Project studio/music room - construction phase

[kominak]

After placing the cloud 20cm futher back into the room (away from the soffit) I was able to simplify it a bit:

cloud_raytrace2.png

I still wasn't able to get away with just one section of the cloud, but 2 sections are better than 3...

I just found out there is a small crisis regarding mineral wool - waiting/delivery times are several months long!
I still have some that I will use for RFZ wings but the choice will be limited for the cloud.

How thick should the insulation in the cloud be? I guess I'd like to address the midrage bumps (100Hz -1kHz) in the frequency response of the room. Would the 8cm thick insulation work?
Another question: How wide should the cloud be? Is there a limit where it starts to interfere with bass traps in the corners?

[kominak]

Yesterday I finished the rough work on "wings" - those constructs on both sides of the soffit that help to create RFZ zone in the mixing position:

wings_stiched_small.jpg

New set of measurements is here: Wings measurement

The frequency response got a little bit prettier again and the listening experience is even more enjoyable - the left/right and front/back space is even more defined.

I finally found out what was the reason for that weird frequency dip around 10k in previous measurements: when using usb mic I should not use 2 sweeps when making measurement with REW - there's a timing/latency problem when using different devices for playback and for input and using multiple sweeps, it's described in the REW manual.

So - when using USB mics, stick with single sweep!

[kominak]

Small progress here - I finished bass traps in the wal/ceiling corners plus absorbtion layer in the middle part of the celing. The area where the hard backed cloud will be (in the front) was left bare, same as the area in the back of the room, where the recording will take place - I'll deal with that later.
Due to stiching of multiple photos the room does look longer than it really is - the bass traps are 3m long, the absorbent ceiling area is 2m long:

ceiling_stiched.jpg

Measurements are here: Measurements with ceiling bass traps

It seems to come together nicely - at least from my inexperienced point of view. Here's waterfall graph with standard 1/3 octave smoothing:

waterfall_ceiling.png

The response is still not within +/- 3dB that ITU/EBU call for, but still better than I hoped for when I first started with this (much) less than ideal room.
The RT60 time is still a little high (250-300ms) as oppossed to the target of 180-200ms, so I think still more absorption is needed (see those long parallel walls in the first picture...).

Here's the impulse response (0,1 ms smoothing):

IR_ceiling.png

There's still some spike above -20dB level within first 5ms, which i think is reflection from the celing, I hope the cloud will solve this (and still, this is not too bad - the ITU spec for control rooms specifies reflections within 15 ms be at least 10 dB down).

The next step will be the hard backed cloud above the listening position, and I'll try to add more absorption around/behind it to dampen the low frequency reflections, as per Stuart's recommendation.

[Soundman2020]

It seems to come together nicely - at least from my inexperienced point of view. Here's waterfall graph with standard 1/3 octave smoothing:

For the low end of the spectrum, don't use smoothing: It hides all the modal stuff. At best, use 1/24, or preferably only 1/48. It's fine to use more smoothing for the high end. I normally use 1/6 for the highs, 1/12 for the mids, and 1/48 for the lows.

- Stuart -

[Gregwor]

The RT60 time is still a little high (250-300ms) as oppossed to the target of 180-200ms, so I think still more absorption is needed (see those long parallel walls in the first picture...).

Did you look at your spectrogram? It appears that your decay is ~180-200ms!! It's coming along great. I'm sure the room will be amazing in the end.

Greg

[kominak]

For the low end of the spectrum, don't use smoothing: It hides all the modal stuff

Here's the unsmoothed low end of the spectrum:

waterfall_ceiling_lowend_unsmoothed.png

And here's the whole spectrum:

waterfall_ceiling_unsmoothed.png

Did you look at your spectrogram? It appears that your decay is ~180-200ms!

Hm, that's the type of graph I'm not familiar with, but I think I see what you mean:

spectogram_ceiling.png

Still, there's a lot of flutter echo in the room due to big parallel surfaces that are not treated yet. I was thinking about placing 10cm absorption there in checker board pattern, but now I'm afraid it will deaden the room too much...

[Gregwor]

I'm not sure if it's a calibration thing on my end, but when I opened your file, this is the spectogram - showing between 180 and 200 ms!

Ceiling Specto.jpg

Greg

[kominak]

Hm, that's strange. Here are my (default) settings:

spectogram_settings.png

I still don't quite understand this. If this should show the RT60, shouldn't the "Scale Range" parameter be set to 60 dB?

Either way it looks like I should proceed carefully when adding more absorption and probably start to treat the problematic areas/modes in more surgical manner instead of adding more broadband panels.

