IB Distortion

I was browsing around the various audio forums when I saw a thread about a Tuba HT, folded horn subwoofer. The builder showed what looked remarkably like an REW screen measuring distortion "live" on his YouTube video. If memory serves he was measuring 22Hz at 122dB! His house and courage must be far stronger than mine!

I wondered why I hadn't tried measuring IB distortion myself. It couldn't be that difficult to produce some meaningful distortion figures of my own for my 8 x 15" IB.  I certainly remembered that REW could produce distortion graphs and figures for up to 8 harmonics but the actual method had become vague from lack of practice. Thankfully John Mulcahy's REW help files covered the subject in sufficient detail to make me dangerous. It seemed easy enough to make it work at a basic level even if my understanding of the purpose of the variable controls were still a bit fuzzy.

I have used italics in the paragraph below to indicate REW controls and settings: This may help those unfamiliar with REW but bent on trying to replicate my measurements technique on their own subwoofers or even their speakers:

Having already Calibrated REW using my Galaxy 140 SPL meter and my IB on pink noise (with the SPL meter at the listening position ) I knew I had a safe level to work from. The REW Generator is opened and set to produce a sine wave with Frequency follows cursor for ease of use.  Spectrum is selected from the test options just above the graph box. Then the Distortion measurement box down below the graph box is ticked. Once the chosen test tone emerges from the subwoofer the red Record button is pressed. Now a dancing, "boxy" curve is produced in the REW graph space. I played around with the settings then chose RTA 1/6 octave to provide the best looking graph. (IMO of course) The number of sample Averages is also available as an option. This steadies the constantly shifting curve until (eventually) it becomes fairly fixed and  "snapshot" can be taken and saved to Windows Documents.  

As it is a horribly rainy day my wife (The Head Gardener) is stuck indoors. So no window breaking SP levels were attempted for the moment. Still, I think the results are interesting enough to share even though they only peaked at around 100dB(C). If this figure changes I may have summed up the courage to try even higher levels.

I was actually quite disciplined, for once, and saved the graphs and images at each test frequency. It is quite normal for me to omit recording the blindingly obvious. So being able to resuscitate REW graphs and measurements, at will, is a vital characteristic of this amazing program. I haven't quite grasped how to Copy and Paste the numerical values for instant regurgitation onto a blog so it will mean typing them out longhand. The numerical distortion figures and test conditions all appear just to the left of the REW graph window. So they aren't lost provided the individual graph is saved. I had hoped I could copy and paste the figures for all distortion harmonics straight into a useful format here but it seems not. The time it takes isn't the problem just that I fear inaccuracy creeping in as I carry them across. (this involves maximising the REW page, noting the figures, then maximising the blog page again to enter the details in the table) For the moment I have simply listed frequency, Total Harmonic Distortion and SP Level. I may list the separate harmonics later if I have sudden burst of conscience.

First, an image of the (All) Measured plots for 10Hz upwards @ 75dB until lunch intervened at 50Hz precisely. This is simply all the curves for each test superimposed on one graph to give you an idea of what I was seeing on the individual tests.

The vertical pillars in the graphs indicate the level of distortion harmonics around each fundamental from 10Hz up to 50Hz. The shoulders of all test frequencies are remarkably "clean". As is witnessed by the very low numerical figures for harmonic distortion listed in the table below. No change was made to volume levels throughout the nominal 75dB range. The variable SPL readings are the natural result of output at each frequency depending on the IB's own frequency response. Later I wound up the volume on my preamp to achieve much higher levels. I also stepped up the Galaxy 140 to the next higher range. (50-100dB) Switching ranges on the Galaxy affects the meter output voltage which REW reads. So it is best to match the range to levels actually being measured. Or the measured curves fall below the bottom of the graph window.

Thanks to my flimsy baffle wall vibrating there are some serious structural rattles at higher SPLs which may be contributing to the measured harmonics! I shall snatch higher SPL readings when my wife is absent so do remember to reload the page to ensure you get the latest version. (assuming you are a regular visitor here) The 10Hz measurements were taken last.

