly. However, the corners had me confused and I also realised that I should probably get the back side stapled BEFORE wrapping the sides over. I ended up going with this triangular deal for the corners. I had a feeling that there is some other nicer way to do it but at the time I couldn't figure it out. Turns out that when I did the very last corner on the second round on the second absorber I finally figured it out. But then it was too late so I did the triangle anyways. Learn by doing! By this stage I had realised that the dirt cheap fabric from Ikea matches the colour of the wallpaper in my living room really well so I went back to the store to buy another three pieces of fabric so I could wrap the first panel with one more layer and then do two layers on the other panel as well.

Left panel with one layer of fabric, right panel with two layers.

Both panels finished.

Both panels finished.

The panels blend in nicely with the wall and don't attract too much attention. I had first planned to go with a coloured or more textured fabric but I'm glad I did it like this. As I mentioned previously my intention was to place these all the way against the back wall with only a 20mm gap for the baseboard down by the floor. Turns out this doesn't work because behind the right speaker there's a power outlet that stick out quite a bit. I ended up having to move the TV-bench and the speakers forward a bit to make enough room. Now the panels ended up 70mm from the wall and this changes the efficiency of them a bit.

Blue line is with 70mm air gap, green with 20mm air gap.

Not a big difference but it does give slightly better absorption below 200Hz. Good for me, because that's exactly what I needed. However, if we compare this new result with the fluffier material from before we get a different picture.

Blue line is the dense panel with 70mm air gap, green is fluffy with 70mm air gap.

Now my dense panel gets beaten by the fluffy panel from around 150Hz and the fluffy panel is way better from 150Hz up to 2-3kHz. This shows that you really need to measure and model BEFORE buying your material. Depending on your situation you might need something different than what you first thought.

Also, something that I didn't consider beforehand but I realised later is that a fluffy panel may sag a little from gravity over time. I'm not sure if it will, but the denser ground insulation panels are incredibly rigid and shouldn't deform over time. That could be something worth thinking about too. Weight is also something to consider if you want to mount the panel on the wall. Ground insulation slabs are HEAVY! My pack of three slabs weighed in at over 20kg. A fluffier material would way much less and be much more wall-friendly. I will end this post with some measurements with/without absorber panels and in the very end I will make a parts list and write the total cost for everything.

Left channel measured at ~1m on-axis with and without absorber panel. Very small difference in frequency response. 1/12 smoothing

Left channel RT60 with and without absorber panel. The biggest difference is between 150-450Hz, as
can be expected since the sound starts to wrap around the speaker at ~500Hz according to the GIK Acoustics article. 

Frequency response at sweetspot with both channels. Red is with absorber, blue is without.

RT60 at sweetspot with both channels. Not as big of a difference here but there is at least a slight difference.

Parts list:

• 1 pack of 3 70mm 600x1200mm Rockwool ground insulation slabs: 519SEK (~€50/$58)
• 2 pieces of 4.2m 28x70mm wood: 66SEK (~€6.50/$7.40)
• 4 pieces of 1.5m "Bomull" fabric from Ikea: 115SEK (~€11.30/$13)
• 1 pack of staples: 59SEK (~€5.80/$6.60)
• 16 wood screws: 0SEK (had those already)

Total cost: 756SEK (~€74/$85) and then I have one slab left so I could build a third absorber for only ~90SEK (~€8.80/$10). So price per unit turned out to be 378SEK (~€37/$42.50) when building two or 282SEK (~€28/$31.50) if building a third one.

Tools required:

• A saw to make accurate 90-degree cuts (or 45-degree cuts if you feel ambitious).
• A drill to pre-drill the holes for the screws. Makes assembly a lot easier and with less risk of cracking the wood.
• Screwdriver
• Staple gun
• Scissors for cutting the fabric.
• Earplugs (recommended because staple guns can be quite loud).

