XO help to shelf rising response of metal dome tweeter response above 10k?

[alecf]

Hi, I'm simulating a high pass filter in WinPCD, but need some help with this Titanium dome tweeter.
Can someone suggest how to flatten its response above 10K?

This is my circuit so far. Thanks!

[PWR RYD]

Try an RC network (resistor in series with a capacitor) after the HP filter. It would be in parallel with the tweeter (like a Zobel). Maybe a 4 uF and 4 ohm as a starting point.

[alecf]

Thanks, that works well. I also found that a "Series RL After Crossover" did a similar flattening. Is one solution preferred?

[rpb]

Make a notch centered near 20k. ie, RLC in a parallel leg. Start with 1 ohm till you can see the notch location clearly. Then raise the resistance.

[Wolf]

I would think the CR/zobel will be the lest expensive method in this case, but whichever one works best should be preferred.
Wolf

[alecf]

Thanks, all. That makes sense, Wolf.

[LOUT]

It might be important to keep an eye on the impedance above 20khz (depending on the amplifier) when using an RC-to-ground parallel with the tweeter, because that RC will probably lower impedance above 20khz.
If the tweeter and/or XO already has impedance high enough to compensate up there (so it doesn't drop below what your amp can handle), or if your amplifier doesn't mind low impendace above 20khz, then this shouldn't matter at all.
Even if it does happen to be a problem, I'm guessing the tiny L (for an L+R series alternative) is only ~$3.

I think increasing the R in the CR to a large enough size to avoid a problem (IF there is a problem) might quickly become too large where the C isn't allowed to do the job you want it to...which is why I'd lean more toward the RL instead of increasing the R with the RC if impedance happens to be a problem.

Sorry to jump in so late when you're trying to decide. I'll also warn that I'm pretty new to all this, so disregard what I'm suggesting if someone more experienced (aka basically anyone else on this forum) says different.

[alecf]

That's an interesting insight. Here are the different impedance affects. Top is RL, bottom is RC for the same kind of top-end flattening - quite a difference

[PWR RYD]

I don't see a problem with either of those impedances.

[LOUT]

Is your HighPass XO (before any RC or RL business) something like this?

Sorry, I'm having an embarrasingly difficult time figuring out how to read that particular software in the first post...not your fault nor the software's, it's just very different from what I'm used to looking at.

Can you share what RC and RL values you're using for the additional high-frequency taming (not shown in the little XO above) as well?


I agree, it doesn't look like a problem. I'm just trying to SIM along at home and I'm having trouble figuring out how to get the same impedance results.

[alecf]

Yes, that's more or less it. Here's the full circuit values and schematic from the software

[LOUT]

Thanks for taking the time to clear that up for me. It looks like your 5.9ohm resistor is both making sure impedance doesn't drop to ~<3 by 40khz+, and its resistance is needed for the cap to bleed highs the way you want in the first place...so win/win.

EDIT:
Might be a dumb suggestion after you've done all that fine tuning, but I think this might give similar results while using more commonly available part values and fewer parts....though it does diverge by up to 1.5-2db between the 1.2khz FS and 2khz, and rolls off a little steeper below 1khz.

The DCR of the inductor doesn't seem to affect the response much, so I think an air-core or solid-core could work if you have a preference.

[alecf]

That's very impressive, LOUT, thanks for tinkering - seems like you are an expert! I'm a novice. I think your circuit looks really good and is much simpler.

Here's the response it gives



I didn't know one can use a parallel cap like that on its own - how did you come about with this topology? And what software do you personally like to use for design?

[LOUT]

More like, I got lucky between the little bit I know and brute force.

I think the parallel cap I used is doing the same thing as your far right Cap+resistor...just without the resistor in-line with it. That makes mine give a continual downward curve while I think yours still curves down but eventually straightens out (not that it matters much either way when the difference is happening above 15-20khz and is really small). I wasn't worried about adding another resistor for mine because everything that gets to that cap already has to go through the earlier 16ohm resistor anyway, so that 16ohm R makes sure the impedance up high stays plenty high for a 4/8/16ohm amp.

I used VituixCAD free software. It has an SPL and Impedance tracing option built in, so I copy/pasted the no-XO impedance you shared to use with it. I input your XO, then setup another copy to mess around with while trying to get the copy to align as closely with your result as possible...which was made a lot easier with both lines sharing the same graph/space.
Vituix also makes it pretty easy to copy/paste images of the XO (and anything else) to upload to websites. Which only took me an entire year to realize because I'm a dummy who had literally take pictures of the screen with a phone before, lol.

[DeZZar]

I assume at some point a woofer/midrange will be introduced into this XO?

Just bare in mind that this network will likely go straight out the window as soon as you start to combine the response with another driver. It's really not possible to work up a network on the drivers individually. It's just a nice theoretical exercise - the network for both together is more than likely to be dramatically different.