Directiva r1.2 design and build

[Alan J]

And another - fewer components, lower crossover point and accordingly a larger LF inductor), along with a softer HP slope, and less of a wrinkle in the DI. Here, the drivers are out of phase at the crossover point and vertical directivity in the crossover region is biased upward. In the current intended room, these will sit on a 36" high counter at the front of the room so that may be advantageous.

The larger LF inductor means, in addition to the larger part, more series resistance, though I planned for 0.5 ohms in the preliminary bass alignment and I think I'll be within my crossover volume estimate. The plan is to check back before ordering parts, and I have the option of doing near field measurements with a mock crossover volume.

 

[Alan J]

One more, and that's it for today... Very similar to the last one but returning to a 4th order slope on the tweeter, and the crossover point back up to around 2 KHz.

 

[terryforsythe]

I may have already stated this, but it is worth repeating. Use air core inductors to avoid core losses!

 

[Alan J]

I may have already stated this, but it is worth repeating. Use air core inductors to avoid core losses!

Yes, definitely.

My entré to speaker building was a pair of Dennis Murphy's Affordabe Accuracy monitors, his original Pioneer LR-22 retrofits. Dennis used the Pioneer iron core in those due, I think, to tight space constraints and a touch of economy. But I'm not venturing into that "is this OK in this circumstance" territory!

 

[Alan J]

I'm back from family event time and ready to complete this project. Below is the top candidate crossover, with which I'm pretty satisfied.

My priorities were

• SPL in the listening window - smooth and with prescribed modest downward slope. Due to the intended placement, the listening window is from 0 to 20 degrees rather than +/- 10.
• Sound power - smooth with prescribed downward slope

Slopes are 4th order. In the crossover region, the central lobe is tilted up a bit, which is a good thing considering the placement.

Key off-axis angles - seated position (+5), standing further back (+15), and floor and ceiling first reflection angles:

This seems to be OK, especially with a rug and in a pretty lively room.

The total volume of the components, using 15 gauge air core inductors and 400 V polypropylene caps, adding 20% for the mounting board and wires, is 215 ml, less than half my half-liter budget. The 380 mOhm inductor is also below the half ohm I was allowing for. So the final bass alignment should be fine.

 

[Alan J]

As I prepare to get my order in to PE for Monday morning, I've looked back at the active crossover prototype that sounded so good. As with the passive design in the prior post, the listening position here is 140 mm above the tweeter axis. The prototype had less tolerance for ears above the design axis (standing further back in the room), yet we noticed no issues when we tested. When we did our listening tests on the prototype, I'd had the DSP set up to tweak the tweeter levels but ended up having no need to do so. This adds to confidence in the passive design.

 

[Alan J]

OK - I promise to stop fiddling. But I'd love to hear opinions on voicing options.

In Post 145,

• R2 changes the mid-band level, creating a broad hump or valley from around 1-5 KHz
• R3 changes the slope in the treble region, tilting the SPL curve past 2 KHz

In the variant below, which has essentially identical response as shown,

• R2 changes the DXT level, shifting the entire response beyond 2 KHz up or down while leaving the slope close to flat
• R3 changes the slope as before

Which is preferred?

 

[Paweł L]

I can spend months fiddling with x-o and different configuration until I get tired and farther improvement can't be found. Some drivers are 'special' in their behavior and need special corrections. Some one wanted me to use Kartesian 32mm some tweeter with SB17 alu - nothing worked, as they both exhibit very bright character. Eventually I used the Kartesian with the old SB17 paper cone, but there's still something about that tweeter which makes listening rather tiring in the long run. By now I have plenty of caps, coils and resistors to build a few pairs of speakers.
Anyway, just make one version, if it's a bit off, try to EQ it to the point that you like the sound and then, redesign the X-O. Differences of 0.5dB, and slope characteristics are actually quite noticeable especially around 1-5kHz where the driver's pass bands meet - directivity, power response and all this shabeng..

 

[Alan J]

That's sensible. I do expect it to sound very much the same as the DSP-based prototype with which we were so happy. With that said, just in case, I expect to place a wooden block inside to take up the crossover volume and start with the crossover outside for easy modification (Wago connectors, I suppose).

With the ability to do modest tweaking in mind, what I'm looking at here isn't which configuration is best (they're essentially the same); it's which configuration gives the best ability to tweak with changes in resistors only.

 

[Alan J]

Parts are ordered. I went with Post 145 which should allow enough shaping if needed. Parts should be in hand at the end of the week.

 

[Paweł L]

Parts are ordered. I went with Post 145 which should allow enough shaping if needed. Parts should be in hand at the end of the week.

