DIY cardioid monitors

Gotcha! I chose to ignore the fact that you have the mid ''breathing' sideways.
Still, sometimes theory is misleading and best sound is found somewhere unexpected.
Give the low xo point a shot, when you have time, I'd like to hear your thoughts at this point

 

After I realized that I had all the parts I needed for a rudimentary first shot and that it wouldn't take long to simply hook it up, I did just that. Within 15 minutes or so it was up and running. XO frequency should be around 1.5kHz-ish.

The response isn't too bad except for a large bump around 700-800Hz.
Listening position RTA:

... yeah. So take my first impressions with a grain of salt. Unfortunately, according to the simulation, there's no easy fix for this for now. Not with parts I have and even then it would seemingly take a complete redesign of the crossover.
It is nice to see that the tweeter is mostly linear, though I would prefer not to see such a dip at 10kHz and the following bump. I'd prefer a response closer to that of the 2 way. Maybe with a waveguide?

With that said my first reaction upon hearing it was to make a convolution filter in RePhase to correct for the excess phase that the 2nd order filter introduces. That should give you an idea.
To put it short: It lost most of the magic that I was used to, more so than that weird 1st order MT filter. Maybe it's a matter of getting used to the "normal speaker" sound that it has now, but I think it's pretty boring this way.
To be more specific: It lost that speed, the transient handling and the three dimensional imaging. The visceral impact the bass once had is now muddied, imaging diffused. I expected that by connecting the tweeter in inverted polarity I could keep it, but the 0.15 Ohm of the series coil for the mid/woofer likely does no good here.

With the convolution it regained some of the imaging mostly. Also some of that hyper focused sound. But not the speed, transients were still kind of soft, etc. And of course it's not as coherent as having a single driver covering the whole MT region.

The cap I'm using now is of somewhat worse quality, but I was surprised by how good it still seems to be. Waxed paper instead. There's a boring old cement resistor in parallel to the tweeter now and I could exchange it for three or four parallel high quality wirewounds I have. Maybe that could help a bit. But in general I think the tweeter just doesn't match the widebander in speed and resolution. I've tried listening to the tweeter on its own again a couple days ago and it just didn't seem as resolving.

___
Keep in mind that I'm after a very specific sound and have been listening to first order speakers for 8 years now, all widebanders. Even the ones on my desk have mostly been widebanders and if not I always use RePhase and EQ APO to correct their excess phase for a nice looking step response. Though if I'm honest the widebanders I've had on my desk in the past have been shitty, lol.
I don't want it to seem like I have this magical recipe for speakers and the others are inferior, no. The tweeter is smoother sounding. And with a speaker built for wider dispersion - narrower baffles for example - I think you could get much better imaging from a 2nd order XO than I can with a cardioid midrange mated to a tweeter.
Plus my speaker is kind of ridiculous with no HP on the widebander. It makes pretty much the same excursion as the woofer and there's no way around it. But in the end it's the sound I'm looking for. Yes, even the small speaker. If I'm honest I still listen to Hathor more than Amon.

 

Finally did the "XO12" upgrade I've been meaning to do for a while now.

XO12 upgrade consists mainly of better caps all around and a slightly different tuning of the notch filters.


Sound:

• There's a big deal of placebo at play here, so take this with a grain of salt:
• I expected more smoothness to the treble, but it wasn't as dramatic as expected.
• However I think it sounds quite a bit more transparent. Imparts less of its own coloration. It could be argued that the caps I'm using are still colored sounding, but parts quality should be more than sufficient for high Q notch filters. Though I'd normally use two Mundorf Supreme resistors here.
• Expanding a bit on that thought I think the tonal palette of the treble is wider now. Timbre is more true to life.
• Similarly dynamic contrasts are more evident.
• All of this really only applies to the treble. For example sibilants seem better integrated. And honestly the entry level MKPs (WIMA) were already sufficient for this application. It's not like with a tweeter series cap.

The woofer cap didn't seem to do all that much. Its purpose is to filter the metal driver cone breakup, so it's not really part of the low pass per se.

• It's not at all like when I replaced the standard electrolytics to the huge oil filled motor run capacitors for the woofers in the OB. In that case bass got significantly tighter, more defined. Better note delineation.
• I already had a WIMA MKP here. Still, this vintage hybrid PIO does some good.
• As above I think the tonal palette is richer now. But it's really hard to tell when the widebander crosses so much lower than the frequency the capacitor acts on. Crossover is at 200Hz, capacitor works from 2kHz on.
• Is it a waste of money? I don't think so still. Midrange transparency probably went up half a notch. Maybe also from the notch filter capacitors?

I also made two tweeter crossovers so that I can more easily switch between the 2nd order and 1st order crossover and put them in the tweeter enclosure. I changed the values slightly. Lowered XO frequency for the 1st order to ca 4kHz and the 2nd order has somewhat less of the 800Hz bump now.

The tweeter crossover is not really an upgrade in parts quality. It's more about usability and some FR tweaks.


Here I focused more on the in-room response this time around than the listening axis response. As a result the 1st order XO is somewhat brighter than neutral and the 2nd order is somewhat warmer than normal in standard measurements.

In terms of sound the 1st order crossover wins by a long shot for me. 2nd order sounds neutered.
The tweeter does handle high SPL better than the widebander, but it's not that dramatic. Also the ceramic tweeter sounds more brittle than the Alu sandwich widebander, for lack of a better word. I think it imparts more of its own coloration than the widebander. More of a hard cone sound. Maybe removing the phase guide could help.
That it's a downgrade in technicalities to me is something I already mentioned. Less immediacy, worse imaging and resolution. 2nd order sounds smoother, but I think artificially so while 1st order actually sounds less smooth to me than the 2 way in a way.
The 2nd order doesn't sound quite coherent to me while the 1st order does quite well in that regard. And it's not just dispersion, it's mainly the note handling, attack and decay characteristics and imaging.

So why did I bother with testing the tweeter again? I heard a nice tiny bookshelf speaker (Fischer&Fischer SN70) with a 15cm alu cone Seas Prestige midwoofer and a Scan Speak Classic 28mm textile dome. Presumably L16RN-SL and D2905/970000, respectively, That tweeter should be better than the ceramic I have here. Ultimately I liked what I heard from the lower to middle midrange, found it somewhat incoherent and found the tweeter far too soft sounding. Maybe one of the TOTL Satori ring radiators would win me over. I want to give the BlieSMa TW25Be a listen somehow, but ultimately adding a tweeter probably isn't to my liking.

 

MEASUREMENTS

Spoiler: FREQUENCY RESPONSE IN ROOM

Three way 1st order looks like a major win here, but it doesn't really sound like that. I'll blame it on the comb filter effects.
For the 2nd order I had to sort of balance having a large 800Hz bump with a bit of a suckout in the upper mids.

Spoiler: EXCESS PHASE

As expected neither the 2 way nor the three way with the 1st order XO have excess phase. 2nd order is 90° at 2kHz, so that should be the crossover frequency.

Spoiler: STEP RESPONSE IN ROOM

Note that the first reflection already happens around 1ms, but I still think it's somewhat useful to look at.

I didn't take measurements at closer distances to get good looking distortion measurements, but the tweeter simply has less distortion than the widebander in the treble, so the order should be pretty clear. 2 way has the most treble distortion and 3 way 2nd order XO the least. But I'm not too sure how audible that really is at normal listening levels.

I think the measurements are actually less interesting than they may seem at first. But I think I did find a good balance for the 1st order XO and it shows.