How quiet for quasi anechoic measurements?

[digahill]

Hi all - I'm following the excellent sticky post quasi-anechoic measurement guide to understand the frequency response of my Overnight Sensations build. My goal is to take measurements of a test box I built, which isn't very well air sealed nor does it have any polyfill. I'm curious how this test box will ultimately compare to the finished build in a month or two.

I live on a busy road, so finding a very quiet space isn't really possible. I'm taking these measurements in my garage, where you can still hear low rumbling road noise and also the low hum of the power amp I'm using for testing. This is about as quiet as I can (easily) make the space.

The impulse response I'm getting doesn't show a very clear reflection spike around 6-8ms like I see in all the examples I've seen on the forums. I assume other people are taking valid measurements in environments that aren't perfectly quiet, but maybe I'm wrong.

My question for those who know more - does the attached impulse response look like my calibrated mic might be bad (or something else in my signal chain) or could this just be due to the environment not being quiet enough?

Another note: I was taking measurements at a pretty high SPL, in order to improve the signal to noise ratio. But the SPL of this mic is not calibrated.

 

[RayDunzl]

Post an RTA of your noise level, then post an RTA of the test tone from the speakers, so we can see the relative levels with your uncalibrated mic.

Reflections should poke out from the noise if your tone level is loud enough to swamp the noise.

Not sure what you are expecting to see in your highly reflective garage, though.

Impulse with reflections in my room. Not sure why there is no "noise" before the impulse.

The flight path of the reflections is about 1ms delay for each foot of additional distance over the direct sound.

Noise in my room a few days ago.

 

[amirm]

You would need to be away from all reflective surfaces, not just the floor. To wit, if the ceiling joists are close than the floor, they show up as spikes earlier. Ditto for the car next to it. People tend to do these outdoors where the only reflection to worry about is the floor.

On general noise, that can be dealt with by having loud enough signal and using a microphone distance that is short enough to achieve high SNR. I think you are already doing that.

 

[BenB]

The graph you posted doesn't show much noise before the impulse. Constant background noise will typically be visible there. You don't appear to have much of a reflection-free zone on the graph, even though the set-up looks like it should give you one. Is there acoustic treatment (Wool, poly-fil, something like that) inside the speaker cabinet? Maybe the speaker and mic are too close to things above them, even though they are far off the floor.

Honestly, when I have a set-up like that, and get measurements like you showed it's usually because I forgot to plug the measurement mic into my laptop, so it used my laptop mic instead... which gets reflections off the screen, and off my belly, and my lap, etc. I'm sure you aren't as forgetful as I am, though, or if you are, you would have realized your mistake before posting. It is a mystery.

If you are worried about background noise (which I don't see strong evidence that it's impacting you), then you should use the longest signals available on your measurement software. That will improve the SNR of your measurements.

 

[a4eaudio]

...
Honestly, when I have a set-up like that, and get measurements like you showed it's usually because I forgot to plug the measurement mic into my laptop, so it used my laptop mic instead... which gets reflections off the screen, and off my belly, and my lap, etc. ...

Haha, glad to see I'm not the only one who has done this!

 

[JSmith]

People tend to do these outdoors where the only reflection to worry about is the floor.

JSmith

 

[Keith_W]

Welcome to ASR! Many people think that the benefit of an anechoic chamber is the signal to noise ratio. If it can suppress background noise, then the SNR will improve, right?

Not so. The benefit of anechoic chambers is that reflections are suppressed.

Improving the SNR is trivially easy and well within reach of us hobbyists. It is not so much the "noise" part than can be suppressed, but the "signal" part can be improved. There are two ways to boost the signal - you can increase the height (SPL) of the signal, or increase the width (time) of the signal.

I do not recommend increasing the SPL. This is because loudspeakers are nonlinear devices, and are likely to distort or dynamically compress at high volume. I measure my loudspeakers at about 70-75dB and I check it with an SPL meter.

It is far better to increase the width of the signal. You can do this by choosing a slower sweep, or by taking multiple sweeps and averaging them. This is the noise floor of a single measurement:

And this is five measurements combined:

Both waterfalls have the same scale.

 

[digahill]

Great thoughts. Thanks everyone for your replies! The temps around here have dropped back down to the teens this week, but once it warms up I’ll give the measurements a try outside. It’s will be noisier, but reflections should be limited.

Speaking of reflections, my impression was as long as all surfaces were at least ~5 feet away the room could be very reflective without negative impact to the result. Since that gives me a 6-7ms window reflection free to work with.

However, the sweep definitely takes more than a few milliseconds, so I’m not sure how that works.

I’ll also verify I wasn’t using the laptop mic. That’s a great thought.

If anyone is interested, my project files for the build as well as a bunch of pics are here: https://github.com/digahill/overnight-sensations

 

[Keith_W]

Please read this post on how to take measurements. That post has a section on how to calculate the lowest frequency you can measure which is uncontaminated by reflections.

Then read this post on how to determine measurement quality. In particular, you can see the reflections in your measurement if you know where to look. You can also determine the longest wavelength that you have measured which is free from reflections.

 

[digahill]

@Keith_W Thanks for that article about measurement quality. That was a huge help.

I have now collected some measurements that I think I can trust. Thanks all!

 

[RickS]

Welcome to ASR!

Looks as though you got some experienced help. Would be interesting to see your results.

Was your intent just to validate the build or do you plan to modify these? Given the age of the design, am surprised there are not more measurements posted. The designer did not even bother with impedance or off-axis metrics (though no off-axis is not unusual for an older design).

 

[digahill]

I was just trying to learn a bit about measurement and see how my results compared to the designer’s. I haven’t done the bass measurement yet, but my 200hz to 20khz response doesn’t really match Carmody’s, unfortunately. I think my cabinet build is really solid. So maybe I messed up the crossover or something. The speaker still sound much better than the sound bar I was using for my TV though. Night and day. When I get back to my computer, I can grab the REW screenshot and post.