Introducing Directiva - An ASR open source platform speaker project

[Rick Sykora]

As for the z-offset delay, there is a pretty good minidsp app note on how to measure it. You need to have all your wiring done correctly and use the analog input. Inputs 1 and 2 are left and right respectively and outputs 1 and 2 are the tweeter and woofer respectively. With this wiring, the minidsp routing should be thus:

Setting up the sweep with a specified acoustic timing reference, after you do your measurement, go to the impulse graph, click on the box on the upper left and change the y-axis units to be %. You will also need to zoom in on the x-axis until you can see something around the 0 time mark that looks like this:

The delay needed is the difference between the time at the right peak minus the left peak time - 1 ms. For this case the result in 1.1 - 0 - 1 = .1
This is an approximation and I usually try the setting a little above and below to get the best reverse null measure. In this case, ended up using .12 ms for the stock Directiva.

 

[Rick Sykora]

Measure the delay (will show details in next post) and after tweaking around it for best result, here is the combined response and the reverse null...
View attachment 120656
This is not a full crossover implementation. Still need to do some boost at both ends.

I will also add that I tried the same delay optimization process with a lower crossover frequency (1.6 kHz rather than 2.4 kHz) and the resulting null was much less. This is not conclusive as there may be other crossover points to come, but am sticking with the higher cross for now.

 

[Rick Sykora]

While the fancier cabinet has been put on the backburner until I can work out the next steps for the PR tuning, @ctrl wanted some measurements to compare directivity to his sim. So, thought I would share as they look promising. Notably with crossover in place, no sign of 2-4 kHz rippling seen in earlier on-axis measures.

As for some conditions, these were done at 78 dB, 1m and 3 ms gating. Used the crossover settings from the stock cabinet...

One caveat is I do not have a turntable or any measured indexing capability, it is just a careful manual repositioning of the speaker. I check the mic alignment using a laser pointer. The top line is on-axis and the middle one is about 45 degrees off-axis.

 

[ctrl]

Here are Rick's measurements normalized to the axis frequency response to better estimate the potential of the v17a version.
According to Rick, the blue or purple curve should represent about 45°.
That looks pretty good!

Now let's compare the simulated frequency responses for version v17a with the real ones Rick measured.

You can see that the simulation predicts reality quite well up to 7-8kHz.

Compared to the simulation, Rick measured at about 10°, 30° and 40°.
(Pay attention to the scale)

 

[Lbstyling]

While the fancier cabinet has been put on the backburner until I can work out the next steps for the PR tuning, @ctrl wanted some measurements to compare directivity to his sim. So, thought I would share as they look promising. Notably with crossover in place, no sign of 2-4 kHz rippling seen in earlier on-axis measures.

As for some conditions, these were done at 78 dB, 1m and 3 ms gating. Used the crossover settings from the stock cabinet...

View attachment 120787
One caveat is I do not have a turntable or any measured indexing capability, it is just a careful manual repositioning of the speaker. I check the mic alignment using a laser pointer. The top line is on-axis and the middle one is about 45 degrees off-axis.

Looks to me like the brown line would be the best angle to listen on (a la geddes style), Perhaps somewhere between brown and purple (whatever angle that is) it also looks like it would not require any if tilt at that angle either. CD response can sound bright when EQed flat on axis.

Aesthetically, this could be a good thing as the speakers may end up pointing forward with no toe.

Your beating the SIM by quite a margin in terms of CD over 9khz

Very good!

You may have answered this, but what is expected max SPL?

 

[Rick Sykora]

Looks to me like the brown line would be the best angle to listen on (a la geddes style), Perhaps somewhere between brown and purple (whatever angle that is) it also looks like it would not require any if tilt at that angle either. CD response can sound bright when EQed flat on axis.

Aesthetically, this could be a good thing as the speakers may end up pointing forward with no toe.

Your beating the SIM by quite a margin in terms of CD over 9khz

Very good!

You may have answered this, but what is expected max SPL?

Depends on how low you like your bass. With the current tuning, sim can hit 100 dB at 40Hz before exceeding excursion limits. Not done with tuning or eq, so not sure where it will land. Do not anticipate any issues at 1 kHz though. Am keeping an eye on distortion, but no problems so far with one speaker at tested levels. One speaker is pretty loud at just over 1m!