Time required: It depends on how good tools you have, how good you are at using these tools, and how skilled you are in general. I had a terrible saw and did all the drilling and assembly in the small kitchen in my apartment and wrapped the frames on the largest open space available in my apartment. I probably spent 2-3hrs on each panel. If you are more skilled than me (which isn't very unlikely) and have a better workspace you could probably finish a panel in 30-45 minutes. 

Experimenting with a quieter power source for the MiniDSP

As exams are approaching I find myself spending more and more time by my desk with faint background music on as I study. As I lean forward over my books my head comes within 40-50cm from the speaker drivers so any noise will be very audible.

I have been running the MiniDSP hooked up to my system even though I am not using any subwoofers at the moment. This is because the mid bass becomes a bit overpowering in my room as soon as I turn up the volume a bit. I have been using a small wall wart as a power supply for my MiniDSP 2x4 for years now but the noise haven't bothered me at all since I haven't been listening near-field. If I sit a couple of meters away from the speakers I can't hear any noise at all.

I did some googling on the problem and found a thread where some guy suggested powering it with a 9v battery to see if it was the wall wart that was introducing the noise, because apparently batteries are supposed to be a quiet (but not very convenient) power source.

By chance, I happen to have some 9v batteries lying around from my previous O2+ODAC builds.

The sacrifical cable to the left, a 2.5mm male to 2x2.5mm female cable.

I also had a spare cable which I ordered just a couple of weeks ago. The plan was to use it as a replacement cable for my Koss Porta Pro headphones, but I realised I should get something sturdier (which I have, just waiting for my heat shrink to arrive from China).

I started by cutting off the useless 2.5mm female connectors. I mean, have you ever used one of those? Well, I haven't.

Shielding/ground and signal wire.

Apparently the cable had a shielding/ground wire and a signal wire, which makes sense after all. I decided to use the outer ground wire since it was thicker and I didn't have to peel an extra piece of insulation.

The next problem was how to connect the wire to the battery, without holding it in place with my hands. One of the connectors on the battery allows for winding the cable around it but the other one does not. But as it happens to be, batteries are not the only things I get left after building an O2+ODAC combo. I get spare battery connectors as well. Might as well sacrifice two of them for this experiment.

Wire soldered to a battery connector.

This way I can attatch the cables to the battery without holding them in place. However, another problem appeared. These connectors are meant to be soldered in place on a circuit board, fixing them in place so they can't rotate and touch eachother which creates a short circuit. My solution to this was electrical insulation tape.

Tape-wrapped and insulated battery connectors.

I used my trusty soldering iron to melt the tape in a few places to fuse it together so it wouldn't fall apart as easily.

Finished "product".

After finishing the battery connectors I simply connected the other ends of the cables to the MiniDSP. No fire, no smoke, no weird sounds. All good. Next step - connecting it to the amplifier. No fire, no smoke, no weird sounds. All good so far. Turn on the amplifier and wait for the "click" that turns on the sound. Dead silent, no background noise. Wow, did this really work? I turn on some music but the speakers are still dead silent. I unplug the battery and plug in the USB cable. Background noise appears again along with the music.

My first thought was regarding the age of the battery. I think the battery I tried was from my own O2+ODAC build several years ago. I take it these batteries are not high-end and maybe they lose charge by time? I fetched my DMM to check the voltage.

3.91 volts, the MiniDSP 2x4 requires at least 4.5 VDC.

Turns out I was right, the battery had lost too much charge. But as it happens to be, I had seen some other batteries in my box full of newer O2+ODAC supplies while I was getting all my tools which where in the same box. I fetch the newer batteries and measure them. The first one measured around 7.5 VDC and the other measured 8.83 VDC as you can see in the picture below.

8.83 VDC, the MiniDSP 2x4 can handle up to 24 VDC.

I connect the new battery to the MiniDSP, I connect the DSP to the amplifier again and as I turn on the amplifier and hear the "click" I can instantly hear the background noise again. I put on some music to see if it actually works with a battery and it did work, however not any better than the wall wart.

