Preliminary IMD measurements for subwoofers (1 Viewer)

[Ilkka R]

Here are all components up to 6th order created by 30 Hz and 78.5 Hz fundamentals.

108.5
48.5
18.5
127
138.5
187
217
97
265.5
168.5
205.5
11.5
67
175.5
247
295.5
41.5
284
198.5
344
325.5
145.5
374
277
254
37
422.5
228.5

And here are the harmonics of those two fundamenltals.

 Harmonics
 2nd60157
 3rd90235.5
 4th120314
 5th150392.5
 6th180471
 7th210549.5
 8th240628
 9th270706.5
 10th300785

 

[Ilkka R]

Here is the same list for 30 Hz and 70 Hz fundamentals.

100
40
10
110
130
170
200
80
240
160
180
20
50
150
230
270
50
250
190
310
300
120
340
260
220
20
380
220

And harmonics.

 Harmonics
 2nd60140
 3rd90210
 4th120280
 5th150350
 6th180420
 7th210490
 8th240560
 9th270630
 10th300700

 

[dave alan]

No, we're not. We're trying to figure out what percentage of output of a given sub is inharmonic non linear distortion (vs harmonically related non linear distortion, which is less offensive and even preferred by some, and well covered by the test for THD).

The tones should be deduced based solely on this criteria.

Dave

 

[Ilkka R]

By definition harmonical tones are multiples of the fundamental. And the THD is being calculated by comparing the total SPL of these harmonics to the fundamental.

30 Hz and 70 Hz (we can also try 30 Hz and 78 Hz) played at the same time create a number of inharmonic tones (inharmonic to fundamentals, but no to each other), which are compared to the total SPL of two fundamentals. And this way we get the IMD-%.

I think we mean the same thing, you are just confucing things with your 'musical harmonics'.

 

[dave alan]

I think I'm clear now, after you have confused things with your geek speak.

When you say 'fundamental' you mean the stimulus tone or tones.

When you say 'harmonic', you mean integer multiples of the fundamental, or stimulus tone or tones.

When you say 'non harmonic' you mean ALL other frequencies that are excited, besides the previous mentioned fundamental and harmonic frequencies.

Spectral contamination is simply a next step that uses multiple stimulus tones to examine the combination of all non linear distortions, viewed as a sort of noise floor rather than calculated as a percentage of total output.

So...what is Sub-C, or why won't you say? Seems silly.

Dave

 

[Edward J M]

While the same can be said for IMD, a larger database needs to be built for SC before any interpretation of the test results can be performed. Once you defined the level of the noise floor (this is part subject to individual interpretation), the %SC is easy to calculate.

 

[dave alan]

From the def I use, '...Since it is NOT HARMONICS of the original signal...described as "harsh and grating"...'

If the stimulus tones are harmonically related, then some resulting tones will also be harmonically related, so the device under test may show relatively high IMD but not sound as harsh and grating...thus the confusion to a musician

I would hope that the end result is that the frequencies in the formula are created to the same level in a given sub, regardless of the fundamentals used.

Dave

 

[dave alan]

It's gonna be difficult to build a database if Ilkka doesn't name the subs

Dave

 

[Edward J M]

He's only responding to comments from others that hiding the identity of the subwoofers (especially if they are competing internet brands) show impartiality and generates less hysteria and allows us to focus on the test results instead of the brands.

While I can certainly appreciate that POV, I personally don't buy into it...provided the tests were conducted under the same conditions (using the same test gear and software) for both products - which was the case here. The cat is 2/3 out of the bag in this particular case, but personally I don't care what Subwoofer C is - it's the test methodology we should be focusing on.

Ilkka obviously recognizes the limitations of indoor measurements and this thread was merely a tool to get IMD on the table and spark discussion and obtain constructive criticism and suggestions for improvement before hitting the wide open spaces for true ground place measurements. Then the identity of all tested subwoofers will of course be revealed.

I think the next step is to test 30/70 vs. 30/78 at the same sound pressures and see how things shake out.

 

[Parker Clack]

Roger and Robert SW:

I have increased your post count so you can post links now.

 

[Edward J M]

Yes, you are correct, but in this case 30/70 are not harmonically related in the mathematical sense - i.e., multiples of 30 do not overlap 70 or multiples of 70. And the same applies to 30/78.

What Roger was saying is - that for certain fundamental tones - some of the IM components end up being harmonically related to one (or both) of the fundamental tones, even though the fundamental tones are not harmonically related to each other.

For example, using 30/70, 3F2+F1 = 240 which is also the 8th order harmonic of 30 Hz and would thus be disregarded under the original example. However, using 30/78, 3F2+F1 = 264, which is not harmonically related to either fundamental and can thus be counted.

This is where using the Fibonacci sequence Roger recommended adds value to the process - because it eliminates the problem of IM components equaling harmonics of either fundamental. It is also why I use it for SC experiments.

 

[Shervin_Sabri]

Ilkka.....could "subwoofer C" possibly be the Monolith?

I agree with Ed 100%, if the tests were conducted in the same procedure, you should name the subs. Who cares about the brand!

 

[Edward J M]

My guess/vote for Subwoofer "C" is a Tumult-based DIY sub that Ilkka built. Failing that, my second vote defaults to the evergreen Bose Acoustimass module.

 

[Jack Gilvey]

In a sense, a competent, objective reviewer producing objective data is merely a messenger. If people get hysterical over the results in a knee-jerk defense of a brand or person, let that speak for itself. If said results are largely favorable toward a particular product, that also speaks for itself...attacking the messenger won't change it.

 

[Ilkka R]

Yes! Sorry about the geek speak.

 

[Ilkka R]

Guys,

I'm much more comfortable naming all the tested subs after I start making GP measurements. It is essential that the environment doesn't affect on the results at all and unfortunately I can not promise that while indoors. The purpose of this thread was to rise some discussion about IMD and how to measure it, not to test which sub has lowest IMD numbers or start collecting a database.

I do like the fact that 30 Hz and 70 Hz are musically harmonical frequencies, but I will run some tests with suggested 30 Hz and 78 Hz, and see if IMD levels change. I will pick the one which produces more IMD.

 

[dave alan]

Cool...

Now, in looking again at the speadsheet, there is f1+f2, which is OK, but later is 2f2+2f1, which would be the H2 of f1+f2, regardless of the fundamentals chosen.

Example:

30 & 70Hz...
(f1+f2) 30+ 70=100Hz
(2f2+2f1) 60+140=200Hz

30+78Hz...
(f1+f2) 30+ 78=108Hz
(2f2+2f1) 60+156=216Hz

In this case, I assume that the H2 of any 'inharmonic' is OK?

Dave

 

[Ilkka R]

Well naturally some IMD components are harmonically related, but it doesn't matter and can not be avoided, since the equations are what they are. Important thing is that they don't match with any harmonic of those two fundamentals.

 

[SteveCallas]

I know you've discussed test frequencies extensively over the past couple of pages, but I would imagine if using 78hz, the question has to be asked just how significant will the results be? The bulk of people crossover at 80hz, so even 70hz is already going to be reduced a bit compared to the output of a 30hz tone during "real" usage.

 

[Edward J M]

There are various permutations of the 0.618 "Golden Ratio" which Risch used for various SC/multitone experiments. Multipliers of 1.1618, 1.2618, 1.3618...etc. can be used, for example.

Using 1.7618 yields 30/53 Hz and 2.1618 yields 30/65 Hz, for example. I don't have the IMD spreadsheet in front of me to check for suitability, but you get the idea.