I think that John's slotted traps could help me with the modes in the low mids, but I'll have to call on Stuart's expertise to calculate them correctly.
Here are some of them, beautifully crafted (2 last pictures): Slotted traps in Marc's studio

[Soundman2020]

I'm not sure if it's a calibration thing on my end, but when I opened your file, this is the spectogram - showing between 180 and 200 ms!

You shouldn't use the spectrogram for measuring decay times. The RT-60 graph is meant for that. The spectrogram is usueful for pinpointing modal issues and SBIR, as well as a few other things, but it's not calibrated to 60 dB decay.

Here's what your actual decay looks like:

Martin--REW--RT--50..11k--ceiling.png

Still, there's a lot of flutter echo in the room due to big parallel surfaces that are not treated yet. I was thinking about placing 10cm absorption there in checker board pattern, but now I'm afraid it will deaden the room too much...

Your room response is looking pretty good overall! Especially considering the small size and non-ideal dimensions. The decay times are close to ideal, in general, but you could go a little lower on the high end, and a little higher in the low mids (always tricky). It's a small room, so you don't want the room too "bright". There's a "hump" in the high end that needs attention. Not sure what it is yet. And there's some strange stuff going on in the mid range that looks like it could be flutter echo. So I would suggest building some not-too-large absorbers for the side walls, with just 2" of OC-703, not sealed, and maybe some random slats on the front. You could try interspersing those with shallow poly-cylindrical diffusers, as well. For example, on each side wall have a tall poly with a pair of absorbers on each side, one up high and one down low. That will help to deal with the flutter, while not killing too much of the decay. Do a TRUE poly shape (catenary curve), NOT the typical circular-based cylinder. Make it at least 18" wide (45cm) and at least 6" deep (15cm) at the peak. If there's enough space to do 20" (50cm) and 8" (20cm), that would be better, but I have a feeling that will be too big for your room, aesthetically and practically.

Something like this (very rough sketch!))

Martin-K--Side-wall-rough-suggestion-S42.jpg

That0s not a true poly I drew there! That's a squashed circle. You need a try poly, but I didn't have time to do that. For reference, the dimensions are: poly: 50cm wide x 140tall x 15cm deep at the center. Absorbers: 60cm wide x 40 cm tall x 5cm deep. Front ones roughly in line with your head, poly about 15cm behind the rear edge of those, and the rear absorbers roughly 20cm behind the poly. Front absorbers are 60cm from floor and 120 cm from floor (lower edge). Rear ones are 50 and 130 from floor. Poly is 40 above floor. Those are all aprox. and not too critical. You could experiment to see if you can find better locations for them, but somewhere around there should be good, and in roughly the same pattern. Do the same on both walls: mirror image. Don't put slats on the absorbers yet: wait until you have tested them without slats.

This might not be enough, but it's a start. That's only about 2m2 of absorption, and you'll probably need more but take it slowly, one step at a time.

Your room is probably going to be a good candidate for digital tuning, because you are treating it so thoroughly. If you are interested, e-mail me.

It's working out very nice!


- Stuart -

[kominak]

Great insight, as always - many thanks Stuart!
I still have to finish the overhead cloud, but once that is done, I'll work on the side walls.
I'll be interested in digital tuning once I run out of possibilities in basic room treatment, that's for sure - I'll shoot you an email once I'm there!
Also, a big thank you for encouraging words, it feels great knowing the work was not in vain...

[kominak]

The overhead cloud is done:

overhead_cloud_stiched_small.jpg

The middle part between the studs is hard backed(18mm plywood with 10cm thick mineral wool underneath):
The measurement done is here:Overhead cloud measurement

There's still one -20dB spike at 4.13ms on the impulse response graph, which I think is the reflection from the ground:

impulse_response.jpg

Hopefully this will be dealt with using angled console.

There appeared a dip in the frequency response graph at 157Hz, which buggs me a little:

freq_measurement.jpg

I was thinking it is correlated with that reflection from the ground, but my calculations didn't confirm that - Stuart do you perhaps know if this is a mode related issue?

Happy building everyone
Martin

[kominak]

I spent some time pondering on ideas for side wall treatment and I think I need to make some small adjustments. Here's the look of the side wall at the moment:

side_wall_empty.png

Stuart, please correct me, if I'm wrong, here are my thoughts:
The RFZ "wings" end roughly in line with my head, so the first set of absorber panels need to be pushed slightly to the back of the room, I think? Then there's a stud in the middle of the wall, supporting the ceiling structure, that I'd like to hide within the diffuser (because it causes some early reflections at 13ms as I just found out...). Something like this:

side_wall_filled.png

The front pair of absorbers got slightly wider (because of the position of the diffuser), the rear pair got slightly shorter (the dotted line on the wall indicates, where the door lie on the opposite wall - if I want to have the same treatment on both sides, I have to end here - not sure, how important that is).

