DiscoCAT Speaker Design

[S-Man]

Ah yes. Very jagged!
Shame there are no measurements of its affect though.

 

[a.palfreyman]

Have you measured with grille frames off to confirm that they are the culprit?

 

[S-Man]

Have you measured with grille frames off to confirm that they are the culprit?

Yes, it is apparent in the pics in post #11. In the tweeter plots in particular the 5.6K and 11.2KHz reinforcement is visible, with some cancellation above and below 5.6KHz.
The bass/mid also suffers.
The problem is not just the frame, even without it the box edges cause trouble (despite the 1/2" roundover). I was hoping that I could spread the diffraction of the frame around to some extent and end up with a better compromise than without a frame at all. I may need to try some jagged shenanigans to achieve this.

 

[a.palfreyman]

I'm surprised you are getting same from midbass...unless it's from the foam / cone edge which appears to be about 120 diam on this driver.

 

[S-Man]

The effect on the midbass is less dramatic and would appear to be more of the cancellation effect.

 

[S-Man]

On to the 4th XO I have built and the one currently in use:

This version has a separate series filter to effect the BSC (R$ L2 C5). The overall FR is quite similar to the previous one.
It also (deliberately) reduces the XO point to about 2.5KHz... in the hope of providing better off-axis response.
On the downside, the phase tracking is less ideal than the 3rd XO.

I've been running with this XO for a few weeks now. 1st impression was that it sounded very similar tonally to V3, but somehow it just sounded more "right". I still have V3 built up and ought to revert to it to compare to V4, however I am enjoying V4 so much I don't feel that motivated to change.

The higher resolution of the DCs versus the CAOW1s has resulted in me to have a complete reappraisal of my system. I have ended up preferring different power amps, reverting to my op-amp active crossover and different interconnect cables.

Overall, I'm rather pleased with this project!

 

[James]

I'd be bothered by the 5.6kHz peak. Have you considered LR4 acoustic instead of LR2?

One of my design philosophies is to use as few LCR components as I can get away with. You may be able to reduce filter circuit complexity and improve integration with a smaller overlap - assuming you can nail the relative phase.

Oh, how I miss designing and building loudspeakers ...

 

[S-Man]

Hi James, yes I am cautious of the 5.6K peak, but not sure how audible it is. I really ought to measure horizontally off axis at the angle I actually listen at - the speaker backs are parallel to the front wall.

If you would like to suggest an LR4 XO, I will happily send you the FRD and ZMA files of the drivers in the box. Drop me a pm if you want them. I won’t guarantee I will build them, but it would certainly be interesting to see how an experienced designer would do them.

After AP’s posts I’m think of making some sort of shaped mdf inserts to try to spread/reduce the peak. That peak is not in the nearfield measurements so it’s almost certainly a box/frame edge effect.

 

[a.palfreyman]

That peak is not in the nearfield measurements so it’s almost certainly a box/frame edge effect.

That was my thought too.

 

[James]

If you would like to suggest an LR4 XO, I will happily send you the FRD and ZMA files of the drivers in the box. Drop me a pm if you want them. I won’t guarantee I will build them, but it would certainly be interesting to see how an experienced designer would do them.

I would dearly love to, but sadly one of the reasons I hung up my copy of LspCAD is it won't run on Mac OS, which I migrated to over 10 years ago.

After AP’s posts I’m think of making some sort of shaped mdf inserts to try to spread/reduce the peak. That peak is not in the nearfield measurements so it’s almost certainly a box/frame edge effect.

Effect of grille frame and grille material (with some felt diffraction control efforts) on the 15W:
Green = no grille

You don't listen nearfield, so the baffle effects need to be taken into account. That is also why, as I'm sure you know, in-situ FRD measurements are necessary. My suggest of LR4, particularly if crossed closer to 2kHz will largely ameliorate the 5.6kHz peak without the need for a notch filter. I can't quite see whether the tweeter is similarly affected by the baffle or grille, so do take my recommendation with a wee pinch of salt.

As you already have the FRD and ZMA files, it shouldn't be too much trouble to sim an LR4 network.

 

[S-Man]

Shame about the OS.

Most of the 5.6K bump is caused by the tweeter reflection off the grille frame and possible also the cabinet edge. I can’t see how an LR4 slope (on the 15W) will help all that much.

I think I’m going to try to smooth the baffle/frame somehow.

 

[James]

Most of the 5.6K bump is caused by the tweeter reflection off the grille frame and possible also the cabinet edge. I can’t see how an LR4 slope (on the 15W) will help all that much.

You're right, it won't. I didn't see it earlier, but your last response graph make is as clear as day. As you were, then.

 

[chiily]

Oh, how I miss designing and building loudspeakers ...

I bet you can still smell the sawdust; go on then, go on, you know you want to James....

 

[James]

I bet you can still smell the sawdust; go on then, go on, you know you want to James....

I can make sawdust anytime, and in fact I do. My wife has no shortage of ideas on what I can do with my stash of leftover wood. What I miss is the thrill of chasing a particular design objective and going through various iterations in my mind on how it can be achieved. Then there is the fun of scouring for the right complement of drivers to suit, and that's well before sawdust is made.

I'm now reminded I have an incomplete second PFM-Special design languishing in a cupboard. All I'm missing is a pair of tweeters that I borrowed for the E-IX. I really should finish it and see if the design (2-way with sealed Seas CA-22RNX 8" mid-woofer) is worth publishing. The drivers are still available.

 

[Mike Hanson]

I'm still pushing to finish my PFM Specials (initial model) this fall, and I would be sorely tempted to build the second iteration if you finished it.

 

[S-Man]

I have reverted to the 3rd XO. I don’t think it’s as pleasing as V4 .

However, I have hit a snag, there is just a touch of over-emphasis in the upper midrange. This manifests as a slight exaggeration when vocalists give it a bit of welly. The CAOW1s don’t do this! (LS50s do, and probably worse than the DCs).

I’m not sure if it’s a small peak in the FR, or if it’s a resonance or time-domain issue. More measuring and tweaking needed!

 

[James]

I'm still pushing to finish my PFM Specials (initial model) this fall, and I would be sorely tempted to build the second iteration if you finished it.

The good news is that the PFM-S2 is a far simpler, traditional floor standing box design. If anyone has a surplus pair of Morel CAT378 tweeters, I can swap you some cash for them. @orangeart, do you have any in your E-IX BOM inventory?

 

[a.palfreyman]

This manifests as a slight exaggeration when vocalists give it a bit of welly.

Cone break-up?

 

[S-Man]

Cone break-up?

Possibly. The 15W is supposed to be lower distortion than the CA15RLY used in the CAOW1, however my measurements show an increase in distortion in the 1.5 to 2KHz region. It's only 0.5% THD though, so unlikely to be audible.

I think I will try increasing L4 from 1.04mH to 1.3mH. This take a dB or 2 out of the 1 - 2KHz region and actually improves the phase tracking slightly. It also moves the XO point down to about 2.3KHz.

 

[Fatmarley]

It's well worth checking the off-axis response. I've had brightness in the past that didn't show on-axis.