Preliminary IMD measurements for subwoofers (1 Viewer)

[steve nn]

Well I dont want to go so far as to say "cant we all just get along" but considering the real subject at hand (IMD) maybe it would be best just to move on. Most of us know from past experience nothing is to be gained except deeper wounds yet.

 

[Tom Vodhanel]

No problem at all Ilkka, As I already stated...sorry for hijacking your thread. There are so many falsehoods being spread about SVS lately...I just try to correct them as I'm reminded about them.

Tom V.
SVS

 

[Ilkka R]

Tom,

We're cool. And I know what you were trying to do, because I saw those quotes at AVS. But if someone hasn't seen them might be a little bit surprised what they are doing in this thread and where did they come from all of a sudden.

 

[steve nn]

Tom if I were in your shoes..I would most likely not be able to post at any of the Forums due to wanting to stand up for my Company and self..it's totally understandable. Only God knows how I would react. I dont think I would be as restrained as you generally are and thats a compliment.

 

[dave alan]

Ilkka,

Some further thoughts after the rude interruption...

In your report, you glossed over the reason(s) for the selection of 30 and 70 Hz. I wish you would expand on this. I am also a bit surprised that you and Ed didn't run it past a musician.

I also wonder where you got the definition for IMD (credit?)? It's not the best def I've come across, and actually a bit misleading, IMO. This def better represents my point...



30 and 70Hz is actually very, very close to an FMin chord without the fundamental.

This is the worst case scenario for an IMD test and may well be why you concluded that IMD and THD are closely related.

IOW, the tones used are not inharmonic, IMO, or at least not inharmonic enough for the test to be valid.

If you choose 2 better tones for the test (maximally inharmonic), you may see less of a relation to THD and better results for or against your hypothesis

In fact, why stop there? Yourself & Ed know how I feel about a spectral contamination test for subwoofers. It would combine the two tests (IMD and SC are very closely related) and go a step ahead toward quick eval of the sound of a sub.

Also, you may want to eliminate 50Hz from your calculations.

Dave

 

[Edward J M]

Hi Dave:

You need to remember Ilkka's post represents preliminary research into this subject.

The fact that he shared this research with everyone doesn't bother me in the least. With that said, it also shouldn't imply that this work is final. Peer review is an important part of research, and Ilkka is asking for constructive comments from everyone.

If you are suprised we haven't asked for a suggestion from a musician on the selection of the two test tones, perhaps the more appropo question in response would be why haven't you offered one?

The IMD spreadsheet I developed (and shared with Ilkka who subsequently tweaked it to included additional IM components) allows the selection of any two frequencies and it automatically re-calculates all the IM components. It would be extremely easy to select two other test tones. Indeed this has already been done; I've tried literally dozens of combinations of two tones.

My selection of 30 Hz and 70 Hz was initially strictly meant to satisfy a few basic conditions:

1) They are not harmonically related, which allows the separation and isolation of harmonic distortion components from intermodulation distortion components.

2) 30 Hz is easily still within the lower effective operating limits of most "real" subwoofers.

3) 70 Hz represents typical upper bass frequency content that would be seen by nearly all subwoofers using the popular 80 Hz XO frequency.

4) 30 Hz is above the tuning point of nearly all "real" reflex subwoofers, and thus does not bias reflex over sealed (or vice versa).

If you have a suggestion for two different test tones, I'm all ears (pardon the pun) and I know Ilkka would re-run the test data in short order.

Regarding your comments about SC - this is one of many tests I've been working on which is not yet ready for prime time. I've actually got reams of test data on SC using Spectra/Pro in extremely high resolution mode on several subwoofers I've recently tested. The acquisition of data and the development of criteria to evaluate the performance on an absolute and relative basis are two entirely different things.

Regards,

Ed

 

[Jack Gilvey]

Is it not the case that any distortions introduced by the amp will be manifest in measurements of the subwoofer?

 

[Ilkka R]

Why's that? Of course I checked the noise level and 50 Hz hum was at low level. It doesn't affect on IMD -% at all. And at higher test levels actual 50 Hz component goes higher than the noise.

