Speaker rebuild project - x-over questions

[Brian Steele]

Hey everyone, I'm looking for your feedback on the x-over design for a pair of loudspeakers that I'm rebuilding with new drivers in a refinished cabinet.

The woofer is the Dayton Audio DSA135. The tweeter is the Tang Band 28-847SE, a pair of which I purchased 13(!) years ago, but has been sitting in storage ever since.

Some notes:

1. The woofer's FR show a hump in its response that starts just below 1 kHz (baffle step?) and a high Q resonance around 7.65 kHz.
2. The woofer's THD shows a small hump at midrange frequencies that peaks at 2.5 kHz
3. The tweeter's FR shows the impact of baffle diffraction effects from 2.5 kHz to about 6 kHz, with the biggest dip at 3.5 kHz

I was initially trying for an acoustic x-over point around 2 kHz (to reduce the impact of the woofer's THD on the response), but I think I found something workable around 2.5 kHz that cleans up some of the issues caused by the diffraction effects on the tweeter's response. What do you think about the solution I've come up with? Any "gotchas" that I should be considering?

[Brian Steele]

...and the x-over details...

[Brian Steele]

Oh, I forgot to post the sim'd response of the proposed x-over....

[rpb]

I don't see any problems, but I have a suggestion. Let your ears be the judge though. On the tank notch at 8k, I'd hammer it hard rather than make it follow a target curve. So the R5 resistor could be dropped either entirely, or down to a couple ohms to keep it amplifier friendly. Take that peak down as low as you can.

[donradick]

Pretty much agree with rpb.
Take measurements with and without the trap to verify what it's doing.
If you drop R5 some, it will increase the Q of the trap.
that dip at 3.5K is troubling, suspect you have some cancellation going in, because the woofer has pretty good output there.
Play with R3 and L2 if and try to get the phase lined up, and eliminate that dip.
The center frequency of the woofer trap looks perfect.
https://leleivre.com/rf_resonance.html
Measure the tweeter by itself.

Other than the dip at 3.5K, FR looks pretty good.
When you get the phase lined up, the imaging improves a whole lot.

[Brian Steele]

The dip around 3.5 kHz (and the smaller dips above that) appear in the tweeter's raw response curve. I think they're diffraction effects. I've seem a similar effect with another build I did, with a narrower baffle that effectively moved the notch up to 4 kHz. I can reduce it by moving the x-over point up a bit more, but then that moves the x-over point further into the zone where the woofer's THD rises, which I want to avoid. Mission seems to have avoided this issue by using a small waveguide around the 1" tweeter in their 751 series of speakers, because even though the baffle is about the same width, there's no sign of the dip in their tweeter's response curve. I've been looking around for a tweeter that has a similar waveguide built in, but there's only a 2.6" mounting diameter to play with. I'd bet that the dip likely won't be noticeable anyway.

R5 affects both the "Q" at the notch frequency and the slope, which in turn affects the phase tracking at the x-over frequency. 20 ohms seems to be the best compromise between the two, but I am going to play with different values for R5 to hear what the results sound like.

Funnily enough, my main concern is that "huge" inductor L1. 2.3 mH seems kind of big. I'm trying to figure out if there's a way to get the same or similar response curve on the woofer without having to use such a large inductor, which will need to be iron-core in this case.


Thanks for the feedback everyone.

[rpb]

Most 8 ohm speakers use about 2.5mH if full BSC. If you add a notch centered around 1k, the coil will be smaller. Try in your sim, a 1.5mH, 50uf, and 6 ohm notch. I'm going from memory, so play with the values. It will flatten the 1k range, and allow a higher xo, and smaller coil, about half as big, or smaller.

[Chris Roemer]

If you've measured them with the tweeter down, w/them on the desktop (like your pic), you might be getting some bounce cancellation (off the desktop edge).
Measure them once upside down (from the way they look now).

A Zobel can lower your coil value (maybe).
I came up w/this -
2nd order HP w/"L" : 12uF series cap, 0.50mH shunt coil, L-pad: SR=4n(ohms) / PR=4n
2nd order LP w/Zobel : 1.5mH (low DCR) series coil, 20uF shunt cap, Zobel = 8n+10uF

My Fc is closer to 2.0k than 2.5k, and although my BSC is about the same as yours (total), my woofer rolloff shoulder is shallower.
I've got the same diffraction "dip" from 3-4k that you've got on a similar project using a compact dome on about a 5" wide baffle. If I'd have offset the tweeter only 1/2" to the side, that dip could have been a lot less.

[Brian Steele]

A minor update to the x-over, using some parts that I have on hand...

[Brian Steele]

Most 8 ohm speakers use about 2.5mH if full BSC. If you add a notch centered around 1k, the coil will be smaller. Try in your sim, a 1.5mH, 50uf, and 6 ohm notch. I'm going from memory, so play with the values. It will flatten the 1k range, and allow a higher xo, and smaller coil, about half as big, or smaller.

I tried that. Looks like I had to drop C to 30 uF to flatten the woofer response. I adjusted the filter on the tweeter to compensate. This has resulted in a shallow dip around 2 kHz as shown, even with good phase tracking through the x-over. I'd come across this same issue with an earlier iteration of my x-over.

[Brian Steele]

If you've measured them with the tweeter down, w/them on the desktop (like your pic), you might be getting some bounce cancellation (off the desktop edge).
Measure them once upside down (from the way they look now).

A Zobel can lower your coil value (maybe).
I came up w/this -
2nd order HP w/"L" : 12uF series cap, 0.50mH shunt coil, L-pad: SR=4n(ohms) / PR=4n
2nd order LP w/Zobel : 1.5mH (low DCR) series coil, 20uF shunt cap, Zobel = 8n+10uF

My Fc is closer to 2.0k than 2.5k, and although my BSC is about the same as yours (total), my woofer rolloff shoulder is shallower.
I've got the same diffraction "dip" from 3-4k that you've got on a similar project using a compact dome on about a 5" wide baffle. If I'd have offset the tweeter only 1/2" to the side, that dip could have been a lot less.

Thanks, I used that information, and fiddled around with it a bit, and came up with this. Looks interesting, with a nice low acoustic x-over point. Only issue I see is a lot of the tweeter diffraction issues end up remaining in the response curve. The second cap (C4) in the tweeter circuit seems to be required to have proper phase tracking through the x-over.

[rpb]

Hey, give us some files to play with.

[Brian Steele]

Here ya go: http://www.diysubwoofers.org/project...tech2/pita.zip

[rpb]

Here ya go: http://www.diysubwoofers.org/project...tech2/pita.zip

Where's the files? ?????

[Brian Steele]

Should be available now, sorry. For some reason the link is case-sensitive for zip files.