Some XO advice

[gbyleveldt]

Hi Guys,

I've been a long time lurker, absorbing all the info you guys have to share.

Anyway, I've been tinkering with the XO for my first speaker project and thought I would post it here to get some of your input/suggestions as to the results. Please feel free to tear it apart, any input would be greatly appreciated :B :B

In a nutshell, it is a little TM that I'm planning on using for the surrounds in my HT setup. Once I'm happy with these I'll tackle the fronts for the system. The woofers are Seas CA18RLY and the tweets are 27TDFC's. I've gone for an electrical LR2 XO and a "think" I got a pretty reasonable LR4 acoustic response. The XO point is around 2k. I did not add too much BSC to the circuit, as these are meant to be the surrounds and I want to keep the sensitivity up as the mains will be quite sensitive (2.5way planned). The tweeter is mounted offset from the woofer, but both are flush mounted on the baffle.

First Pic - XO with values
Second Pic - The SPL, Phase and impedance plots
Third Pic - Polar response at (Brown) 1.5k, (Green) 1.75k, (Purple) 1.9k, (Blue) 2.1k, (Pink) 2.4k
Fourth Pic - Off Axis at (Brown) 0deg, (Green) 15deg, (Purple) 30deg, (Blue) 45deg

The reason for starting this thread is about my concern about the phase response having a little burp just after the XO region. It makes me think that the phase integration hasn't come off as good as I had hoped.

Any thoughts?

Thanks so much!!! :B :B

Cheers

Gert

[TacoD]

Firstly your first filters already exhibits more skill than most of us start with (including myself ). Great job. The power response (lobing of the filter) also looks oke to me. Do not give the baffle an angle I would say.

I think 2 mH gives a bit too much BS, I would try 1.8 mH. (this is my experience with the CA18RNX). I would also add a zobel network for the CA18, something like 10 uF and 10 ohm will do.

The LCR in the tweeter network, is this for killing the resonance freq. of the tweeter?? If so, it would suppress some harshness which is present in these Seas tweeters.

[augerpro]

Looks good to me, but I'm pretty new at this too so take that for what it's worth

A couple questions 1) The HBT for the tweeter looks like you started pretty low. Nothing wrong with that as long as the HBT phase matches the measured phase. At least under 20k so the phase around the XO is correct.

2) Did you measure teh drivers from one point in space, i.e. did you leave teh mic in the same position for all your measurements? If you did you can't use the polar plot function.

3) Reverse the polarity of the tweeter in the CAD screen and plot that to see the phase integration.

[gbyleveldt]

Geesh Guys, Thanks for all your kind words!!! :B :B :B Much appreciated ;x(

@Taco:
I've actually built the boxes already without sloping baffles, so it's a little late now It seemed a little ambitious for a first project anyway. The 2mH inductor was actually a 1.8mH inductor, but I changed it so that I could use standard components. I guess it won't be a problem taking off a few turns to return it to 1.8mH. It'll shift the XO point back to 2k instead of 1.95k. I'm a little confused at your suggestion of the zobel network though (I'm still a little new to this, so please bear with me ) I'm assuming you need it to tame some of the raggedness around the 900Hz region of the CA18RNX? Well, that raggedness is not there so much on the CA18RLY (I used the Seas spec sheets for comparison), or maybe I have the use of a zobel quite wrong? The LCR on the tweeter is to tame the 12k hump in the 27TDFC's response. I've attached the XO transfer function in the first pic to illustrate the result.

@augerpro:
I spent quite a bit of time trying to match up the HBT phase with the measured phase on all the drivers. Again, this is all pretty new to me so the results aren't guaranteed :B I did quite a bit of reading WRT the position of the mic when measuring and finally decided to leave the mic aimed at a central position of the baffle for the far field measurement of both drivers. My (ill) logic dictated to me that this should give the best result for the least amount of effort. I understand that there are trade offs doing it this way but I felt comfortable doing it this way with the amount of knowledge I have now. I've attached the reverse null in the second pic for your perusal. That dip is around 20dB, not great, but I'm assuming that is one of the trade offs with mounting the tweeter offset and on a flat baffle. In designing the XO, I made sure to include all relative acoustic offsets (in SE), so I'm hoping I should be ok.

As a side note, I've been using the Digital Filter function is Sound Easy to verify all my results and I'm almost happy with it. I still need to learn how to associate what I hear with what I see on the graphs.

Again, thanks for all your help ;x( ;x(

[cjd]

A few thouhts:

1.8mH is a standard value...

Don't worry about the top end response on the tweeter.

Phase integration can improve - start worrying a little less about target crossover frequency. But, it's also not bad at all, so not something specifically to worry about.

The potential issues that I see remaining revolve around supression of the distortion inherent in this mid-woofer. What happens if you push your target to 1800Hz?

Sometimes, that third component in the crossover can make a HUGE difference - and you don't need to do the same on both. Mid could be 2, tweet 3. Or the reverse.

Do you have a graph that shows the phase for *each* driver (same graph, as that shows how well they track and over what range) That blip suggests you're using minimum phase perhaps, which requires offset and the like. It does not look like an artifact of the crossover specifically, but of the frequency response. In which case it's just what it is. Phase is inherently a derivative of frequency response.

C

[TacoD]

Firstly your first filters already exhibits more skill than most of us start with (including myself ). Great job. The power response (lobing of the filter) also looks oke to me. Do not give the baffle an angle I would say.

I think 2 mH gives a bit too much BS, I would try 1.8 mH. (this is my experience with the CA18RNX). I would also add a zobel network for the CA18, something like 10 uF and 10 ohm will do.

The LCR in the tweeter network, is this for killing the resonance freq. of the tweeter?? If so, it would suppress some harshness which is present in these Seas tweeters.

A straight baffle is oke .

Drop the LCR on the tweeter, I doubt it sounds better (those components also have their signature).

The zobel is to suppress the raising impedance, this ensures that the other filter parts work more effective.

[gbyleveldt]

Thank for all the excellent ideas, Guys!

I'll be wipping open my text books again and fiddle with the XO and report back what I find. I'll also shift the XO point a little to see how things improve.

Cheers

Gert