I am a bit disappointed with the results, I was really hoping to get rid of the noise with a quieter power supply. I guess it must be the MiniDSP itself making the noise and I will have to delve deeper to try to find a solution to this, because otherwise I wont be able to use the MiniDSP in my current setup.

An evening of experimenting and measuring

This evening I have been experimenting further with the room's acoustical properties.

I started with moving one of the subwoofers forward about 1 meter and 0.5 meter to the right so it's against the wall. It is still facing forward though.

Unfortunately I haven't saved any measurements and I can't be arsed to make new ones now, so I might add a picture tomorrow or some other day. The big differences were that the problems at around 100Hz was reduced drastically and the peak at ~50Hz reduced slightly. Of course I added a time delay to the subwoofer so they are acoustically aligned.

I used the EQ-function in REW to get the proper values for the EQ biquads. REW's graphical interface is still going crazy even though I have updated Java and my grahpics card drivers. I don't know why, but it's annoying as hell. I managed to export the filter values though and I imported them to the MiniDSP. The Q of the EQ bands were much lower now than before, which of course is good, and in general less EQ was needed. I got a much more flat response now and another "dimension" has been added to the music now that I have the 100Hz region playing along as well.

I think I mentioned before that I had an electrical buzzing sound from the Behringer and that I was going to send it back. Well, I haven't heard any buzzing at all today. I will wait with sending it back, the shipping label is valid for another ~3½ weeks so there's no hurry. And if I shall send it back I will wait a couple of weeks and send it just before I go to Germany for ten days.


Another thing I did was to see the impact on the sound when adding lots of pillows and cushions to the room. I did two measurements, one without pillows, and one with pillows.

Without pillows the room looked like this:

And this was the measurement.

If you wonder why the subwoofers are still in the same place as before it is because I did this experiment before I started moving the subwoofer. The measurement is with the subwoofers playing and the EQ active. Take note on how the bass is somewhat flat actually. Ignore the black line, that's just the microphone calibration curve which I forgot to hide.

I then proceeded with filling the room with pillows. I couldn't get the whole room in one picture but there are more pillows to the right which aren't visible.

And this is how it measured.

Much less phase issues than without the damping, and the dips in the midrange are not as severe as before. High frequency levels are slightly reduced, not much though.

Both at the same time. Blue before, green after. No smoothing.

1/24 smoothing. Blue before, green after.

Conclusion: Did it sound better? Well, I'm not sure. Arbitrary damping might be good and it might make things worse. The measurements show that the damping reduced some phase issues and reduced the amplitude of the dips in the midrange. Another thing I did not include was reverberation time. I don't know how to measure that so therefore I decided to leave it out. Though my ears could tell me that there was much less reverberation with the pillows than without them.

Hours with REW and the MiniDSP

I've spent most of the evening playing around with REW, the MiniDSP and my UMIK-1.

I started with measuring the response. Pretty flat since I had already set the main EQ stuff a couple of days earlier. But there was lots of room for improvement and I still have some nasty stuff going on at ~90-400Hz.

Today I have tried different crossovers, inverting phase and fine-tuning the EQ. This is the result so far, presented with 1/24 smoothing.

Right now the subwoofers are crossed over at 120Hz with a 24dB/octave Butterworth filter. The main speakers are crossed over at 80Hz with the same type of crossover as the subwoofers. Might be it's better to set it a few Hz higher to flatten the tiny peak at ~84Hz. And by the way I don't think the graph can be trusted below maybe 30Hz because right now the fans of the Behringer amp is blowing straight towards the mic (I have the muff on) but the background noise is slightly high because of it.

I also noticed an electrical buzzing sound coming from the Behringer, from behind the power button. It doesn't sound very healthy so I'll contact Thomann about it. I don't want to switch fans if I have to return the amplifier.