The dimensions of the poly are: 60cm wide, 15cm deep, 120cm high - mainly, because I found a flexible poly sheet that's 250x120cm (it's flexible only in one direction). It's 7mm thick, made of soft wood. Or is it better to use regular 4mm hardwood ply?
I found an information, that homogenous wood bent between two ends will create a natural catenary curve, so there's no need to make any sort of "form" for this poly - is it true?

It's hard to get 5cm thick 703 these days around here (I would have to wait a couple of months) but I have some 8cm panels left - it's OK to use them, I guess(without cutting them to 5cm thickness)?

[Soundman2020]

The RFZ "wings" end roughly in line with my head, so the first set of absorber panels need to be pushed slightly to the back of the room, I think?

Then there's a stud in the middle of the wall, supporting the ceiling structure, that I'd like to hide within the diffuser (because it causes some early reflections at 13ms as I just found out...). Something like this:

The front pair of absorbers got slightly wider (because of the position of the diffuser), the rear pair got slightly shorter (the dotted line on the wall indicates, where the door lie on the opposite wall - if I want to have the same treatment on both sides, I have to end here - not sure, how important that is).

That's fine. It's always good to keep the room symmetrical if you can, even behind the mix position. In front of the mix position is critically important, but if you can do it behind as well, that's great.

The dimensions of the poly are: 60cm wide, 15cm deep, 120cm high - mainly, because I found a flexible poly sheet that's 250x120cm (it's flexible only in one direction). It's 7mm thick, made of soft wood.

Perfect! There's a special type of plywood, usually called "flexible ply", but also has various brand names, and that's the stuff you need. It's specially made to be very "bendable" in one direction, precisely for this type of application where you need curved surfaces. It sounds like you found exactly what you need.

I found an information, that homogenous wood bent between two ends will create a natural catenary curve, so there's no need to make any sort of "form" for this poly - is it true?

Right. It might not be a perfectly mathematical catenary curve, but it will certainly be close enough for your purposes. Here's a panel I did for a studio in New Hampshire a few months ago:

catenary-poly-corner-panel-OVERVIEW.jpg

catenary-poly-corner-panel-CONSTRUCTION-DETAIL-SML.jpg

In this case I scaled it "forwards" just a bit, so it is slightly deeper than a true catenary curve, as I needed more exaggerated diffusion. There's also a part on the back that you can see in the detail view, but you won't need that. In this case, the panel was movable (rotatable on a vertical axis), acting a bit like a gobo, so either side could face the room, but yours will be fixed to the wall, so you don't need that part. You could if you wanted to! If you can spare the extra depth to do that, you'll get better diffusion down to lower frequencies (or better scattering, to be more technically correct...) Just add a frame at the back, as deep as you can (at least 5cm, preferably 10 or more), and cut multiple large holes or slots all around the edges of the frame, to expose the internal insulation to the room, then cover that frame with nice fabric. That would help. Of course, your room isn't very wide, so that might not be physically feasible. The effect is small (but useful), so if you can't do it, it's not a huge deal.

- Stuart -

[kominak]

If you can spare the extra depth to do that, you'll get better diffusion down to lower frequencies (or better scattering, to be more technically correct...) Just add a frame at the back, as deep as you can (at least 5cm, preferably 10 or more), and cut multiple large holes or slots all around the edges of the frame, to expose the internal insulation to the room, then cover that frame with nice fabric. That would help. Of course, your room isn't very wide, so that might not be physically feasible. The effect is small (but useful), so if you can't do it, it's not a huge deal.

Oh, I knew I should have waited! I already finished the diffusers this morning
In the end I found out that the plywood is flexible in the other direction than I first thought, so that allowed me to more closely follow the design you posted on August, 3rd:

side_walls_stiched_small.jpg

It's hard to get the perspective right from these stiched pictures, but you get the idea. The whole system is quite unobtrusive and the polys look very elegant, despite them being slightly larger (The poly is 140cm high, 50cm wide and 17cm deep, filled with mineral wool). The absorbers are 50x40cm and 60x40cm, 8cm thick.

The best news is: IT WORKS! The slapback echo is gone and the room doesn't feel dead. Here's the waterfall graph:

side_walls_waterfall.jpg

and the RT60:

side_walls_T60.jpg

Here's the mdat file: Side walls measurement

It actually worked rather too well, probably because the absorbers are thicker. The RT60 time is around 165ms now, my target zone is around 180-200. That could be brought back with some slats, but probably better after the flooring and desk is in place (and the fabrics all around the room)?

I wasn't expecting it, but the sound changed quite dramatically - it's like the fog has lifted. I thought the speakers were revealing before, but now - this is revealing. The backing vocals suddenly became separate singers, where I heard one distorted guitar I now can hear several different guitars playing at once. Interestingly even the bass frequencies are a little bit smoother.

