After much tinkering, my Natalies sound great,why didnt the stock x-over work for me?

[fjhuerta]

I'm absolutely confused here.

After much modeling, measuring, modeling, building, listening, and tinkering, I simply love my Natalies.

But it doesn't make any sense.

So many people have built the stock crossover and loved them.

I followed the plans. The baffle dimensions are identical (although they are towers). The only thing I changed was the box depth, and I don't think that should affect sound so much.

Basically, I had to change 5 values to get them sounding right: L3 = 0.160mH, L2=1.5mH, R3=1 Ohm, L4=0.300mH, C10=30uF.

The main issue I had was with the tweeter. I had a very pronounced "bump" at about 2 KHz, and the tweeter was far too high in level. I had to bring it down 2 dB or so. Actually, the tweeter response curve was so far off from the spec, I wonder if I had a defective pair, or the batch I received was completely different from the ones that were used to make the design. I have never seen any driver being so different from the published specs - this was way beyond wrong.

Now, after measuring them, they are about +-2dB from 200 Hz to about 14KHz (IIRC).

But I don't get it, and I'd love to learn whether I did something right or wrong while building them. All I changed was, as I said, the depth of the box. Everything else is identical. The crossovers use 15GA inductors... I can't imagine what I did wrong.

But... darn, the Daytons sound beautiful. For the lack of a better word, the instruments' "textures" sound so... right with them. They are really incredibly detailed.

[Brian Walter]

Javier,

The Dayton RS28 has been reported to have had quality control issues, so you may have been received a pair of drivers that didn't meet spec. It would be interesting to see or hear how they sound and measure with replacement tweeters. That would certainly tell you a lot.

Brian Walter

[kingpin]

How long ago did you buy the RS28.
I don't know when they started, but the ones that I received in Aug have a quality control marking on them. They have to read between a certain spec or they are not sold.
The rating is on the side of the tweeter like this.

Image not available

[soho54]

I have two sets I recieved in February 06 that have this sticker on them. All the checks were in the same spot. In the middle somewhere, like -.2 or something I think.

I ran sweeps on them and they matched the posted ones.

[fjhuerta]

Hi all!

I got the tweeters around August or September, I can't remember. They do have the stickers, and they match on both units.

I attached a plot of the tweeter. That hump is terrible, and I can't help but wonder if it's a result of me not rounding the baffle.

Strangely enough, both tweeters measure the same, the finished speakers measure the same, and whenever I take measurements of PE speakers, my plots are identical to theirs. So weird... I just hope to learn what went a bit wrong this time.

I'm also attaching the plot of the finished speaker, with the new values, and the simulated frequency response graph, with the measured response graph at 5ms and 1 meter. It all adds up, apparently... which, as I said before, is too strange.

[Dennis H]

That hump is terrible, and I can't help but wonder if it's a result of me not rounding the baffle.

That's what it looks like to me. Dennis Murphy compared rounded and unrounded baffles. His baffle was much smaller (5.5" MT) so he had a peak at 4K and dip at 7K without rounding but the principle is the same, just lower frequencies with the bigger baffle. You can model it in Baffle Diffraction Simulator from the FRD group.

http://murphyblaster.com/content.php?f=cabinets.html

[fjhuerta]

That's what it looks like to me. Dennis Murphy compared rounded and unrounded baffles. His baffle was much smaller (5.5" MT) so he had a peak at 4K and dip at 7K without rounding but the principle is the same, just lower frequencies with the bigger baffle. You can model it in Baffle Diffraction Simulator from the FRD group.

http://murphyblaster.com/content.php?f=cabinets.html

:T :E

I also thought not rounding the baffle would not cause too many issues - but those graphs surely remind me of the ones I got when I measured my speakers with the original crossovers.

Thank you very much for the link. At the very least, this should be a warning for all not to build the reference designs without using the recommended boxes, too (or at least replicating the rounding).

I modeled the diffraction on the KHF tool in LSPCad after reading your post, and indeed, I got a hump at 2KHz. Not as pronounced as the one I measured (perhaps off by 1dB), but it's close enough for me... I'm pretty sure by now diffraction is the culprit here.

Thanks a lot! I'm beginning to understand what's going on

[Jed]

Try measuring off axis too. That diffraction hump might actually disappear or not be too much of a concern 15 degrees off axis.