Help with subwoofers cut off

[Davide]

Hi everyone,

I finally made my two DIY subwoofers operational, made in the following way:

Driver:
- Scan Speak 30W/4558T00

Box:
- Acoustic suspension design (Vas/Vb=3.1)
- 30mm MDF walls + 19mm internal bracing
- 463mm baffle width (perfectly cubic shape)
- 60 litres net internal volume (very very well sealed)
- Stuffing with Visaton Polyester Whool (approximately 105 grams)
- Resulting Vb = 67 litres
- 3 x Rubber feet (diameter 40 mm, height 10 mm)

Amplifier:
- Hypex NC502MP amplifier (external)

I sized the internal volume with the spreadsheet provided by Scan Speak.
I made it 60 liters, increasing the apparent liter to 67 with stuffing (always calculating this with the spreadsheet they provided).
See below:

The subwoofers are placed roughly in the center of a 6.7 meter wall, and 1.2 meters apart (not optimal, I know, but that's what I can do). The other side of the wall is 4.2 meters, while the height is 3.2 meters).

Now I performed some quick measurements to see the response, and this is the result, complete with modes estimated by REW.

Up to 45Hz everything is fine, further down there is this drastic drop in both subwoofers.

From the simulation I should have obtained a gentler roll off, around 12dB per octave.
I don't know if I'm dealing with a room-cancelling effect, or if there's something wrong with the design.
In theory an acoustic suspension shouldn't drop that steeply.
Furthermore, the final Fs should be right at 34Hz, where there is that big drop...
I'm thinking that the rubber feet have a dampening effect on those frequencies.
Or is my microphone faulty?

What do you tell me?

PS. no DSP was applied. This is raw response.

 

[antcollinet]

So your subs are less than 1/4 wavelength apart at 70Hz. The fact you've placed them centrally in the wall means you can also be getting some symmetry in the reflections, increasing the chance of co-incident in phase reflections.

I know you say it is the best you can do - but as a diagnosis test try putting one of the subs in an acoustically completely different place (Eg towards the end of the opposite wall.

Don't worry about cable routing for the test.

 

[ppataki]

Perform a measurement from 1cm with the measurement mic at 0 degrees
This will show you the response (almost) without the room

I also believe that your sharp dip above 40Hz must be room-mode induced

 

[sam_adams]

The response drop you are seeing is due to the Windows audio settings for the output and the mic. You need to turn off all the enhancements in the Sounds Control panel item for both the input and output in order to get full-range from the devices.

 

[Davide]

The response drop you are seeing is due to the Windows audio settings for the output and the mic. You need to turn off all the enhancements in the Sounds Control panel item for both the input and output in order to get full-range from the devices.

No, the audio travels entirely on ASIO.

Perform a measurement from 1cm with the measurement mic at 0 degrees
This will show you the response (almost) without the room

I also believe that your sharp dip above 40Hz must be room-mode induced

I've thought about this too and I think I'll check this out first. Thanks

I know you say it is the best you can do - but as a diagnosis test try putting one of the subs in an acoustically completely different place (Eg towards the end of the opposite wall.

Moving these boulders is truly a feat on its own, I'll leave it as a last chance...

However, I specifically tried to calculate the room modes with REW, and actually, as you can see in the first image, there is that little red bar near the dip. But I didn't understand if it should be a cancellation or a reinforcement. Red is labeled "axial".
It seems to me that there are actually dips corresponding to the other red bars.

 

[alex-z]

In addition to the near-field mic sweep, I would recommend checking the final impedance to make sure your absorption material is having the affect you expect.

 

[Davide]

In addition to the near-field mic sweep, I would recommend checking the final impedance to make sure your absorption material is having the affect you expect.

Is there any easy way? Seeing REW guide seems complex to me...

