Discussion on how speakers should be measured

[oon_the_kid]

Hi,

The purpose of this thread is to reconsider how we test our loudspeakers so they better reflect what we hear. I would like to seek your comment and inputs on how we could do this better.

I have the Dayton Omnimic V2 system, however, I wouldn't rate myself as expert.

If you look in a standard loudspeaker test, there are generally no indication that a full range speaker will sound better than a multi driver in the mids. Although in our experience, they generally do.

I have some proposals, which I think may make testing of speakers more representative to how we appreciate music and not just looks good for electronic engineer ( I am one BTW).

1) Frequency response at 1W and 1m
Typical measurement is at 1w at 1m or 2.83V at 1m frequency sweep. I would say it is not representative. I would suggest instead of compering frequency response at same power, they should be compared at the same loudness at 1KHz, possibly at 90dB at 1m. For some speakers with a 105db sensitivity 1W is way too loud and those who are at 80dB sensitivity it would be too soft. Then they would be compared at the same loudness and better reflect sonic comparison at the same normal listening level.

2) Frequency response at different loudness.
A music is not only one note and one loudness. The volume should be reduced by 10 db and freq response taken, and the experiment is repeated with lowering of volume. The microphone could be placed nearer (~10cm).
This is repeated until there is no sound produced or reached the noise floor. May reflect more of concepts such a micro texture and such. This will also reflect how the speaker will handle soft passages. I have only measured the Faiatlpro3fe22. But the frequency response changes as the volume goes down.

3) A string of 10 waveforms at bass frequencies. To see how long many waveforms it takes to reach full volume and how long it takes to decay off. Measured in the oscilloscope mode. Could tell a bit on how fast the attack of the bass will be.

4) A sinewave with a small notch and bump in part of the waveform. Measured in the oscilloscope mode. If you have a good set you should still be able to see the notch and bump in the microphone measurement. After all, we don't listen to sinewaves, and it is only appropriate that the hifi system reproduces the waveforms that was given to it.

Any thoughts?

Oon

 

[wesayso]

To learn more of what happens at the listening spot you'd need to measure there.
And at several other spots around it. How does your speaker interact with the room?
Can we improve on that?

One thing we can do as DIY is to choose a concept that works in our particular space.
Making a theoretical perfect performing speaker is one thing, but drag it into your room and all things change. So if you want to know why it sounds so good (or bad) compared to another speaker, measure both and compare at that most important spot.
How speakers interact with the room determines a lot of how they will sound. We DIY people can tailor that to our specific situation. But you'd need to be willing to dive deeper than looking at FR alone.

Measure it up close to see the potential. Measure at the listening spot and learn why the phase deviates so much. Learn to read your room and come to know where and when it influences what you hear.

At least, that's how I would do it...

 

[lcsaszar]

Here is the method I used:

- Use a pink noise generator fed into a power amplifier
- The power should be 1W on a 8 ohms dummy load (it can be measured with a true RMS voltmeter)
- Measuring distance was 2.5 ... 3 m, i.e. at listening position
- Use a 1/3rd (1/6th, 1/12th) octave filter (FFT with averaging or a Leq sound level meter)
- You get the frequency plot with room influence, this is the best you can get

Distortion, power compression, IM, phase (better said group delay), polar diagram, energy storage (waterfall diagram) etc. are more difficult to measure. Modern computer processing helps a lot.

 

[oon_the_kid]

To learn more of what happens at the listening spot you'd need to measure there.
And at several other spots around it. How does your speaker interact with the room?
Can we improve on that?

One thing we can do as DIY is to choose a concept that works in our particular space.
Making a theoretical perfect performing speaker is one thing, but drag it into your room and all things change. So if you want to know why it sounds so good (or bad) compared to another speaker, measure both and compare at that most important spot.
How speakers interact with the room determines a lot of how they will sound. We DIY people can tailor that to our specific situation. But you'd need to be willing to dive deeper than looking at FR alone.

Measure it up close to see the potential. Measure at the listening spot and learn why the phase deviates so much. Learn to read your room and come to know where and when it influences what you hear.

At least, that's how I would do it...

Makes sense that we can tailor it to our room response. It might be also arguable that the speaker will be built to give a flat frequency response when fitted in a standard room, say about 40m^2. (8m X 5m). After all we we don't live in an anechoic chamber. Maybe some of the hifi manufaturers already factor that into consideration, and already designed the speakers with room gain in mind.

However my intent for this thread is to look at tests that might differentiate speaker/speaker drivers from each other.

For example, if I were to give you 2 speakers with equal freq response (even polar), you would expect them to sound the same. But in reality, they will sound completely different. So in that essence, the tests we have done is insufficient to differentiate this two speakers.

One observation I have made through the years, is that many power hungry multiways with 10 inch woofers sound terrible when played softly. Sure they sound fantastic when played loud. But they sound so veiled when you listen to them softly late at night.

The opposite is true for a full ranger (say for example a TC9) , where it tends to sound good when played loud or soft. Although in some cases, when it gets loud (Lowther) , it tends to shout!!!

But where does it show it show in the specifications that the multiway sounds poorly at low volumes?

Similarly, for subwoofers, we know there are fast subwoofers (for music) and slow subwoofers ( for home theater). So what's the test to show which is which.

In short, I am looking to see if we could as DIYer with measurement microphones and software, devise devise certain tests that will better differentiate between speaker/speaker drivers.

Thanks.