 

[ChrisBee]

Forgive my intervention in a fascinating thread... but why the absolute fixation on comparison testing of single, commercial subwoofers?

While it is interesting to compare commercial examples of such exceptional perfomance... the simple addition of a second subwoofer, a third or a fourth will significantly reduce most of the comparison criteria from the argument. Wouldn't two or four lesser (and hopefully cheaper) subwoofers add up to a single, cutting-edge example?

True, the overall costs might be somewhat higher. But so what, in the context of absolute performance? Isn't this thread really just an argument about best performance value for money of single, competing, commercial examples? Rather than a discussion about the best obtainable sound quality?

Poverty of choice (and certainly of pocket) seem not to affect many mature AV/HT fans. Pairs of high quality subwoofers are by no means unusual. Though not everyone enjoys Frank's broadside of firepower. Others might well choose another route to bass nirvana. As he might well himself if he had heard one by now. Baffled? :wink:

Understanding exactly how subwoofers work and what makes one better than another is certainly interesting reading. The work of those who carry out these fascinating tests should never be undervalued. But let's not forget that adding a second, third or forth driver to the equation will reduce most of the factors under discussion to a fraction of the figures mentioned here. The effects of which will certainly be very audible.

Comparisons of single subwoofers is like comparing motorcycles. You can improve the camshaft and raise the compression ratio to your heart's content. But his very average 500 will still stomp all over your highly-tuned 250. :b

The Genelec is an extreme example of multiple commercial subwoofers stacked sideways. Isn't it time that stacking became the norm? Rather than squeezing the last drop out of single subs by pushing the technology and enclosure size to the very limits?

Single subwoofers are already pushing the limits of domestic acceptability in size. Surely their overall weight per unit has long gone beyond what is reasonable, or even sensible, in a typical domestic AV environment?

It's not that I don't value the overall improvements in driver performance that this commercial competition brings to all of us. It's just that I feel the competiton is somewhat artificial in the context of obtainable sound quality if limited to a single driver, enclosure and amplifier.

Regards

 

[Robert SW]

Hi ChrisBee,

In regards to:



I think most of us following this thread with INTENSE interest know that. It is just that we all are really "techy" types .

Bob

 

[dave alan]

Late night post...sorry for being so slow to get up to speed in this thread.

Ed...thanks for the explanation. I am aware of the fact that there is no standard for IMD testing of a subwoofer and that you only need 2 (or more) tones, and that those tones may be arbitrarily selected, so your thought process in selection of 30 and 70Hz is excellent info.

Actually, after getting some time to go over the frequencies used in the spreadsheet, I'm less inclined to disagree with the fundamentals chosen.

Still, there are 2 main points in my thoughts regarding inharmonic fundamentals:

First is the fact that a subwoofer typically is asked to simultaneously reproduce the low end of a music track and the random synth tones used for LFE. The fly by in AOTC comes to mind, but there are many such examples of these sorts of inharmonic tones being summed and sent to the subwoofer for playback.

Given that, it seems logical to select inharmonic fundamentals to best simulate the subwoofer under stress to measure IMD.

Second, when the various sum/difference freqs are looked at there is generally only a screening for repeated frequencies and not harmonic relationship. The test is run to reveal inharmonic tones that are added as non linear distortion to the input signal, the most offending kind, as they are more easily recognized as not original because they are non harmonic. A good example of this is the car horn. They were designed as a 2 horn sound and purposely configured to create a dischord, rather than a harmonic chord. This dischord captures your attention much more quickly than a harmonic dual tone.

If you look at a list of frequencies that are calculations of the sum and difference frequencies created by the simultaneous playback of 2 fundamentals, how do you calculate whether or not they are inharmonic to either fundamental?

It's not as simple as saying that 100Hz is inharmonic to both 30Hz and 70Hz because it's neither 30 nor 70Hz.

Example: As I first posted, A# (29.13Hz) and C# (69.25Hz) combined in a harmonic way to a chord that is not offensive to the ear. If you select 30 and 70Hz, it's very close to the example if the guitar (as an example instrument) is ever so slightly detuned.