Many thanks Stuart - your design works, as always

Martin

[Soundman2020]

The best news is: IT WORKS! The slapback echo is gone and the room doesn't feel dead.

I was sort of hoping that would be the case!

It actually worked rather too well, probably because the absorbers are thicker.

I expected that: That's why I mentioned adding the random slats in the other post. You might need that, but wait and see...

The RT60 time is around 165ms now, my target zone is around 180-200.

165 is a little low, yes, but not shabby at all for that room. It could come up a bit, especially in the low-mids, but I wouldn't be too concerned about it. More important than hitting a specific overall time, is getting the times consistent between adjacent frequency bands... and you are certainly doing that! (see graphs below)...

That could be brought back with some slats, but probably better after the flooring and desk is in place (and the fabrics all around the room)?

Right. Fabric will take the edge of the highs as well.

I wasn't expecting it, but the sound changed quite dramatically - it's like the fog has lifted.

Yup! Diffusers will do that for ya, and polys are especially nice for doing that. Side wall treatment in general is very helpful here.

I thought the speakers were revealing before, but now - this is revealing. The backing vocals suddenly became separate singers, where I heard one distorted guitar I now can hear several different guitars playing at once.

"I love it when a plan comes together!". That's the major goal for a control room: clarity, clear stereo image, and accurate, precise sound stage. You got it! Well done! Even some so-called "pro" rooms sometimes can't achieve that convincingly.

Interestingly even the bass frequencies are a little bit smoother.

Right! Because you are treating your width axials for the first time. Before this, there was nothing to deal with the modes running across the room (as well as the flutter), but now there is.

Many thanks Stuart - your design works, as always

Thanks for the kind words, Martin! I'm pleased that you are seeing (and hearing!) good results, after all the work you put into this place. It's all paying off, now.

OK, so let's take a look at what you accomplished with those polys and absorbers, with some more detailed REW analysis. Here's a series of "before and after" graphs, showing various aspects....

First up, and perhaps the most impressive, is full-spectrum, unsmoothed waterfalls.

BEFORE:

Martin-K--REW--WF-20-20k--Ceiling-done.jpg

AFTER:

Martin-K--REW--WF-20-20k--Side-polys.jpg

You can clearly, clearly see just how bad it was, with all that nasty ringing going on at specific frequencies. There were around 20 very large, very long spikes there before, and all of that has gone now. The difference is day-and-night. That's where all your clarity is coming from. That's why the "fog has lifted". All of that "mush" is wiped out, so now you can hear the pure, clean, direct sound.


Next up: Spectrograms of only the low end.
BEFORE:

Martin-K--REW--SP-18-500--Ceiling-done.jpg

AFTER:

Martin-K--REW--SP-18-500--Side-polys.jpg

Maybe not so dramatic as the waterfalls, but you can still clearly see what you accomplished, and why the bass sounds smoother now. The low end is coming along very nicely!



Now for decay times (RT-60):
BEFORE:

Martin-K--REW--RT--50-10k--Ceiling-done.jpg

AFTER:

Martin-K--REW--RT--50-10k--Side-polys.jpg

Now you can see what I'm talking about regarding decay times consistency between frequency bands. Before, it was all over the place! Each frequency band had a different decay time, with large differences. Eg: look at the difference between 400 Hz and 500 Hz "before": 400 Hz was over 430ms decay, while the very next band up was 240ms, and the very next band down was 215ms. Those are MAJOR swings, well over 200 ms. And AFTER, they are all practically identical: 400 Hz is 135 ms, the next band up is 160ms, and the next band down is 145 ms. Differences of just 15 to 25 milliseconds! Specifications say that it should be no more than 50ms difference between adjacent bands and you are already twice as good as that! This is more important than hitting a perfect target time: smoothness beats pure numbers here.


And finally, one that isn't commonly even checked by many people, but is really important: Phase response:

BEFORE:

Martin-K--REW--FR.PH--18-2k--Ceiling-done.jpg

AFTER:

Martin-K--REW--FR.PH--18-2k--Side-polys.jpg

It's not so much the actual phase that we are interested in here, but rather how the phase CHANGES over the spectrum, compared to the ideal phase response. That's what the "excess phase" graphs shows (it's more complex than that, but a simple explanation suffices here). As you can see, before the polys, you had major, huge jumps in phase response, but now the phase response is smooth and clean: phase is rotating exactly as it should.

So, yeah, I think it's safe to say that those devices are doing what I designed them to do! And the REW data is just confirming what your ears are already telling you: that's a really nice, clean, smooth, accurate, room now! Congratulations!


- Stuart -