The results are in . . .

Hello again everyone,

I need to discuss two things. Firstly, we had a great competition last Saturday and I think everyone had a lot of fun. It was a different event than most people are used to but it was very helpful for me to get some real world data that can help calculate whether or not displacement is an accurate indicator of SPL (see the previous post for more info on that issue). So the first item of business here is to discuss the final scores and what they mean as it relates to the SD vs. displacement discussion.

But, since I am fan of doing things backwards, I want to announce the second round of competition for the summer. It will be held on August 20th, 2011 here in Logan. The exact location is TBD (we are working on getting a larger parking lot than last time). The format will be largely the same, so far as classifications and such go, but the qualifying details for the classifications will be changed slightly. Read on for for those details. I will post all the specifics on the blog in the coming days, and also I will make an announcement on the Lynn's Facebook page. So keep your eye out!

Now to the important part: the results.

I entered everything into an Excel sheet and averaged scores, displacements, SDs and finally ratios. Here's the table:

Competitors will recognize their competitor number and be able to know where they are in the list. Now, using this data I plotted a graph to show the distribution of the scores and evaluate which would be the better indicator, or predictor, of performance. Here's that graph:

Clearly the displacment plot, the blue line, exhibits less variation than the SD plot. Also notice the R^2 value is slightly higher on the displacement plot, and ideally an R^2 value should be closer to 1 to indicate higher linearity. That the displacement curve shows less deviation and a higher R^2 value supports the claim that this is a more accurate indicator of actual SPL performance.

That said, let me qualify a few things. First, displacement is very frequency dependant. I didn't delve into that much on the previous post but the graphs I used indicate that peak SPL varies per driver and only occurs at certain frequencies. Competitor number 25 proves this point (as you'll likely notice two other competitors do). His was one of the anomalies in the graph. Simply put, his score was acheived by capitolizing on the resonant frequency of this vehicle rather than relying on the displacement potential of his woofer. While most competitors didn't go the lengths of evaluating their car's transfer function, he did, and in this case we see that displacement modeling only works in theory. It has limited value in practice simply because there are too many other variables that effect real world performance. For serious competitors who will consider their vehicle and environment as an integral part of the system, frequency becomes the great dependant variable and will ultimately trump the value of displacement. If a competition used frequency weighting as part of it's criteria for classifications, and these criteria focused solely in the lower tone ranges (say, 40 hertz and below), Xmax times SD would become the determining factor in SPL performance. Many of the cars in last Saturday's event prove that point as most of them used music or tones that played in this range. In most cases, these tones are below any resonant frequencies and thus the woofer is left to its own devices, namely its net displacement, to make SPL. But given the sample size we can't assume that this will always be the case. As with any scientific endeavour, the outliers often disprove the rule. In this case, the anamolies belie the possibility that displacement modeling can't accurately, or fairly, be used to determine classifications.

While I still consider displacement the more accurate methodology to evaluate real world subwoofer performance, my opinion has shifted to focus on non-competition settings, and the fact is that for our purposes it isn't going to work. There are two reasons for this. First, as I just explained, there are other variables that effect its value. Second, and probably more importantly, it isn't more accurate enough to make it worth the added effort to calculate. Notice the R^2 values; the displacement model is really only about 15% more accurate, and that's based on this sample size. I suspect that, with more data analyze, the difference would shrink to something closer to 10%. That said, it's a lot of extra work for the judges to crunch all those numbers, especially the day of the show. If this factor was the true determining feature of SPL capability, by which I mean if the accuracy of displacement modeling was 2 times greater, or more, than SD modeling, this effort would be worthwhile and I would continue to do it. That it is only really 10-15% more accurate means that it isn't worth the extra effort. Fairness can be maintained using SD only.

All of this to say that for purposes of our next event we will be returning to the more traditional SD classification system as employed by DBDRA. So the breakdowns will be as follows:

Street I:
1-10"
2-8"

Street II:
1-12"
2-10"
3-8"

Street III:
1-15"
2-12"
3-10"
4-8"

Street IV:
1-18"
2-15"
4-12"
6-10"
8-8"

Street X:
Anything bigger

We will also use the Modifided rule system as before; however the power requirement will be moved to 1000 WRMS instead of 500. Otherwise the guidelines will remain the same (charging system upgrades, sound reinforcement, purpose-built SPL box and RMS power). I will develop and post more detailed explanations of these rules so that there is less concern over what they mean specifically. I don't get the feeling that anybody had any concerns over the division between Street and Modified, though, so perhaps this is unneccesary. It seems that the only real debate revolved around the displacement model which we will abandon for future events.

Thanks again to everyone who came, and I really hope to see you all out again in August!

Risk