The question in my mind is...are we really calculating the inharmonic tones caused by the simultaneous playback of the 2 fundamentals, or are we simply calculating the sum and difference frequencies, harmonic and inharmonic?

It follows that, if we really are going to create a standard for this test and use it's results to evaluate part of a subwoofer's performance, should we be concerned enough to select the frequencies that would create the most inharmonic tones in the sum/difference spreadsheet, or simply be happy to let the spreadsheet calculate the sum and difference tones, harmonic or inharmonic, to 'X' order?

BTW, Ilkka...my saying 'worst case scenario' was a very poor choice of words...sorry.:b Please insert the required IMO, I think, my thoughts, etc., into all of my posts on this subject.

Dave

 

[dave alan]

Man, I'm gettin' old, but I swear I'd already done that. It was indeed a calc of sum/diff using 2 particular tones. I'll check my files. Possibly I didn't actually forward it to you, but I sure thought I did.

At some point in this thread, a sample set of tones (or 2 or 3) will come up for review, I promise.

Dave

 

[dave alan]

This post is aggravating me and it needs to be deleted.

Dave

 

[RogerW]

Just a suggestion, Ilkka, but why not use tones based on the Fibonacci sequence as per Jon Risch's paper. Since I'm not allowed to post a link, do a Google search on "A NEW CLASS OF IN-BAND MULTITONE TEST SIGNALS" and look for Jon Risch's site.

Say, use F1=30Hz and F2=F1*2.618 or 78.54. Then the IMD products should not coincide with any of the harmonic or other IMD components. Just omitting an IMD component that happens to coincide with a harmonic seems a little arbitrary to me.

 

[Edward J M]

Indeed, a variation of the Fibonacci series listed in the Risch papers is what I used to generate the test frequencies for the spectral contamination test tones.

30 Hz and 78.5 Hz is easily doable and still satisfies the other previously listed criteria; thanks for the suggestion, Roger.

These two frequencies yield the following IM components:

108.5
48.5
18.5
127
138.5
187
217
97
168.5
11.5
67
247
284
198.5
344
374
277
254
422.5

The other interesting thing not completely shown on Ilkka's screen shot is the generation of a fairly large VLF IM component (10 Hz previously, and in this case 11.5 Hz).

A single VLF tone (say 5 Hz) is something Mark Seaton had suggested introducing as a 3rd modulating test frequency. I was hesitant to do so because (due to high pass filtering and/or near-DC protection) many commercial powered subwoofers will not even respond to such an input (or if they do the output will be insignificant).

Perhaps the generation of a VLF IM component - which will to some extent interact with the other IM components - is a reasonable compromise in this area. With that said, it is unlikely these secondary and tertiary IM components will be of sufficient sound pressure to affect the overall %IMD calculation.

 

[RogerW]

Ed,

You also need to revisit the spreadsheet, and put back the missing IMD components that were excluded as clashing with harmonics. e.g., 3F2+F1, 3F2-F1, 2F1-3F2, 2F2-3F1 etc, etc,

It is also worth noting that some of the higher order components still fall within the measurement zone. e.g., 5F1-F2 so I would consider extending the spreadsheet to include any of those that are relevant, say limiting oneself to combinations up to 5th harmonic above which the amplitude is likely to be so low as to contribute little to the total IMD %.

Finally, if individual IMD components can be kept above 10Hz, they should fall within the calibrated measurement range of our respective test rigs, and therefore measurable with some accuracy.

 

[Edward J M]

Yes, the mic, sound card, and software are all accurate to somewhat below 10 Hz, so 11.5 Hz shouldn't be a problem.

 

[Ilkka R]

I'm not sure IMO would help that, but what ever...

 

[Ilkka R]

That is a good suggestion. But like I said to Dave, when using 30 Hz and 70 Hz fundamentals, 2nd and 3rd order components are "clean", so if we need to omit for example two components from the 4th order components, it won't affect on total IMD at all (maybe 0.0x% at most).

 

[Ilkka R]

There was a pretty strong 10 Hz component at highest 120 dB level. For example Ultra produced around 82 dB there (~1.25% IMD).