Almost a year ago I started this thread after having great success optimizing a design and component selection for crossovers for the DBL. We were all pretty much in agreement that the midrange and tweeter are at best mid-grade units. But with these crossovers and finally introducing a balanced power isolation transformer (10 kVA worth) they took on a greatly enhanced degree of purity and extension that pushed back my inclination to muck around with the drivers.
Months ago I had some custom built 5.25 inch drivers made at circa $200 USD a piece as their 6.5 inch bretheren are amongst the most neutral and transient preserving drivers I have heard. These were custom designed by the very guy who initially designed my crossover and were the product of over 5 years R&D and many, many iterations before they actually met the specifications he provided.
So yesterday I took my DBL upper enclosures over chez Daryl to see how they would respond to replacing their midrange units and what revisions of the crossover would be required.
First off...the driver was too tight of a squeeze to drop directly into the DBL cutout...which had to be filed ever so slightly to allow insertion. Secondly the screw holes were maybe 1 mm different in location from the stock drivers, to the holes of the replacement drivers had to be elongated via a slender file-more work.
Once the new driver was inserted a frequency SPL sweep was done. Oddly enough there was a huge suckout (17 dB) at 1.2 kHz that no amount of crossover modelling would adequately correct. As this driver is about as linear as could be, this deviation HAD to be cabinet related. The stock driver had an in-cabinet response that was rather more linear and certainly had no such suckout.
So the stockdriver was then mounted to a flat surface and a free air SPL frequency sweep done. This sucker has a 22 dB resonant peak at the very 1.2 kHz. However the way it is mounted in the cabinet is on a compliant baffle that has rubber sealed slits around it which primarily serves as a mechanical notch filter rather than to mechanically decouple it from cabinet vibrations as claimed. Which means the ONLY way to replace this driver would to be to find one that has a huge peak at the same 1k zone-which excludes about any other 5.25 inch driver extant. I briefly contemplated sawing out the baffle containing this complaint mount and replacing it with a glued-in fixed board that would then allow any driver to be mated to the DBL, with a very good idea that this would allow the benefits of a superior driver to be quite evident. Sanity prevailed and I returned the enclosure back to its rightful place top portion of the DBL frame. My monkeying with the crossover and internal wiring is about as far as I am willing to go.
I can only speculate what the result would have been if Naim had decided to use a higher quality, more linear and costlier midrange driver mounted in a more conventional fashion. Intuition says that having a compliant mount would only cause loss of detail and diminished transient reponse.
And here are the sweeps....
Stock midrange driver SPL sweeps. Free air = purple; in cabinet = green
The upper sweep shows the free air response (purple) and in-cabinet reponse (green). Note the free air response has a very large resonant spike centered at 1.2 k with shoulders to 800 Hz and 1.7 kHz.
The green in-cabinet sweep has no indication of any residual peak, as resonance (and musical information) in that region has been sucked out. The efficiency of the lower midrange is circa 5 dB higher than that of the upper midrange as previously noted.
SPL sweeps of aftermarket midrange driver in cabinet (red) vs stock driver free air (purple).
The lower (red) sweep shows the in-cabinet response of the aftermarket midrange driver with a 17dB suckout at the 1.2 k derived from the compliant mounting. This is compared with the free air reponse of the stock driver showing the peak in precisely the same place as the cabinet-derived suckout of the former.
Oh well....it was worth a try.
Having said that, the DBLs are sounding great as I type...but still I wonder. And wonder.