 

[ZolaIII]

@Davide nope there is no easy way.
You need to do the proper placement (1,2,4 m...), elevation and angle (45° vertically) for a microphone with psycho accustic smoothing catching two or three outputs with lowest background noise you can. You also need deacent SPL meter. For the as close to natural response you would have to do open field (out door's) measurements like for instance Eren douse. Something is wrong around 70 Hz as antcollinet said. Those driver's are a bit fussy and hard to drive on bothom end. The response dB slope roll off will depend more of how hard you drive them it's only ideally 12 dB per octave for compression enclosure. Anyway more than that... fundamental oscillation frequency for rubber is 23 Hz and in the future use silicone (whose is 9 Hz and has better property's).

Scan-Speak 30W/4558T00 - 12″ Subwoofer

12″ Subwoofer, 17 Hz, 350 W, 86 dB, 4 Ω

loudspeakerdatabase.com

Please make an effort regarding doing a proper setup and measurements and post full measurements (FR with phase and RT60 at least). Don't worry you will get there and even after basic room mode fix (bottom fundamental peak) it will sound good.

 

[Davide]

You need to do the proper placement (1,2,4 m...), elevation and angle (45° vertically) for a microphone with psycho accustic smoothing catching two or three outputs with lowest background noise you can.

Sorry, but is this for measuring impedance? Or do you mean to have better response mapping?

You also need deacent SPL meter

I have a Beyerdynamic MM1 calibrated in frequency and sensitivity, so the SPLs I measure are effective (cal file applied to REW).

Something is wrong around 70 Hz as antcollinet said.

That certainly, but 99% of it is an effect of the room, and I'm not surprised. What amazes me is the response from Fs (theoretical) down.

Those driver's are a bit fussy and hard to drive on bothom end. The response dB slope roll off will depend more of how hard you drive them it's only ideally 12 dB per octave for compression enclosure.

In theory in that band, from Fs down, I am Xmax limited, not power limited. So I'm looking for an explanation...

Anyway more than that... fundamental oscillation frequency for rubber is 23 Hz

I had informed myself about it, and from the online documentation it appears that the rubber reaches a maximum of 20Hz. Normally should be 5-15hz.
But I'm also considering this problem (probably I'll support the subs on something else to see if anything changes).

Thank you all so much for the support in the meantime!

 

[sigbergaudio]

Which microphone are you using? Start with a nearfield measurement as suggested by others. Ideally measure outdoors, putting the mic on the ground 1m from the subwoofer.

 

[Davide]

Which microphone are you using? Start with a nearfield measurement as suggested by others. Ideally measure outdoors, putting the mic on the ground 1m from the subwoofer.

Beyerdynamic MM1, with factory cal file.

 

[sigbergaudio]

Beyerdynamic MM1, with factory cal file.

Let's see a measurement ~5cm from the driver then.

 

[ozzy9832001]

So your subs are less than 1/4 wavelength apart at 70Hz. The fact you've placed them centrally in the wall means you can also be getting some symmetry in the reflections, increasing the chance of co-incident in phase reflections.

I know you say it is the best you can do - but as a diagnosis test try putting one of the subs in an acoustically completely different place (Eg towards the end of the opposite wall.

Don't worry about cable routing for the test.

This is actually the case in my room as well. The subs perform best when one is centered between the 2 speakers and the other placed in the front corner. They are only maybe 3' apart.

 

[ZolaIII]

@Davide not for impedance. It's to improve accuracy and extended regarding highs mostly. Impedance falls to 3 Ohms on those but trouble is the high to low (impedance) angle when saying not easy to drive. More measurements you do it becomes easier (and boring eventually) and you get to know your space better.
You first need to patch that hole at 70 Hz (which I am afraid means finding a better placement), then see how much is it room and how much woofers response the bottom peaks (47, 54 Hz) are and iron them out (DSP - PEQ). The down slope will be little less out of the cliff looking then. Point is play with it a lot.

 

[sigbergaudio]

Perhaps also measure a bit louder, around 80dB. And use 45-105dB on the Y-axis and 20-100hz on the X-axis. A more common way presenting it = easier to read and comment.