Did you check how the impedance looks like with the tweeter section in this configuration. I'm asking because I never checked how moving the resistor R3, parallel with the coil, in front of R2 affects the circuit. I usually have it (R2) placed right in front of the driver so it dissipates as little wasted heat as possible. And yeah usually you can unwind the coil make a tap for a smaller value and wind the rest back, say make 1.5mH and 1.3 with the coil. Saves money and time while prototyping. There was actually Italian design which had tapped coil and switch for tuning. I also have some additional resistors, small caps 1-1.5-2uF on the side for parallel value mods.

 

[Alan J]

Yes - looks fine...

R2 (Post 145) dissipates the most, but not a lot...

I did try R2 on the other side of R3, in parallel with C3, in order to have R2 more directly control the Purifi level. In that configuration, I couldn't do quite as well with lobing.

 

[Paweł L]

My concern is the almost -50° phase around 3-4kHz and then staying -25°, most likely due to capacitor in parallel with the tweeter. Most stable amps will take it, but the less phase rotation especially capacitive load angle the better, switch the EPDR option ON in impedance graph section. Did you try changing the slope by adding more resistance to the L2 coil, I have around 2ohms added to coil with that pesky Kartesian to line up the acoustic phases and smooth out the 2.5-5kHz so it doesn't beam so much, no waveguide though. Maybe you want the opposite in your case. Sometime different C2-L2 combinations values yield better results. Definitely passives are so much 'fun' ...

 

[Alan J]

Thanks for raising the question, Pawel. Here's the impedance plot with EDPR. The minima are 2.7 Ohms and low frequencies and 2.2 Ohms at the high end. Removing C3 results in an upward slope to EDPR at high frequencies but doesn't change the minimum.

Output power seems reasonable (with the peak voltage bumped up):

This alternative...

Has minimum EDPR of 2.7 Ohms, with a corresponding slight decrease in generator power at the high end. As noted above, this alternative doesn't do quite as well with lobing in the crossover region.

 

[Alan J]

I had a DM request for the VCAD file, so here it is.

Note the -140 mm driver offset in the Room tab. This is for the relatively low placement of the speakers in the intended room (tweeter axis 140 mm below ear level when seated). The options are below (don't recall what gets saved). Note the 0-20 degree listening window and user vertical angle list. I use the latter to look for problematic shifts when standing. With the -140 mm offset, I look at +10 degrees and am trying to keep all shifts below 3 dB.

The saved R1 is Post 145 (though maybe not without some further fiddling with values) and R8 is the same with the component values that were available when I ordered.

 

[Alan J]

Here's the first measurement of the crossover as built [EDIT: this is on-axis SPL], overlayed with the VCAD prediction for the measured component values. It's become common to go from measurements to model predictions and have the predictions come so close to real world measurements. It's still a delight every time.

Per the VCAD model, I'd benefit from shaving a bit off of R3, for which I need to wait for stock at PE. Other than that, I'm not at this point seeing anything that needs to change.

 

[terryforsythe]

Good job!

 

[Paweł L]

I'm so intrigued that the change in baffle shape with it's slanted edges around the tweeter changed the freq response so much in the right direction, resulting in much easier X-O with less need for compensation above 5k. And I bet now the spkr is also more manageable around crossover point with way better directivity pattern. Alan if you gonna do some off-axis measurements, please post them. I'm almost ready to run my baffles over the table saw...

 

[RickS]

I'm so intrigued that the change in baffle shape with it's slanted edges around the tweeter changed the freq response so much in the right direction, resulting in much easier X-O with less need for compensation above 5k. And I bet now the spkr is also more manageable around crossover point with way better directivity pattern. Alan if you gonna do some off-axis measurements, please post them. I'm almost ready to run my baffles over the table saw...

Thought we explained this tradeoff in r1, but I drove it towards a simpler cabinet. It is a hobbyist speaker and so left fancier cabinetry to other hobbyists (as Alan has done here).

Alan has been very disciplined with regards to his effort here and appreciate him continuing in the spirit of Directiva.

 

[Alan J]

The basic plan (which I thought I posted but must not have hit the Post button):

1. Crossover build. I may go with separate boards so that both inductors can lay down flat.
2. Bass alignment and near field measurements
3. Off-axis far field measurements - planning on going all the way around, as I'd love to verify the sound power predictions
4. Driver mounting, etc. with unit B (Purifi is broken in; DXT is still in the packaging) and stereo listening tests

Regarding the crossover - I went back to the no-delays (all IIR) active crossover design and yes - the DXT filtering is much simpler. Referring to Post 146, the two notch filters go away. It does then need more attenuation and a low pass with custom (< 12 dB/octave) slope.

Normally, I'd say that I can get some quick measurements in the +/- 45 degree range. But I'm getting ready for some travel and It's likely to be two weeks before any of this happens :-(

[EDIT: I'll grab some partial data right now.]