A good tweeter to match 5" SB Acoustics Magnesium Woofers

[DDF]

Btw, super powerful sprinkling resistors and caps around and cheap. C across l creates zero, usually bigger c vs l for same product of the two increases Q, increase inductor L or add resistor to make zero less deep (the two methods don’t produce identical results which is a problem with pcd). Just watch out for what RC shunts do in the ultrasonics At a concert now, string quartet, no better practice for voicing a speaker. Btw is your new avatar from an old Gories album? You mentioned working in Detroit, saw many a show at the Greystone, Asylum and Saint Andrews Hall and amazing record stores up near ten mile back in the old days

[augerpro]

I'm somewhat interested in the new 6" ceramic version SB17CAC-08 and am wondering about a good match. Maybe Hiquphon's?

I think if you were looking at the CAC, I would pony up the couple extra bucks for the Rohacell cone. Otherwise, in most of the CAC line I'm just not seeing what the advantage is over the NAC for the extra cost. No ideas on tweeters, lots of really nice and affordable ones out there.

[DDF]

Is the ncore recall related to caps? Sent from my iPhone using Tapatalk

[xmax]

The pads I design use a cap in the shunt, it's great for a good safe transfer function, but these guys can't seems to get that far.
Hypex has a Ncore recall, its a huge mess for my buisness and Hypex, I'm sure they will get through it but they are not even
answering emails right now. It's a shame because the amps are so advanced and sound so much better than
what I thought Class D could sound! I have almost given up on this forums lack of natural filter understanding
other than your posts. Thanks.

[DDF]

Autocorrect madness Ncore, not neorealism Sent from my iPhone using Tapatalk

[DDF]

Glad to help. If you get a chance, part 2 and3 describe simpler filters and set tweeter roll off lower. My experience matches your in that I think it’s easy to stress a tweeter trying to fill in the power gap. Cleaner and more powerful the amp (neorealism!), easier it is to hear this Sent from my iPhone using Tapatalk

[xmax]

You guys made me use windows and learn Xsim to confirm what I am referring to, without the 10 ohm resistor "bypassing" the 23 uF cap,
it is very similar to not having the 2 split coils/caps. Great design and thank you for clarifying the friendly transfer function on the tweeter.

(If the resistor was not required in the transfer function, it could have basically been done with 2 caps and coils instead of 4,
a good study in filter design)

[dlr]

Hi Dave, I found using the trap as a separate element only for bump EQ and not for roll off will often (but not always) add an inductor.

You're right, it often will. That's when I let if float and fix other elements in multiple attempts to find the best (read minimum parts count) end result. Bumping one element one way or another sets up the optimizer for different results, so it's hit-n-miss until you hit the right combo. Either way gets you there.

I've had Calsod since '89 but stopped using its optimizer ~ 15 years ago, and it entirely quite a while ago. I feel I can make higher quality decisions about on vs off axis tradeoffs than the weighted functions in an optimizer.

I stopped using it as my main SW long ago as well, started using Jeff's PCD at the beginning satge, then most often finished in SoundEasy. The latter, however, is frustrating. I've never taken to the GUI.

An optimizer doesn't "know" when a bump is diffraction and I don't eq diffraction flat at just one point in space.

True as well, that's why I never design any system without adequate diffraction control (roundovers and felt). Then you can almost dispense with consideration of the off-axis. Were I to design a system and not use it, I would use it temporarily to get good on-axis that would leave the off-axis problems to be almost exclusively driver directionality issues. The last thing would be to finish without any felt in place. It would then created on-axis issues, but at that point you know that the only thing at issue is the diffraction influence.

Xsim also makes any topology possible, replicating the CAD feature of Calsod not supported by PCD.

I really wanted to be able to code it this way, but WinPCD is done all open source and I can't code that sort of GUI on my own. I decided that trying to closely emulate the PCD was the best option to make it easy to move over to WinPCD.

