I'm new here, but would like to share the results of an experiment described below that showed to me that a cap constructed out of several, smaller (different value) capacitors put in parallel offered a superior result over the single value cap. With all the smaller caps being of the same brand and type that is!! So my estimate is that 2 x 7,5 uF will have a better result than the single 15 uF. Even better will be something like 1x 8,0 and 1x 4,7 and 1 x 2,2 and...
About one year ago a friend and I, who share the same speaker (the infamous magnetostatic Apogee Scintilla, 1 ohm version), made an extensive comparison between multiple caps in parallel versus a single cap. Capacitors used in this experiment: copper cased Obbligatos and Sonicaps. First we tested the caps in the treble section.
The Scintillas treble section (6 dB slope, >3000 Hz) contains two 10 uF caps. We made an external set up in which we could switch quickly between the two 10 uF caps constructed with:
1) single 10 uF Sonicaps
2) single 10 uF Obbligatos
3) 10 uF caps built up out of Obbligatos: 4 x 2,2 uF, 2 x 0,47 uF and 2 x 0,047 uF
We played some cds with a lot of HF information. Especially a cd with metal bells hit by a metal stick reveiled some interesting differences.
The difference between the single Sonicaps and single Obbligatos turned out to be relatively small with the Obbligato being the better cap. I feel our result was in about the same league as the judging of Tony Gee (8,5 versus 10) (google for captest and humblehomemadehifi). The biggest difference however was between the single cap and the multiple cap. The multiple cap turned out to reveil micro-detail like the singing of the metal after it was hit, much better than the single cap. Also the spatial information was better with the multiple cap, the soundstage expanded in all directions compared to the single cap. The most prominent feature however I experienced was that the music as a whole became more enjoyable/natural sounding. Of course it is very hard to tell, but I would not be surprised that the performance of a multiple Obbligato in Tonys test would have been rewarded somewhere around a 12-13.
In front of the midribbon of the Scintilla (300 3000Hz) is one big capacitor: 240 uF (yep, Im saving my money to have this cap one day replaced by a multiple parallel Duelund CAST Ag PIO...). In the stock Apogee xover this cap already is a multiple cap (24 x 10 uF metalized polypropylene Spragues in parallel). My friend replaced these Spragues a few years ago by Sonicaps (24x 10 uF) and he was very pleased with the result of this. Recently he took out 30 uF of Sonicaps and replaced these by multiple Obbligatos. This experiment turned out to be very dissapointing. Disappointing as especially some strange things happened to the imaging. It shifted. After a few weeks the Sonicaps were put back in place and the problem was solved.
This experiment made me conclude that 1) multiple parallel caps perform significantly better than single caps, and 2) it is essential to use one brand/type of capacitor.
BTW: I think it was on the Duelund website that I once noticed a picture of a xover made by some super-high-end loudspeaker company that built their caps with different value VSF-caps...
Kees