Thought I'd do a bit of a write-up. It's been a while since I posted here, and I think I've finally figured out some speakers that will keep me happy for a long time.
I've had these speakers for a few years now, and they've been pretty much unchanged for most of that time. When I first built them, they looked like this:
Which is a Seas H1252-08 midbass and a B&C DE250 on an 18Sound XT120 horn. An unlikely combination, really: those 8"s are intended for 3-way designs, so you can get away from the spectacular cone break-up:
What appealed to me about that combination, though, is that the cones and diaphragms were acting as true pistons over almost all of the audible range (IIRC, the DE250 has a little bit of breakup above 15kHz. My hearing isn't great up there, so I was happy to let that slide), which is something you don't see very often.
I made a very brief attempt to listen to the 8"s wide open, and immediately turned the music off. They're horrible if used this way.
It might be worth noting that the cabinets are the result of zero calculation in the Thiele-Small area. The depth was set by the HF driver, the width by the woofer, and the height by where I wanted the HF driver to be. Below 100Hz, what I hear is the room + speaker combination, and I always planned on EQing that to get where I wanted to be. A couple of dB either way from a different-sized sealed box is trivial compared to the 10-20dB of EQ that the room requires.
Simulation done after the build, just out of curiosity:
Measured at listening position:
I use EQ to bring down the 40Hz mountain and 70-80Hz bump, and the result is very even bass down to the lowest registers.
Getting these two drivers to work together wasn't trivial:
On the HF driver, we have:
- 3rd order highpass, which was a bit of trial and error to get something where two drivers lined up reasonably well. The only other thing that did that was a first-order highpass, which isn't steep enough for a compression driver running at the bottom of its range
- A notch filter, to attenuate around 3.5kHz. This is a fairly broad stroke of EQ, effectively flattening the tweeter's response out.
- An L-pad, which drops most of the 20dB of sensitivity difference between the two drivers
The woofer has had a similar amount of work:
- 3rd order lowpass
- "Bottomless" notch to kill the 4.5kHz breakup peak. With the electrical slope, the breakup peak was only about 20dB down. C2+L2 effectively present a short-circuit at that frequency, which leads to a lot more attenuation from the crossover, plus it gives the driver a lot of electrical damping at that frequency. I think the breakup is about 50dB down now.
- Finally, a Zobel filter. These woofers have fairly large voice coils with lots of inductance. Adding a Zobel meant the lowpass filter could do its job more effectively.
I kept them at that for a few years, which is a big deal for me: few speakers last longer than 6 months before I'm looking for the next thing.
I was happy with them until I took them to a mini-meet at Colin's, where I discovered floor bounce.
Floor bounce is the name given to a notch in the frequency response resulting from the sound hitting the floor (or ceiling) and arriving at our ears. When that reflection takes half a cycle longer to arrive, you get a dip in the frequency response from the acoustic cancellation. In a typical listening room, that dip occurs in the upper bass/lower-mid area, and is responsible for missing bass "punch". Speakers that do "punch" well typically have a large bass unit mounted close to the floor, so the floor bounce is actually pushed outside of the driver's operating area. It's worth noting that Colin's Edingdales, on account of having lots of drivers arrayed vertically, don't suffer much/at all from this effect.
Once I got home from Colin's, I could hear the problem with my speakers quite clearly. While they managed to go very deep into the bass, that upper bass "hit you in the chest" area wasn't what it ought to be.
Fortunately, the solution is pretty simple: add another woofer, separated vertically from the first. I could have taken a cue from Colin, with 4x woofers per side, but at £85 per driver, that would be pricey.
I opted for a 2.5-way crossover:
There was some extra output in the lower-midrange, and I decided to EQ that away, rather than re-working the midrange crossover. To arrive at the low-pass for the extra woofer, I kept an eye on the transfer functions of the crossover, and aimed to minimise the phase difference between the two drivers in the common operating area. The resulting values in the 4th order lowpass are staggered, which seems to have worked well. Since the crossover is steep and starts fairly low down, there's no need for a notch filter on the woofer. The Zobel did help, though.
I've spent a week listening, and the upper bass has definitely filled in. There's more punch/slam/whatever. The extra cone area means the demand on each driver is reduced, so distortion will probably be lower. I hadn't noticed much of a problem there, but it does mean I can crank things up even further if I really want.
Sound-wise, they tick all the boxes for me. Flat response, able to play loud without complaining, excellent LF extension. They'll play any music at any volume, and sound good (or as good as the track allows) doing it.
I think that's everything. I've enjoyed playing with these speakers, and now that I've got them to a state I'm happy with, I'll be looking at building a pair of nicer cabinets. Perhaps the crossovers will be mounted inside the box this time.
Dedicated to those large boxes at one end of the room
3 posts • Page 1 of 1
#2 Re: Speaker Update
Interesting developments Chris. Looking forward to hearing these again next your over. Talking of which I shall be in touch soon regards another mini meet my place when Nick has a firmed up delivery date for his new Acrylic baffles; expected early in October.
#3 Re: Speaker Update
Further update: 18Sound NSD1095N tweeters, and tweaked the crossover a little.
They're rather good.
They're rather good.
3 posts • Page 1 of 1