BTW while we're talking traps (s-plane zeros), have you seen Thiele's "Loudspeaker Crossovers with Notched Responses" AES paper? He calculates target functions for high and low pass notch filters that are symmetrical, sum to all pass and degenerate to Butterworth (odd order) or LR (even order) when the notches are pushed well past fc. You might find it useful to help set targets for your optimizer, even when using UE as the xover.

I probably need to totally re-write the code that does the calculations for this, it's not efficient, so it won't happen soon.

dlr

[DDF]

I like traps, I used them, but my work flow in that case is to first model a trap only to linearize that spot in the raw driver response, then I work on the crossover proper.
dlr

Hi Dave, I found using the trap as a separate element only for bump EQ and not for roll off will often (but not always) add an inductor. I do it the opposite way, design the roll off first draft, then start adding shunt parts to get rid of peaks that i know are in the driver (and not just diffraction at one point).

I've had Calsod since '89 but stopped using its optimizer ~ 15 years ago, and it entirely quite a while ago. I feel I can make higher quality decisions about on vs off axis tradeoffs than the weighted functions in an optimizer. An optimizer doesn't "know" when a bump is diffraction and I don't eq diffraction flat at just one point in space.

Xsim also makes any topology possible, replicating the CAD feature of Calsod not supported by PCD.

I feel I get better results this way.

BTW while we're talking traps (s-plane zeros), have you seen Thiele's "Loudspeaker Crossovers with Notched Responses" AES paper? He calculates target functions for high and low pass notch filters that are symmetrical, sum to all pass and degenerate to Butterworth (odd order) or LR (even order) when the notches are pushed well past fc. You might find it useful to help set targets for your optimizer, even when using UE as the xover.

[Waboo]

So kendomusic (OP), You seem to have maybe given up, and that's understandable. With all the needless bickering and bs name calling from some members which has created a complete derailing of your well intended thread, I have to ask; did you decide on what tweeter you might want to use.

I'm somewhat interested in the new 6" ceramic version SB17CAC-08 and am wondering about a good match. Maybe Hiquphon's?

[xmax]

1 The resistor is there for that very specific reason, 2 The values of the high pass are drastically different
and so is the transfer function hence the level being very low at Fs. I get it, do you?

[dlr]

Interesting it seems the resistor is the only thing keeping both the chokes and caps "split"

This comment is clear evidence that you really don't understand crossover filters. Even though DDF fully described the function of each.

[Clearly the difference with your pad/transfer resistor is the higher frequency third order filter...Distortion generator, that's funny.

And this one as well, a repeat of earlier nonsense.

dlr

[xmax]

I am too a believer in "broad overlap" I think it can make the 2 drivers sound more cohesive.
That said some designers prefer extra steep filters with hardly any overlap and this approach
can also sound good...

[xmax]

These inductors actually had a third tap...

[dlr]

If the peak is relatively stable in frequency over angle, it's handy to notch it out by splitting the main low pass inductor in two and adding a small cap across one. Still allows broad overlap and well controlled off axis.

I never thought of it as splitting an inductor. Separate traps were avoided in this instance largely because I was trying to keep the crossover cost to a minimum, although the optimization process can result in "traps" as part of the final crossover due to transfer function needs.

I like traps, I used them, but my work flow in that case is to first model a trap only to linearize that spot in the raw driver response, then I work on the crossover proper. Using the optimizer in CALSOD or SoundEasy I then usually allow the trap to float. Sometimes they eliminate an element or two, sometimes an entire trap. So in the end I'm not creating any particular part of the crossover and leaving it for some specific function. Traps usually never remain unchanged and ratios of components are meaningless other than what the optimization process results in. Jeff's PCD and my WinPCD both provide the option of initializing a filter section for a target type, electrically, as a starting point. Those are the ideal "ratios" as defined by the target function, but they will always vary widely when optimizing, as you know, due to the driver response, electrical and acoustical. Only the acoustic output matters in the end. All of which you know.

dlr