 

[Rick Sykora]

Tried a few more changes later yesterday, without any major aha moments. This included:

1. Reducing the volume by a liter (used a sealed liter bottle).
2. The cotton-based damping from the Celuaris SPK5.
3. Using bar clamps to tighten the front half to the back half at the top and bottom (where there is no glue currently).
4. Taping off the PR to ensure no major leaks around it.
5. Damping between the woofer magnet and the brace to (ala KEF) to see if it reduced any known resonances.
6. Ran an impedance check to verify tuning and look for major resonances...

The impedance measure showed that the prototype is close to the SIM. One change was more interesting. I added a sizable m6 bolt to the back of the PR and brought the tuning much closer to the model.

Here is the measurement pic...

It is pretty clean of major resonances but there is something going on right around box tuning. It is showing with or without the added mass. Given that major sections of the cabinet are not fully secured, it is likely cause.

 

[Lbstyling]

Depends on how low you like your bass. With the current tuning, sim can hit 100 dB at 40Hz before exceeding excursion limits. Not done with tuning or eq, so not sure where it will land. Do not anticipate any issues at 1 kHz though. Am keeping an eye on distortion, but no problems so far with one speaker at tested levels. One speaker is pretty loud at just over 1m!

I assume with a HP filter set at 80hz, you can get to 105db? (This is reference level for home cinema use). Tweeter is likely the limiting factor.

It would be nice at some point to gather some data on of it can do it (as so few do), as it's a standard threshold. what power it takes to get there and what the distortion is like at that level would be good to know. I wish Amir tested this.

For films, people typically like higher SPL peaks than music.

It would also be nice to offer a higher port tune for people to swap in for use with a sub.?

 

[MrPeabody]

I assume with a HP filter set at 80hz, you can get to 105db? (This is reference level for home cinema use). Tweeter is likely the limiting factor.

It would be nice at some point to gather some data on of it can do it (as so few do), as it's a standard threshold. what power it takes to get there and what the distortion is like at that level would be good to know. I wish Amir tested this.

For films, people typically like higher SPL peaks than music.

It would also be nice to offer a higher port tune for people to swap in for use with a sub.?

For home cinema use, can't you just grab four of any decent speaker and stack them however is convenient and then wire them in the usual series/parallel configuration? Four for each channel of course.

Oops, I'm sorry. The devil made me do that.

 

[abdo123]

For home cinema use, can't you just grab four of any decent speaker and stack them however is convenient and then wire them in the usual series/parallel configuration? Four for each channel of course.

Oops, I'm sorry. The devil made me do that.

Actually tonality is quite important for speech comprehension, the moment i realize i’m raising the volume too much to understand someone i know i fucked up my EQ.

 

[ctrl]

Tweeter is likely the limiting factor.

I perform such calculations infrequently, so please check.

If the Seas-DXT is equalized to a linear frequency response up to 20kHz, it delivers about 87dB at U0 = 2.83V - according to manufacturer specification. On average, the impedance is 6 ohms.

The long-term power handling of the Seas-DXT is P1 = 55W, peak power handling is P2 = 150W - in each case with an appropriate crossover.
This results in U1 = (P*R)^0.5 = (55*6)^0.5 = 18V and U2 = 30V.

Long term sound pressure level (dB) = 87 + 20*log(18/2.83) = 103
Peak level (dB) = 87 + 21 = 108

The thermal power handling of the tweeter @105dB is therefore sufficient.

It would also be nice to offer a higher port tune for people to swap in for use with a sub.?

With an active crossover, any crossover frequency to the subwoofer can be set.
A second order high pass at e.g. 70Hz is no problem at all.

 

[Zvu]

Bruno Putzeys is using that tweeter in a Kii3 and BXT with a truckload of woofers.

 

[Rick Sykora]

A few more brief updates on Directiva...

1. Found hiss is much less if you turn off left channel input. By default it is not used by REW.
2. I tried a large sheet of 2 inch thick open cell foam dividing the cabinet in half between the woofer and PR. No major resonance reduction unfortunately. EDIT: There were some limited improvements, but no major across the board reduction
3. Replaced the large bolt with a smaller one as added mass for the PR. While no major effect on the lower resonances, the upper ones drop off entirely. EDIT: could not reproduce, so withdrawn. The obvious change is the expected somewhat lower tuning. Lost the measurement when I tore down my measurement setup to do some room measurements. Will share a pic tomorrow.

