Active crossover for 804S and Rythmik subs?


Hello everybody.

I've been thinking about this quite a bit and would like to validate with experienced users. This step entails relieving the speakers and amp from playing below 100/150 Hz, approx. Sorry for the long ramblings.

My system is made of Lamm LL2 pre, McIntosh MC275 amp, B&W 804S speakers and two Rythmik 12"-subs.

I tried the subs as reinforcement, meaning from my pre going straight into the amp and in parallel also straight into the subs. I also use REW (in-room response measuring software), and a measuring mic, and by tweaking the Rythmiks controls achieved a rather nice in-room frequency response.
Lately I tried going from pre to each subs internal x/o and from there to the amp. The subs plate amp include a crossover with a fixed point at 80 Hz, so the amp now sees 80 Hz and up. While the measured frequency response is not as even, I think I like it better than in reinforcement configuration.

So I'm thinking a better x/o and shorter/better interconnects should help, and that the key constraint is the sub's frequency response at the higher frequencies.

A guy over at hometheatershack.com made a ton of sub measurements and tested the 12" Rythmiks. They are flat until 70 Hz and drop 5 dB by 100 Hz, and an additional 7 dB by 200 Hz. http://www.hometheatershack.com/forums/subwoofer-tests-archived/5756-diy-rythmik-audio-direct-servo-12-sealed-56l.html

I can't find measurements of the 804S, but the 804D, the newer model, was recently measured by John Atkinson. http://www.stereophile.com/content/bowers-amp-wilkins-804-diamond-loudspeaker-measurements
Kal Rubinson mentioned on his review the 804D and the 804S both sounded very similar and were spec'd the same. Maybe safe to assume they have similar bottom end response...

So going with the 804D measurements, they drop about 4 dB at 50Hz compared to their level at 100 Hz.

I guess all this means is the best x/o point is likely to be between 150 and 100 Hz. Right?

What else should I take into consideration?

I'm thinking of using a Marchand XM44. Can you think of better options? I would like to keep it under $1500.

And if all the above ramblings were spot on (probably not!), then what slope should I use for the XM44? They offer 6, 12, 18, 24, 30, 36, 42, 48 dB/octave.

Thank you!
lewinskih01
One of the most transparent HI/LO pass filters available, since the 80's, is the Dahlquist DQ-LP1. The High-Pass is passive, the Lo-Pass active. One of the advantages; there is just one capacitor(and perhaps a resistor in parallel, if needed to match your amp's input impedance), which is actually the filter before the mains amp. You can pick the quality of that capacitor, and thus the transparency. The are rare, and thus bring more than when they were introduced, if in excellent condition. I did a variety of mods to the one I used for almost 30 years. Probably the most important, was to change the RCAs to gold WBTs(very simple, if you can solder) This site for info: (http://www.allegrosound.com/Dahlquist_DQ-LP1_AllegroSound.html) These guys used to work for Dahlquist and still offer support: (http://dahlquistspeakers.com/Dahlquist_Speaker_Service%20DQ-LP1.htm) Here's a site that offers the manual: (https://sites.google.com/site/mpbarney/home/dahlquist-dq-lp1)
The Marchand is quite good and since you don't need additional sub controls, should work well.

If you just want a high pass between the preamp and amp for the main speakers, you can look at Harrison Labs.
There's one on eBay: ( http://www.ebay.com/itm/Dahlquist-DQ-LP1-Active-Passive-Subwoofer-Crossover-/221445734709?pt=Vintage_Electronics_R2&hash=item338f31c135 )
Hello Rodman.

Thanks for the input. I have come across this unit before. I remember you liking it.

It is not clear to me how this unit is operated. The frequency knobs are for adjusting the low-pass x/o frequency, right? Is there a way to adjust the high pass if there is only a cap on that side of the signal?

Sounds like a purist approach to active x/o. I wonder why there are no more, newer, newer units employing this approach?

Cheers!