ScanSpeak SPL Graphs

[LewisH]

I'm considering a couple of different subs for my upcoming build. This one is currently the forerunner because I like how flat it is in the lower region: https://www.scan-speak.dk/datasheet/pdf/32w-4878t01.pdf

My thinking is that it might sound more "musical" (whatever that means) than it's brother: https://www.scan-speak.dk/datasheet/pdf/32w-4878t00.pdf (opinions welcome, please). I say this because I only need it to play to 40Hz (I don't have music lower than that - I measured it!) and the curve is flatter in the pdf (maybe this doesn't matter by the time it goes into a box?).

However, I'm curious about the SPL graphs, particularly with regard to the fact that the 32w-4878t01 model seems to graph (in the pdf) with an f3 below the vented box modelling (also referenced in the pdf). I wonder if the mechanics of it just work better with the way they have them set up to generate the graphs (in which case, why not use those to generate the stats?), or if there's something else going on here. Also the impedance peaks don't match the resonant frequencies of the respective speakers in ScanSpeak's SPL graphs across many of their drivers - it's my understanding that the peak impedance is at res frequency.

I don't really know what my question is to the forum, but my gut-o-meter is telling me that either something's wrong or there's (likely) a bunch of things that I don't know that I should in deciding how each of the two options will sound.

I'm keen to hear people's opinions, thoughts, experiences and scientific knowledge to help set me straight.

Actually, I do have another question in this region: In the case of the driver with the slope from 100hz down (https://www.scan-speak.dk/datasheet/pdf/32w-4878t00.pdf), if that's in a vented box tuned to an f3 of 21Hz, my modelling software infers that it will be flat from 100Hz to (say) 30Hz then roll down from there. How does that translate to reality. Will the actual curve slope from 100 steadily down to 21Hz @-3db or will it play flat until it gets closer to the f3 frequency? Maybe I need to model above a flat line to account for the same db drop in the curve in the pdf?

Sorry if that's all a bit rambly. I'm just trying to get this clearer in my head, so I'm vocalising thoughts a bit here. I realise there's a lot to cover in this one post.

Thanks

[mcargill]

They both have the same mechanical limit on excursion, but one has twice the linear excursion limit (Xmax) due to a taller voice coil. This means that it should have less distortion if you have to play it louder, and use that extra Xmax. If you don't need that extra excursion, then the other one might sound slightly better, being a little more sensitive and lower Qts.

Scan-Speak uses the same baffle and same rear chamber for testing all of their speakers. They do not optimize the rear chamber to produce the flattest response. Some woofers will therefore measure flat only because they happen to be of the right specs for maximally flat response in that size box. If you model a different size box, or a vented box instead of their sealed one, then you will get a different response.
https://www.scan-speak.dk/datasheet/...-explained.pdf

If you will be using any EQ at all for the subwoofer (which is almost a necessity, given variations in room response), then the native response of the speaker and box don't matter much, because they can be corrected. That's why it is better to design subwoofers for minimal distortion than for flattest response before EQ.

-Meredith

[billfitzmaurice]

Modeled results are 1/2 space anechoic. That's what you'd get with the sub outdoors at least 50 feet from any buildings. Indoors you'll get considerably better low end sensitivity, along with boundary reflection sourced nulls, so don't take the model as Gospel.

Where ScanSpeak is concerned I wouldn't spend that kind of money on a subwoofer driver. You can spend a third their price and get just as good a result. Then you can afford to make at least two subs, which is a necessity to smooth out room modes and boundary reflection sourced nulls.

[LewisH]

I find myself lured (rightly or wrongly) to ScanSpeak's super flat SPL graphs. When I look at other brands, they look comparatively lumpy. I appreciate that this might be in part due to how they've chosen to smooth the points on the graph. Can you point me to some examples of superior cheaper sub's please? (I appreciate that "superior" is subjected and related to a project / use of the sub, but perhaps you might give some examples anyway?) Thanks!

[billfitzmaurice]

There's a few to be found here. Model them to find what best meets your needs. https://www.parts-express.com/speake...eter/12?page=2

This looks particularly good at a reasonable price, allowing you to make two subs for less than what it would cost for one Scan-Speak. https://www.parts-express.com/Dayton...r-Coil-295-166

[LewisH]

THanks for that. I'll investigate