 

[Rick Sykora]

Not much to report today. Too nice a day to be doing much speaker testing.

Did get the test rig back in place and should be back at it in the morning!

 

[Lbstyling]

Long term sound pressure level (dB) = 87 + 20*log(18/2.83) = 103
Peak level (dB) = 87 + 21 = 108

The thermal power handling of the tweeter @105dB is therefore sufficient.

With an active crossover, any crossover frequency to the subwoofer can be set.
A second order high pass at e.g. 70Hz is no problem at all.

RE: SPL for the tweeter: great!

Having a higher port tune for the sub reduces excursion and hence distortion. These advantages are not compensated for with EQ. It's not a deal breaker by a long shot, but my thinking was multiple port lengths with corresponding EQ profiles for the Fusion is a nice addition. I

have a large KEF floorstander that has swappable ports for different rooms. A nice feature.

 

[Rick Sykora]

I have run through all the different damping materials fairly exhaustively and have reached a point of diminished returns. Am going to open the cabinet and line it with the Sonic Barrier entirely. After that the Acousta-Stuf polyfill seems the next most helpful, but want to move on to some crossover fine tuning and looking at off-axis response.

I have taken some time to do some listening and know some bass boost is in order, but is room dependent, so will make a initial correction and can see where it goes after Amir tests.

 

[Rick Sykora]

As I measured up the cabinet internals to apply the Sonic Barrier, realized I should finish the cabinet assembly. So tore it all down, did a final sanding, air cleaning and wipe down. Really want to ensure no small wood chips or sawdust gets into critical parts of the drivers and passive radiator.

The only downside of the clamshell is lots of surfaces the glue all at once. I started with the side and spread the glue around all the joints with a small paintbrush so it is less likely to drip. Next, I tilted slightly and put glue bead on the front baffle joints and pushed the two halves together. Wiped off the excess glue and clamped (and wiped off glue again). Now it needs to set...

 

[ctrl]

Having a higher port tune for the sub reduces excursion and hence distortion. These advantages are not compensated for with EQ. It's not a deal breaker by a long shot, but my thinking was multiple port lengths with corresponding EQ profiles for the Fusion is a nice addition. I
have a large KEF floorstander that has swappable ports for different rooms. A nice feature.

I don't have Rick's near field measurements of woofer and SBAcoustics-PR yet, so the following is just a rough estimate.

If the Directiva is configured with a LR4@50Hz HP, then the excursion of woofer and PR should roughly correspond to this...

... at a sound pressure level of 105-106dB (this requires about 250W), depending on how the crossover tuning turns out.
The selected crossover frequency to the subwoofer is extremely low (for a bookshelf speaker). The advantage of connecting the subwoofer as low as possible is that the group delay only increases dramatically at very low frequencies.

Since this sound pressure level also represents the (thermal) power limit of the tweeter, the port tuning does not actually need to be changed.


If you only want the lowest possible excursion of the woofer, set the crossover frequency to LR4@80Hz, which reduces the excursion of the woofer significant.
This does not change the maximum SPL, since the power limit of the woofer is reached at 250W.

 

[abdo123]

I don't have Rick's near field measurements of woofer and SBAcoustics-PR yet, so the following is just a rough estimate.

If the Directiva is configured with a LR4@50Hz HP, then the excursion of woofer and PR should roughly correspond to this...
View attachment 121316
... at a sound pressure level of 105-106dB (this requires about 250W), depending on how the crossover tuning turns out.
The selected crossover frequency to the subwoofer is extremely low (for a bookshelf speaker). The advantage of connecting the subwoofer as low as possible is that the group delay only increases dramatically at very low frequencies.

Since this sound pressure level also represents the (thermal) power limit of the tweeter, the port tuning does not actually need to be changed.


If you only want the lowest possible excursion of the woofer, set the crossover frequency to LR4@80Hz, which reduces the excursion of the woofer significant.
This does not change the maximum SPL, since the power limit of the woofer is reached at 250W.
View attachment 121317

Are you into speaker design? I have a little project going on that i would love to have your input on.

 

[ctrl]

Are you into speaker design? I have a little project going on that i would love to have your input on.

As a hobby, yes. Please send me a PM then we can have a chat about it.