10Hz THD = 4.4%  @ 75dB _ 10.33% @ 86dB_33% @  88dB

15Hz    "   = 2.75% @ 77db  _2.75% @ 90 dB_3.65% @ 95dB

20Hz    "   = 2.05% @ 77dB  _2.34% @ 95db_2.65% @ 100dB

25Hz    "   = 1.34% @ 80dB  _0.81% @ 93dB _1.08% @ 98dB

30Hz    "   = 0.3%   @ 81dB  _1.96% @ 90dB_ 0.67%@ 95dB

35Hz    "   = 0.85%  @ 82dB _1.03% @  92dB_1.95% @ 95dB

40Hz    "   =  1.40%  @ 84dB_1.38% @  92dB_1.46% @ 96dB

50Hz    "   =  0.81%  @ 85dB_0.95% @ 90dB_1.22% @ 95dB

I hope this table comes out of the wash in good shape. There is no telling what others see on their screens. I have glanced at my blogs on other computers and it is surprising how such tables are folded to a newline on some machines. I have deliberately made the table narrower than the standard text width.  Guess what? The editing text width is much wider than the published blog! If you'd like the text and images larger just click on CTRL+  few times. CTRL&0 will return everything to normal size.

BTW: 10Hz is the minimum frequency available for these distortion measurements. It is also the barrier beyond which the cursor's control of frequency will not go. The IB doesn't like it either as distortion goes silly with rising output. A few dBs and distortion doubled.

A disadvantage of using prolonged test tones is rapid fatigue at higher SP levels. Using fewer samples for averaging may help here. As does clicking the red button to freeze the graph followed by a click on the generator button to stop the tone. I can still hear some of the higher level tones long after they have ceased!

Some representative harmonic distortion graphs at different frequencies appear below. Left click for much enlarged images. (only 55kB)

Here you can see a worst case scenario as all 8 drivers reproduce 10Hz at 86dB at the listening position. The door to the IB enclosure is moving about half an inch and output rising only very slowly as I crank the volume control on the preamp beyond half way. The baffle wall and door are obviously limiting output. Probably due to being out of phase with the manifold. At 88dB REW was measuring 33% THD!


10Hz again but much better behaved at 75dB.

My wife returned from a quick slosh around the waterlogged garden to report she could clearly hear the test tones rattling the house while she was outside. I have a headache now after several  hours of running back and forth between the computer and the volume control. The odd thing is how different objects and surfaces vibrate in sympathy with particular frequencies. My glass-topped computer desk shakes like a big diesel engine at 25Hz. Tones below 20Hz are silent to my ears so don't bother me. As the frequency and levels rise the noise becomes very uncomfortable indeed. I could wear ear defenders but would worry about not being able to closely monitor structural vibration and loud rattles!

My summary is that distortion is low above 20hz but rises rapidly below 15Hz with increasing output. This is probably structural and exacerbated by the flimsy construction of the enclosure area. I inherited the structure and doors from the previous owner. Only adding a glazed gable end to close off the roof against prevailing winds. This had once been a completely open balcony with a view over the garden below.

These results make me even more determined to distribute smaller manifolds elsewhere to reduce the pressure (sic) on the existing IB enclosure. This would also better distribute the bass and quite possibly overcome the present trough around 150hz.  My attempt to use 8 drivers has killed the IB goose which laid the golden SQ eggs. Such flagrant abuse of the IB enclosure sizing rules deserved no better reward.

Given a solid concrete box in a basement to contain the pressures involved I might have got away with it. Probably not. It would become a large sealed box rather than an IB. IB drivers aren't designed to fight an air cushion. It is not worth reinforcing the baffle wall as it stands because the antique, glazed, double doors are by far the weakest point. Totally rebuilding this area as a heavy stud wall with a single, modern, fire door might improve matters but I have no such plans. More light would have to be found to replace the large area of glass.

My wife informed me that she could clearly hear Bass Outlaws "Stereo Bass" at the bottom of  the garden as I played it at an indicated 100dB(C) on the Galaxy SPL meter.. This never occurred with only four 15" drivers. So the 8 driver IB must be putting far more energy into the roof structure. This is not a criticism of the IB principle. It just demands far more respect than I have been showing. I pushed the boundaries of undersized enclosures almost to destruction. I lost (almost) as much as I gained by doubling the number of drivers. Though the SQ still improved.  Note to self: I must try harder not to break the (IB) rules. ;-)