Oon

 

[bwaslo]

Off-axis (polar) frequency response. You don't only hear the sound the speaker sends directly at you, you hear ALL the sound that goes other directions from it, too, within some 10s to hundreds of milliseconds. That sound isn't negligible by any means, yet that measurement tends to be ignored often.

 

[jerryo]

Measuring at 1 watt at 1 meter will provide technical information about the driver but it will tell you 'absolutely nothing' about how it will sound whilst sitting in the listening position as a stereo pair. The human ear is infinitely more nuanced at measuring sound than any calibrated mic. measuring system.
As soon as you start detailed measuring the sooner you are spending inordinate amounts of time fiddlng with equalising in order to rectify the 'problems' with the response curves. If that's your thing........fine, but don't forget to listen to the music and don't get too caught up in the technical short comings that you might get overly concerned with.

 

[xx3stksm]

It's a difficult question. A loudspeaker usually means both a unit and enclosure. But I think room environment must be included such as the size of room and position you listen to. They affect sound quality much more than a loudspeaker itself.

But It means a vendor needs to sell both a loudspeaker and listening space(a house). Measurement of 1W at 1m can less describe the potential of a loudspeaker.A DIYer can be superior to a vendor because of this.A DIYer can include their particular environment to sound quality.

As to the measurement method, I don't measure at 1W at 1m of course because will not sit there when I listen to music.I set up a mic at my usual listening position almost 10m from a loudspeaker. I use a pure sine wave below 200Hz and warble tone above 200Hz at the level of 95dB.

I sometimes use burst tone with 1second on and 1second off. This is for standing wave. What you hear in a closed space like a listening room where an indirect wave is dominating is standing wave. Progressive wave and standing wave have different sound pressure distribution.

Sound pressure of standing wave is not so proportioned to the distance from a loudspeaker because of an indirect wave. Many indirect waves will finally make up the sound you can listen. The time depends on probably the size of a room.It takes about 0.1second in my environment.

It's something like reverberation.But there are also "pre reverberation". I have found this by burst tone.It's reasonable both exist like the attachment.The measurement data highly depends on the environment. You can measure whole data including this.I think the method is particular one for your own environment.

 

[wesayso]

Makes sense that we can tailor it to our room response. It might be also arguable that the speaker will be built to give a flat frequency response when fitted in a standard room, say about 40m^2. (8m X 5m). After all we we don't live in an anechoic chamber. Maybe some of the hifi manufaturers already factor that into consideration, and already designed the speakers with room gain in mind.

However my intent for this thread is to look at tests that might differentiate speaker/speaker drivers from each other.

For example, if I were to give you 2 speakers with equal freq response (even polar), you would expect them to sound the same. But in reality, they will sound completely different. So in that essence, the tests we have done is insufficient to differentiate this two speakers.

One observation I have made through the years, is that many power hungry multiways with 10 inch woofers sound terrible when played softly. Sure they sound fantastic when played loud. But they sound so veiled when you listen to them softly late at night.

The opposite is true for a full ranger (say for example a TC9) , where it tends to sound good when played loud or soft. Although in some cases, when it gets loud (Lowther) , it tends to shout!!!

But where does it show it show in the specifications that the multiway sounds poorly at low volumes?

Similarly, for subwoofers, we know there are fast subwoofers (for music) and slow subwoofers ( for home theater). So what's the test to show which is which.

In short, I am looking to see if we could as DIYer with measurement microphones and software, devise devise certain tests that will better differentiate between speaker/speaker drivers.

Thanks.

Oon

Quite a while ago I had similar questions to yours. So I set out on a journey to find out more about it. This was my route and mine alone, so I won't say it's the way to go. But it has taught me a lot!

I started with the notion: everything matters! And even more important: Everybody is right! (at least somewhere we all are convinced of our own truths )

But the world of audio is full of contradictions, so how does that work? Personally I started by reading everything I could find. The papers from people like Blumlein, Toole, Keele, Dunlavy, Geddes, Danley but also people like Hiraga, Choueiri and many many more including the opinions of people on this forum. And on the Gearslutz forums as well as I wanted to know the opinions of the Studio people too. Even visited the Audiophile circles to see what they think.
I tried to find common ground in the several opinions out there. That's a pretty hard thing to find right there. Still trying to be open that every one of them could be right, at least somewhere.
After that I went on my own journey to visit the topics that peeked my interest the most. Most of it documented here...

I've learned a lot by doing things this way but by no means I'll say I have all the answers. But for a person willing to look hard enough, a lot of things do show up in measurements. If you're actually willing to dig a little deeper.

Most of the people I mentioned above (just a few names that spring to mind, I missed a whole bunch who would definitely be worth mentioning too) are busy with speakers and their performance.

For starters I'd recommend to read one of the many papers (actually a PPT in this case) by David Griesinger.
Most of his research is about spaces, rooms, venues and their relation to what and how we hear. Reading his stuff will make it clear beyond a doubt that wherever we listen, the environment is going to have an influence. And that influence just might be bigger than many assume it is.

This presentation also hints at the flaws of Stereo, as a concept it has it's own drawbacks. Knowing those limitations is equally important. Though as a concept, I'm still very much in love with Stereo. As it brings me closer to the music I love.

Disclaimer: this is just one person's way of looking at things. We are all free to go on whatever journey we want or choose. If we can accept our differences. My route has been backed with DSP while others would pursue the purity of full analog systems. Live and let live. As long as we enjoy the journey we are on. Like I said: Everybody is right somewhere .