Home Theater Forum and Systems banner

1 - 7 of 7 Posts

·
Premium Member
Joined
·
1,923 Posts
Discussion Starter · #1 · (Edited)
I put together a set of speakers a couple years ago (3-way vented floorstanders with Dayton RS28A-4/RS150S-8/DC300-8) with a xover that made a nice flat response in PCD, using raw driver response data. Since I've came to fully understand BSC, a BSC correction circuit's been added. This shows the difference when you design a speaker using only raw driver data vs. data with baffle effects accounted for.

The gray overlay is what the response was with raw driver data and no BSC-- looked nice! Looks like a 90dB/1W speaker with +/-1dB full range.
The blue line is the actual response with the same crossover but with the actual enclosure, baffle step, and phasing accounted for in the woofer and midrange using the Bagby Response Modeller-- Probably much closer to reality! Notice the deep sag in the LF range, where the lows begin to radiate in all directions casuing a loss in directional acoustic power. This is baffle step. Also notice the waviness added in the midrange, this is due to phase effects after simulating the driver in a small sealed enclosure, including the actual baffle size.



From there, I pulled the summed response and impedance/phase from the more accurate model and created a new PCD project so I could easily model a BSC circuit as a series RL filter. This is what I came up with:



Goes to show, a 90dB ~6ohm +/-1dB "ideal" (or in-wall) speaker is actually an 85dB ~8ohm +/-2dB speaker after put into a box with a BSC circuit.
 

·
Registered
Joined
·
1,747 Posts
Goes to show, a 90dB ~6ohm +/-1dB "ideal" (or in-wall) speaker is actually an 85dB ~8ohm +/-2dB speaker after put into a box with a BSC circuit.
Indeed. What's more, it goes to show how placement sensitive a speaker can be. That's why I love digital active crossovers as well as commercial offerings with placement settings (Revel and Atlantic Technology have this)
 

·
Premium Member
Joined
·
1,923 Posts
Discussion Starter · #3 ·
Indeed. What's more, it goes to show how placement sensitive a speaker can be. That's why I love digital active crossovers as well as commercial offerings with placement settings (Revel and Atlantic Technology have this)
Yup, speaker arrangement and spacing for a multi-way speaker system is critical. That's one of the reasons why PCD is so usefull, because you can design an xover around known driver spacing.
 

·
Registered
Joined
·
169 Posts
fusseli, do you use active or passive BSC circuits? Did you try to use the " linkwitz transformer network " somewhere?
 

·
Premium Member
Joined
·
1,923 Posts
Discussion Starter · #5 · (Edited)
The only BSC I've done is passive and either built into the xover (if starting from response data with baffle step accounted for) and designed for from the beginning, or added after the fact like this one. This one is a simple shelving circuit added in series before the rest of the speaker/xover. Prior to adding this BSC, I faked the BSC with a graphic equalizer in the signal chain.

I started with a calculator and then fine tuned the componet values in PCD to get the most even and balanced compensation. The only way to guess at BSC is to simulate/predict the baffle effect or to take actual measurements of the driver in the actual enclosure being used. Myself, like many, don't have measurement rigs so simulation is the best bet and will surely provide far better results than no BSC.

The speaker in the picture below would be the input to the crossover, or speaker itself if it were a single driver without xover. Since the RL shelving circuit is impedance dependent, the flatter the speaker's impedance, the better the BSC will work. This can be seen above in my plot where the BSC isn't a perfect smooth shelf. At low frequencies the inductor acts as a short circuit, thus passing the lows untouched. At high frequencies the inductor approaches high impedance, so the resistor supplies constant attenuation to the highs.



It would be pretty easy to dial in exact BSC settings for a graphic EQ using the Jeff Bagby Response modeller, since you could match the baffle effect in the simulator. I think it's more proper for the BSC to be a part of the speaker than to use active filtering, though.
 

·
Premium Member
Joined
·
1,923 Posts
Discussion Starter · #7 · (Edited)
The midrange already has a zobel, the other two driver won't benefit from one. A zobel across the entire speaker (input impedance compensation) wouldn't do much, either.

Here is the total xover schematic: 2nd-order LPF/BPF/HPF, midrange zobel, and series tweeter attenuation with shelving circuit out front. Like I said before, this xover was created from raw data and ignoring baffle effects. Baffle step compensation was added after. The only way to get a flatter response form this setup using these drivers would be to redesign a new xover, and even then it might not get much better than the theoretical +/-2dB it is now. I consider this a finished product, maybe I'll start a thread about my 3-way floorstanders sometime.

 
1 - 7 of 7 Posts
Top