Speaking of fans I have decided to use Fractal Design 1400rpm 80mm fans. I bought one for cofffee-money from my friend Robin and I have ordered a second one from CDON. Hopefully it will arrive before the weekend. While searching through my boxes for some cables I stumbled upon two pairs of Noctua low-noise adaptors (basically a 3-pin 10cm extension cable with a resistor mounted on it) which will come in handy if I need to reduce the RPM of the new fans.

I guess I don't have to say it sounds really good now that the massive 50Hz peak is gone. There's more nuances in the bass now, it's not that one-note bass any longer.

Edit: Here's another picture with phase included.

Something is happening there at ~100Hz but I know too little about phase to draw any conclusions. My guess though is that it's because of the room dimensions.

Edit #2: I did some close up measurements of the right channel speaker.

This is with the mic 10cm from the upper woofer, on-axis. Only right channel speaker playing and being measured.

This is with the mic 10cm from the tweeter, on-axis. Only right channel speaker playing and being measured.

And here's a picture of both main speakers running full-range without crossovers and EQ.

As you can see the room has a HUGE impact on the sound. Close-up the speakers are pretty flat but ~2.8m away they are anything but flat. Room acoustics are important, very important...

A package from Thomann arrived

A package from Thomann arrived yesterday. As usual it was five time bigger than needed and so it was stuffed with those huge bubble wrap things. Great, now I have another 15m of it.

The package contained a Behringer iNuke NU6000 and a microphone stand. I unpacked the Behringer yesterday and put it in the shelf but I haven't gotten the MiniDSP plugin yet so there's no need to hook it up yet. I tried the microphone stand though and I did a quick measurement in REW at listening position. Unfortunately I forgot to save it but it looked quite similar to the measurement I did before. Above 400Hz it is acceptable but below ~200-250Hz it goes all crazy. Hopefully I'll be able to fix most of it with the MiniDSP and subwoofers.

Quick measurement of in-room response

Here's a quick measurement I did yesterday. Since I haven't got any microphone stand I had to hold the mic in my hands, which could have had some impact on the measurement. It looks similar to what I have measured before. With my MiniDSP I will try to tame the huge peak in the bass but I don't think I will be able to do much about the dip at 100Hz because it seems to be some room cancellation. Same with the dip at 200Hz. But I don't think the dip at ~650Hz is a room cancellation, any clues to what it may be?

The MiniDSP and UMIK-1 has arrived!

Hello dear readers!

The MiniDSP and UMIK-1 arrived today. I have been very busy though so I haven't had time to try the MiniDSP and I have only tried the UMIK-1 for a quick voice recording (which sounds very promising!). I don't know if I will have time tomorrow, but if I do I will put up some pictures and maybe a recording as well.

Initital impressions of the UMIK-1:

• Heavy and well-built! Well, not heavy but it's no cheap plastic stuff
• Nice little table stand, feels quite sturdy and easy to use and fold.
• Very easy to use. Connect the mic to the computer via the USB-cable that was in the box and you're ready to go. No additional software needed.
• Crystal clear voice recordings

I just did a quick voice recording but from what I heard I think I will use it for not only measuring speakers and room acoustics but also for Skype and Teamspeak. I don't know the directional pattern of the UMIK-1 but I guess it's a bit more narrow than the Logitech C270 webcam that I use as a microphone at the moment.

Here's a picture from the manufacturer.

MiniDSP and UMIK-1 coming!

Today I just ordered a MiniDSP 2x4 and a UMIK-1. I have decided to return my Behringer iNuke NU6000DSP and get a non-dsp verison instead. This way I save about 500SEK. Hopefully I'll be able to sell my Reckhorn S-1 for ~600SEK and my Radioshack digital SPL meter for ~300. This way one can say I got the MiniDSP and UMIK for free. I have no idea when the stuff will arrive, but hopefully within a week or two (it's being shipped from Hongkong).

I'll be away for a couple of days now for a school trip to Slovenia, visiting the wonderful town of Maribor.