Measuring speaker response to design XO on future projects

[gabdx]

Hi,

When using sinusoidal waves to measure each frequency response of the speaker, you place the microphone at 1m and don’t move it. What should you do about the peaks and valleys of the sound waves propagated in the air and hitting the microphone? The 'knots' and the 'peaks' of the waves seem to change place in the room depending on the frequency generated... When a knot hits the microphones there is a substantial drop in sound pressure measured, even if the speaker is linear in the response! Should you compensate for this problem or adjust the speaker until the knot are aligned for all frequency / or compensated by making an uneven sound response???

If you compensate peeks and valleys in the sound graph, will this be true for measurements at a 1.5 m or 2 m?

How, why, and should you even bother to get an even response of the sinusoidal pattern at a particular spot in the room???

I wonder how designers cope with this problem which can affect measurements quite a lot +-Db especially in the top frequencies.

Unlike live music, speakers are mostly unidirectional and generate peeks and valleys, my intuition would be that since music is not sinusoidal these problems DO NOT arise when music is played through the speakers.

[cuibono]

Hi Gabdx

Welcome to the wonderful world of speaker measurements - first off, the reason you are getting irregular measurements is due to the room interacting with the driver. To get around this, you need to 'gate' your measurements. Usually this is done with MLS signals, although one can use sine signals too (it depends on the features of your software). Gating the signal prevents the room reflections from causing nulls and peaks in your graphs, but it also limits the resolution of the lower frequencies, depending on the length of the gate.

If you can't gate the signal, I've found tone bursts to be much more immune to room effects than sine signals, but again, only special software can generate tone bursts, and I use a hardware oscilloscope to measure them.

The main deal with doing speaker measurements are realizing what your limitations are, and how to interpret you data around that. There is a lot of info about it on the web, but it takes a lot of time and effort. Luckily we have forums like this to ask others for help.

I highly suggest getting ARTA if you plan on doing any more measurements - having good software makes all the difference.

[gabdx]

Hi cuibono , thank you for the nice answer. I am glad to find help here; you see I am a newbie in speaker design. I only have a poor microphone and a poor computer noisy sound card (lots of fans) and my past measurements were atrocious.

I asked because I need to find a way to design my 3 way xo with vintage drivers. I have trueRTA, I don’t know how it compares to ARTA, and I can’t find a gate or burst signal option.

In your opinion would it be wise to buy something like the ultracurve/ultramatch Behringer and the tube condenser microphone (supposed to pick up bass frequency) to make good measurements over the entire sound spectrum??? Its lot of money that now I don’t have and I never used any of behringer stuff but the advertisement got me…

[cuibono]

Well, there is definitely a lot to learn - tube and condenser gear is not useful for these sort of uses.

First, try HOLMimpulse for software, it will do just what you want, and is free (it has gated MLS measurement capabilities). Second, you want to look for a electret measurement mic - people usually go for the Behringer ECM8000 or the Dayton EMM-6. If you want to be very accurate (you don't have to be), you'd get your mic calibrated, or buy one independently calibrated. Cross·Spectrum - Calibrated Behringer ECM8000 Microphones for Sale They are about $50 uncalibrated, and $100 calibrated.

The other thing you need is a decent soundcard and mic pre to connect the mic to your computer. There are many good options for $50-100, search the forums for more recommendations. I use the EMU 0404, but would recommend the EMU 0202. I think there are good soundcards with preamps to be had for $50, but I don't know them off hand.

[gedlee]

Originally Posted by cuibono

First, try HOLMimpulse for software, it will do just what you want, and is free

All hail Ask!! He is changing the world with his incredible software. Everyone should have and use HolmImpulse. And don't forget to thank Ask for this!

[catapult]

Measuring in a room is tough. Gating the impulse and converting that to frequency response removes room reflections but it also means the low frequencies are inaccurate. You usually end up doing a gated measurement for the HF and one with the mic right by the cone for the LF. Here's a good article about how to do that. It's written for ARTA but you could do something similar with HOLM. Don't let the math throw you. Basically, he's saying to:

• Take a measurement at 1m or whatever and gate out the reflections
• Take an ungated close-mic'd measurement
• Apply baffle-step compensation to the close mic'd measurement
• Adjust the dB of the close mic'd one to match the 1m one at the splice frequency
• Splice the two together

http://www.fesb.hr/~mateljan/arta/Ap...d-Rev03eng.pdf

[gedlee]

Originally Posted by catapult

Measuring in a room is tough. Gating the impulse and converting that to frequency response removes room reflections but it also means the low frequencies are inaccurate. You usually end up doing a gated measurement for the HF and one with the mic right by the cone for the LF. Here's a good article about how to do that. It's written for ARTA but you could do something similar with HOLM. Don't let the math throw you. Basically, he's saying to:

• Take a measurement at 1m or whatever and gate out the reflections
• Take an ungated close-mic'd measurement
• Apply baffle-step compensation to the close mic'd measurement
• Adjust the dB of the close mic'd one to match the 1m one at the splice frequency
• Splice the two together

http://www.fesb.hr/~mateljan/arta/Ap...d-Rev03eng.pdf

Thanks - I liked ARTAs writeup. Well done.

[gabdx]

I have HOLMimpulse too, just never used it yet, look complicated XD

I need precision on the kind of microphone I need to use. I have some concert microphone bought around 100 $ 10 years ago but it’s for signing, it’s like crap for speakers I think... The condenser microphone isn’t flat to 20 Hz is what you mean I suppose...

With the signal processing I think ill get the behringer ultracurve or something similar, I trust it more than a noisy computer card. Thanks for all the info.

Now ill do research on the gaited measurements. But my question has not been answer yet as if it matters or not to have a straight frequency response concerning the peeks and valleys of waves...

QUOTE: the reason you are getting irregular measurements is due to the room interacting with the driver.

I know there is lot of them but they are also caused by the fact that sinus waves create peeks and knots in the air, if you move in the room you can clearly put your finger on them

[head_unit]

Originally Posted by gabdx

Now ill do research on the gaited measurements. But my question has not been answer yet as if it matters or not to have a straight frequency response concerning the peeks and valleys of waves...

QUOTE: the reason you are getting irregular measurements is due to the room interacting with the driver.

I know there is lot of them but they are also caused by the fact that sinus waves create peeks and knots in the air, if you move in the room you can clearly put your finger on them

It's not sine waves creating peaks and dips. Sine waves move over time, so at a fixed location a purse sine wave would make the pressure go up and down sinusoidally.

But in a room you do have room modes and resonances and reflections and cancellations. That's probably what you're talking about, and that's what the gated measurements remove (except at the lower frequencies).

Once you're making a proper measurement with good equipment, good quality transducers should not have sharp peaks and dips. No transducers will measure perfectly flat, and most will have some peaks and dips, but they should not be sharp and/or huge.

Welcome to the fun and maddening world of speaker design! (SUCKER ha ha abandon all hope) ;-D

[gabdx]

Sine waves move over time , at what speed ?

Just to say I noticed with better amplifiers the peaks and dips are more noticeable, less harmonics I guess.