How to control the upper peak(s) on the Dayton RS125?

[fjhuerta]

Hi!

I'm now building a new speaker, based on dual RS125's and an RS28.

Since the cabinet is so small, I'll need to simplify crossover design, or (gasp!) use electrolytics in every shunt branch.

Considering this... http://www.partsexpress.com/pdf/295-360g.pdf

I have two peaks: one at 7,500 Hz, and the other one at 12,000 Hz (I already took in cabinet measurements.. hopefully, I'll be able to post pics of the cabinets and measurements tomorrow. They did come out nice!).

So far, I've tried two approaches.

1) Second order electrical with a series RLC. No matter how hard I try, I can't seem to really get rid of the peaks. Not even if I center the RLC's frequency response between the peaks. Or if I don't use an R at all. It does help, but not as much as I'd expect. Peaks are at -30 dB or so.

2) Fourth order electrical. I get rid of every peak, and FR is nice and smooth. But I then have to invert the tweeter's polarity in order to get the speaker to sum correctly.

I'm using a 1,700 Hz crossover point. In both cases, phase is not an issue (when I invert the tweeter's polarity in example 2, that is!). I'm not sure I'm doing the right thing, though.

Any ideas?

[Dave Bullet]

Removing the R should really suck out any breakup peak - assuming you have the L and C correct. What values are you using? Not sure with duals but around 8uF and 0.05mH (yes - only 50uH) should be in the range with a single driver...

PS: How did your last speakers turn out? (I can't find your original thread)

[dawaro]

It might be worth your time to take a look at Roman's Microbe MTM design. It does not use the same tweeter you are looking at but you can see what he did to control the peaks.

[mpotoka]

It might be worth your time to take a look at Roman's Microbe MTM design. It does not use the same tweeter you are looking at but you can see what he did to control the peaks.

Link Broken...

[cjd]

2) Fourth order electrical. I get rid of every peak, and FR is nice and smooth. But I then have to invert the tweeter's polarity in order to get the speaker to sum correctly.

I'm using a 1,700 Hz crossover point.

This sounds more than adequate to me. In fact, the RS125 should be up to a higher crossover slope without problem, 4th order.

The next option is a "C-E" style filter to push the breakup down even more.

[Brian Bunge]

On the Microbe TM design Roman uses a single series inductor with a notch filter on the RS125.

[dlneubec]

FYI, here is Roman's Microbe SE design.

[Undefinition]

I followed Roman's lead and tried a series LCR in parallel with the driver. I forget the values, but that in conjunction with a 2nd order electrical brought things up to a nice pretty 3BW slope around 3K. The peak(s) was 30-40 dB down. I'll see if I can dig up the XO values later on.

[fjhuerta]

Thanks!

My last speakers turned out really great. The Neo drivers are really hard to work with, but pairing the 8 and 3 brought great results. Now I have something resembling flat frequency response

I'll attach the graphics so that what I'm saying makes some sense. 6th order accoustical sounds like the best option right now.

[fjhuerta]

Here's the fourth order design. As much as I tried to optimize the LCR values, this is what I got. At -40 dB, I may not have many issues with the break-up point, although I'll try to move the crossover point up to 2KHz. Purple trace is the 4th order x-over; green trace is the 6th order crossover, included for comparison.

[fjhuerta]

Here's the 6th order crossover. No break-up nodes to be seen. Total number of components is the same. I also have to add - the inductor's R is accounted for in this model. I'd also like to use a 2 KHz x-over frequency on this one.

[fjhuerta]

I just noticed one thing... the reverse null is -17 dB on the 4th order, -10 dB on the 6th order.

Apparently, 4th order is looking a bit better. I just wonder if I did the LCR right.

[augerpro]

How does it look off axis?

[max_andrews]

Try using an lpad on the tweeter instead of a single resistor, this will yield a more uniform attenuation, and may improve your phase alignment.
If not, try using a third order electrical on the woofer (remove the last capacitor). You'll need to change the values of the smaller inductor and the remaining capacitor, but this will relax the slope on the woofer a bit and should improve your phase alignment. You may need to reverse the tweeter polarity in this case.

[fjhuerta]

How about a -40 dB null? It seems 1.8 KHz 4th order L-R may be my best choice.






Also, here's the response at 15° off axis (using measurements, not the LSPCad prediction). Since I know the RS-28 is quite a polite tweeter, I aimed for a flat FR at this point.

[fjhuerta]

OK, at this point, I think I can start thinking about building a crossover. I fine tuned by hand the crossover design, and arrived at a combination that yields practically flat response at 15°.

I'm still curious - how do you control a dual-peak woofer, like the RS-125? Even after doing an LC circuit, I couldn't get rid of all the nasties. Do you have to use dual LC circuits? I think that's a bit overkill....

[mmoeller]

I'm not a speaker expert, but I would assume that you would need two separate notch(LC) filters tuned to the right freq. I think I've seen Jon Marsh do that on some designs. Lot's of components.

[Kevin Haskins]

Your 4th order solution looks fine. I wouldn't choose a network based on the reverse null.

You can build one wide band notch to attenuate the entire region rather than two single notches. If you have it 40db down I wouldn't worry about it.