 

[sam_adams]

No, the audio travels entirely on ASIO.

You are operating under a misconception about how an ASIO driver works. Nothing that ASIO4All does prevents Windows from applying its special brand of stupid to the audio signal chain.

Disabling all Windows audio enhancements for the input and output devices has solved user issues, again, and again, and again.

 

[Davide]

You are operating under a misconception about how an ASIO driver works. Nothing that ASIO4All does prevents Windows from applying its special brand of stupid to the audio signal chain.

Disabling all Windows audio enhancements for the input and output devices has solved user issues, again, and again, and again.

I appreciate your effort to provide me with an explanation, and I thank you, but you are off base.
I don't use ASIO4All. I use the ASIO drivers of my Motu Ultralite Mk5. And the device is set to have no WDM drivers exposed (there is the option to do this).
Furthermore the Windows audio services on my PC are also stopped, cause I don't need they (mixer, endpoint, etc).
So the stupid Windows audio system has nothing to do with it, let's leave this point aside.
PS. I have thoroughly tested the Windows audio system and, apart from third-party APOs (present only on laptops and pre-assembled PCs), it is completely transparent without limiter intervention (avoidable with -0.2dB gain).

 

[Davide]

So, I just managed to take measurements at a very short distance from the cone (<5cm).
Unfortunately, I can't do widespread measurements in the listening area in the evening... I'll do it at the weekend, time permitting.

Below are the results, which seem rather correct to me from the point of view of frequency roll off (about 12dB/Oct).
I also investigated IR and GD for my own sake, to verify the consistency between them.

MAGNITUDE LEFT (NO SMOOTHING)

MAGNITUDE RIGHT (NO SMOOTHING)

IMPULSE RESPONSE

GROUP DELAY

I also developed a graph, thanks to a spreadsheet on the Linkwitz website, where the theoretical response of each sub is shown as a function of the Xmax and Voltage/Current limits of the amplifier (NOTE: the actual power delivered has the dimension of Volt-Ampere not Watt, since voltage and current are not necessarily in phase).
As you can see, the SPL limit is given by Xmax below 42Hz (provided you limit the power according to the pink curve), while above it is given by the amplifier (in reality the power limit tolerated by the driver is not considered, which would be 350W, but little changes in terms of SPL).
This graph is very interesting because it allows me to see how much I can pump with DSP without exceeding the Xmax limit (at least in broad terms).
Given the pink power curve it is easy to calculate the dBFS of attenuation necessary for my signal to respect it (point by frequency point obviously).

So, from a design point of view it seems to me that everything is correct and that there is room to compensate a little for the roll off with the DSP.

I still need to better map the widespread response in the listening area, as someone said above, and understand if that sudden drop below 45Hz is persistent and if it is a room effect (but having seen the response of the subs up close, I'd say it has to be, right?).

Then honestly I'm not worried about how to resolve it now, I'll invent a new positioning so that one sub compensates for the other's dips. Then I will let Dirac do his job.
The important thing is to understand if the subs I made are ok in themselves.

I'll update you as soon as I get the other measurements.

In the meantime, thanks again to anyone who will contribute to the discussion.

 

[ppataki]

These measurements look great; I am sure you will be able to solve the issue by finding the right spots for the subs + DSP (Dirac)

 

[Davide]

Unfortunately I have little time to experiment and I decided to cut to the chase: since the response of my subs seems correct when measuring close to the cone, I launched Dirac and looked at what it combined with the frequency response in the listening area.

This is the result with psycoacoustic smoothing:

Not bad apart from that upper bass area which is where there is more congestion of room effects.

I would say that once Dirac ART is released that area will definitely be filled. The main speakers and my subs are perfectly overlapped in those frequencies so they can help each other in theory. I can't wait to try.

In the meantime, here are some digital loopback measurements of the Dirac Live Bass Control filters: