• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Introducing Directiva - An ASR open source platform speaker project

TimVG

Major Contributor
Joined
Sep 16, 2019
Messages
1,181
Likes
2,573
I don’t think a microphone can tell what is direct sound and what is reflections like the human brain does. So there is no point in maintaining direct sound imo. We’re hearing the sum that is arriving at the microphone anyway.

All I can say is it certainly does matter - at least in the testing for personal sake I have performed in the past.

Tests were also performed by a couple of forum members in the past, and both in the first test by @thewas, and the second by @ctrl the speaker with the flattest direct sound (Revel F52 & Grimm LS1) was generally preffered with regards to coming closest to the recording itself. Link to the 2nd test because it also involved a KEF Reference 3 which arguably has the smoothest in-room response out of any of those speakers.


This is also my own preference in terms of speaker design. An as neutral direct sound as possible, within a large listening window, with reflections to match (with emphasis on the horizontal reflections). So this means constant (or close to) directivity over a large bandwidth. Basically as you move off-axis, the sound should simply become more quiet, but remain the same in terms of timbre - that is just my personal opinion of course.
 
Last edited:

Newman

Major Contributor
Joined
Jan 6, 2017
Messages
3,448
Likes
4,211
I don’t think a microphone can tell what is direct sound and what is reflections like the human brain does. So there is no point in maintaining direct sound imo. We’re hearing the sum that is arriving at the microphone anyway.
???
It is tested and proven that flat direct sound FR is the #1 determinant of speaker preference.
 

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
7,424
Likes
7,940
Location
Brussels, Belgium
All I can say is it certainly does matter - at least in the testing for personal sake I have performed in the past.

Tests were also performed by a couple of forum members in the past, and both in the first test by @thewas, and the second by @ctrl the speaker with the flattest direct sound (Revel F52 & Grimm LS1) was generally preffered with regards to coming closest to the recording itself. Link to the 2nd test because it also involved a KEF Reference 3 which arguably has the smoothest in-room response out of any of those speakers.


This is also my own preference in terms of speaker design. An as neutral direct sound as possible, within a large listening window, with reflections to match (with emphasis on the horizontal reflections). So this means constant (or close to) directivity over a large bandwidth. Basically as you move off-axis, the sound should simply become more quiet, but remain the same in terms of timbre - that is just my personal opinion of course.

I'm not denying or arguing against any of this, but at the end of the day the microphone will record the sum that is arriving at it and that sum will be what we actually hear when the recording is being played back.

These tests that you mentioned did not have EQ at all, so obviously the speaker with the best default tuning will be most preferred.
 

Newman

Major Contributor
Joined
Jan 6, 2017
Messages
3,448
Likes
4,211
Well please clarify.
 

TimVG

Major Contributor
Joined
Sep 16, 2019
Messages
1,181
Likes
2,573
I'm not denying or arguing against any of this, but at the end of the day the microphone will record the sum that is arriving at it and that sum will be what we actually hear when the recording is being played back.

These tests that you mentioned did not have EQ at all, so obviously the speaker with the best default tuning will be most preferred.

By that definition however in the 2nd test, the KEF Reference 3 should have won featuring the best in-room response. In the end the Grimm LS1 (which had a flatter direct sound) was the winner, despite still having a dip in the overall soundpower (as any non-concentric multiway loudspeaker).
 

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
7,424
Likes
7,940
Location
Brussels, Belgium
By that definition however in the 2nd test, the KEF Reference 3 should have won featuring the best in-room response. In the end the Grimm LS1 (which had a flatter direct sound) was the winner, despite still having a dip in the overall soundpower (as any non-concentric multiway loudspeaker).

Okay i am quite intrigued now.

Do you think the fact that it's a binaural recording it preserves some of the pshycoacoustics you would hear in-room successfully? The Grimm LS1 extends lower in frequency response, which we know is even a bigger proponent of speaker preference than on-axis linearity. But that extra extension is mainly below 40 Hz where there is barely any content in mainstream music.

Looking at it more closely now, the Grimm LS1 has quite the treble lift on axis, making it's ER / PIR have quite the flat slope, so perhaps what the microphone picked up in that instance was indeed flatter than the other speakers which was preferred more when played back on headphones.
 

TimVG

Major Contributor
Joined
Sep 16, 2019
Messages
1,181
Likes
2,573
Do you think the fact that it's a binaural recording it preserves some of the pshycoacoustics you would hear in-room successfully?

I cannot say for sure, but there appears to be a correlation - the only thing I've concluded with certainty so far is there is no such thing as a universal 'target' curve - simulated or measured.

Looking at it more closely now, the Grimm LS1 has quite the treble lift on axis, making it's ER / PIR have quite the flat slope, so perhaps what the microphone picked up in that instance was indeed flatter than the other speakers which was preferred more when played back on headphones.

I'm not sure how the LS1 were positioned in that test, but there is a setting which intends for the speakers beams to cross in front of the listener to reduce the sidewall reflections.

1655988171790.png


I do remain by my initial point however: The test by Erin is a good effort, and a good test by itself, I would however have like to seen the 'target' be based on the anechoic data, the direct sound, instead of an in-room target curve.
 

DualTriode

Addicted to Fun and Learning
Joined
Oct 24, 2019
Messages
893
Likes
593
Back when Rick proposed the project, I was hot for the possibility of using a sealed box for smaller size, easy integration with a sub (no high pass filter required) and no port pipe resonances colouring up the midrange.

As it turns out, the 6.5" Purifi's Thiele Small parameters require a very small sealed box size.

Classic acoustic suspension design wisdom is that the air in a sealed box acts as a linear (and linearizing) spring, dominating over the driver's suspension compliance (which has its own distortion causing non linearity).

In reality, the thermodynamics of air show that an air spring isn't really linear and if air is compressed too much, it causes harmonic distortion on speaker's output that can easily exceed the distortion caused by the speaker's suspension.

Linkwitz here estimated the % second harmonic distortion caused by air in sealed box as =0.014*SD (in cm^2)*driver displacement(in mm)/box volume (in L).

I simulated a sealed box for the Purifi, and then used Linkwitz's formula to estimate the 2nd harmonic distortion from the air trapped in the box itself, at different listening levels. Here are the results below, compared to the total harmonic distortion of the Purifi vented prototype measured at ASR.

index.php


The sealed distortion is quite a bit higher than a vented and wastes the Purifi's low distortion potential. I also expect the mids might distort when played simultaneously with loud bass because the air itself is distorting (ASR tests don't measure this).

The 6.5" Purifi is all about playing loud with low THD, but it struggles to pull that off in a sealed box because of its unusual TS params.

Hello DDF and All,

Lately there is a renewed interest in this Purifi 6 ½ inch driver.

I have a new pair coming my way in a day or two.

DDF did you actually test the driver in a sealed enclosure? Or did you evaluate the driver in simulation and calculation only?

Looking at WinIsd it looks like with a Qtc or Qb of 0.71 the calculated box volume is 0.21 cubic feet.

For a Qtc or Qb of 0.51 the calculated box volume is 0.5 cubic feet.

It seems that any 6 ½ inch driver will have the same nonlinear air compression issues in a small sealed box.

My purpose is a pair of smallish sealed speakers for my test bench with an expected max SPL at 5 feet of ~90dB’s

Currently I am using Peerless HDS 830875 Nomex 6 ½ inch drivers in parts Express 0.55 cubic foot knockdown enclosures.

I will do some testing and report back.

Thanks DT
 

TimVG

Major Contributor
Joined
Sep 16, 2019
Messages
1,181
Likes
2,573
FYI A variety of drivers has been tested using Klippel equipment by one of the team members and we have unanimously settled on a somewhat higher efficiency driver (compared to the Purifi) to serve as the bass driver (in the 3-way configuration) to hopefully alleviate some of the issues due to air compression issues.

Stay tuned, we're working behind the scenes.
 

DualTriode

Addicted to Fun and Learning
Joined
Oct 24, 2019
Messages
893
Likes
593
I will stay tuned.

Focused on a small 2-way for desktop, every review you look at shows port interference and noise that shows up as distortion after the software calculates the distortion curve.

I found and confirmed the sealed box, due to air compression, distortion calculation on linkwitzlab.com

I calculated with 133 cm squared area moving 2mm on a 14L sealed box at 0.20% distortion. 2mm movement is plenty SPL for my desktop. That is 0.20% is for any 6 1/2 inch driver in a 14L sealed box which is not bad.

I will be testing with APx555, Gras microphones and APx500 acoustic measurement software.
 
Last edited:

DualTriode

Addicted to Fun and Learning
Joined
Oct 24, 2019
Messages
893
Likes
593
Ready to glue and cut holes.

First measurements:

Sealed box 0.56 cubic feet. Total Q ~ 0.5 this box has larger volume than the total Q ~0.71 and less air compression distortion. I am going to test drive this sealed box on my bench for a while to experience sealed box distortion. As in is it audible or problematic at bench top listening levels.

port

resistive port

purifiBox.jpg


That is a ScanSpeak 2608 tweeter in the photo. I have a couple on the shelf Oops to the budget. This speaker will be on my bench driven by a Topping LA90 amplifier. I am not concerned that the woofer will handle more watts than the tweeter.
 
OP
Rick Sykora

Rick Sykora

Major Contributor
Forum Donor
Joined
Jan 14, 2020
Messages
3,517
Likes
7,027
Location
Stow, Ohio USA
Ready to glue and cut holes.

First measurements:

Sealed box 0.56 cubic feet. Total Q ~ 0.5 this box has larger volume than the total Q ~0.71 and less air compression distortion. I am going to test drive this sealed box on my bench for a while to experience sealed box distortion. As in is it audible or problematic at bench top listening levels.

port

resistive port


View attachment 219058

That is a ScanSpeak 2608 tweeter in the photo. I have a couple on the shelf Oops to the budget. This speaker will be on my bench driven by a Topping LA90 amplifier. I am not concerned that the woofer will handle more watts than the tweeter.

Looks interesting, but suggest you give this its own build thread.

If you want it considered part of Directiva effort, let me know and will add it to the project and can give it a name/number.

Thanks!

Rick
 

DualTriode

Addicted to Fun and Learning
Joined
Oct 24, 2019
Messages
893
Likes
593
Hello All,

I am not really looking at a new build yet. So far I have the Purifi 6.5 inch driver. I am putting it into the same 0.56 cubic foot Denovo cabinet. The tweeter will wait in its packaging.

I want to go back and examine the assumptions made about a sealed enclosure and compressed air caused distortion.

As far as I can determine the “simulated” sealed enclosure distortion was based on the formula found on the Linkwitzlab.com site. Change in volume due to cone motion times 1.4, divided by enclosure volume all times 100 equals % distortion. My favorite simulator calculates that for a Total Q of 0.71 for the volume of the enclosure at ~0.2 cubic feet. This is most likely the number used in post #35. If the enclosure is increased to the selected Denovo cabinet, 0.56 cubic feet, the volume is increased by nearly 3 times, and the distortion is reduced by 60%. With the increase of volume Total Q is 0.5, a not uncommonly selected design number.

I want to go back and take a closer look at the assumptions in post #8 and #35 and make some measurements.

Do the index.php simulation results that appeared in post #35 still exist somewhere? I would like to see them

Thanks DT
 

TimVG

Major Contributor
Joined
Sep 16, 2019
Messages
1,181
Likes
2,573
Just an update that both Rick and myself have resumed work on R2. We both had a bit much on our plate over the past few months, and while I can't speak for Rick, I needed to sort out my measurement rig first so that our data is as accurate and comparable as possible.

Having been able to hear and measure quite a variety of speakers over the past months, I can honestly say that R2 will have been worth the wait.
 

D!sco

Senior Member
Joined
Jun 16, 2021
Messages
483
Likes
381
Glad you all are making progress again.

I had taken an interest in the Purifi 5.25" since it had been introduced early this year. It seemed like a nice form factor for nearfield monitors. In VCAD, it matches the SB Acoustics 5x8" PR almost perfectly. There are dozens of crossovers between the larger purifi driver and the SEAS DXT thanks to this project. They all fit in the smaller .23 cu ft cabinet on PE. This was sort of inevitable.
Full disclosure, I'm not sure I'll build these because it's still $1k in drivers. But I would if it were no object. If I keep working on it I'll probably make my own thread. Needless to say, there would be a lot more work to be done. However, all the tools available in VituixCAD have been used, so baffle step and enclosure response merging has already been accounted for. The one I can't confirm is the tweeter directivity. It's not as simple as copying over the Directiva measurements, as far as I know. The boxes are copied directly from the PE site, and the other models are stock/downloaded from the warehouse on sketchup.
Screen Shot 2022-09-01 at 11.37.10.pngScreen Shot 2022-09-01 at 11.37.38.pngScreen Shot 2022-09-01 at 11.39.22.pngScreen Shot 2022-09-01 at 11.40.27.pngScreen Shot 2022-09-01 at 11.49.16.pngScreen Shot 2022-09-01 at 11.50.00.pngScreen Shot 2022-09-01 at 11.50.43.pngScreen Shot 2022-09-01 at 11.52.01.png

The F3 does take a hit, between 5-10 dB depending on how loud you want to be. Weights can help dig deeper, without them the PR will probably cap out in the 50's with a big bump. Higher weights need some kind of highpass filter. I wouldn't go higher than 40g.
No XO
Screen Shot 2022-09-01 at 11.55.31.pngPurifi PTT5.25X04-NFA-01 Six-pack-NoXO.png
Passive XO
Screen Shot 2022-09-01 at 11.56.33.pngPurifi PTT5.25X04-NFA-01 Six-pack-Passive.pngPurifi NFM XO-schema-1.pngPurifi NFM Six-pack-1.png
A more expensive passive with a lot of filters and a 1.25" baffle extension for the woofer's phase correction.
Purifi NFM XO-schema-4.pngPurifi NFM Six-pack-4.png
Active XO
Screen Shot 2022-09-01 at 11.58.43.pngPurifi PTT5.25X04-NFA-01 Six-pack-ActiveXO.pngPurifi NFM XO-schema-6.pngPurifi NFM Six-pack-6.png
 

TimVG

Major Contributor
Joined
Sep 16, 2019
Messages
1,181
Likes
2,573
Glad you all are making progress again.

I had taken an interest in the Purifi 5.25" since it had been introduced early this year. It seemed like a nice form factor for nearfield monitors. In VCAD, it matches the SB Acoustics 5x8" PR almost perfectly. There are dozens of crossovers between the larger purifi driver and the SEAS DXT thanks to this project. They all fit in the smaller .23 cu ft cabinet on PE. This was sort of inevitable.
Full disclosure, I'm not sure I'll build these because it's still $1k in drivers. But I would if it were no object. If I keep working on it I'll probably make my own thread. Needless to say, there would be a lot more work to be done. However, all the tools available in VituixCAD have been used, so baffle step and enclosure response merging has already been accounted for. The one I can't confirm is the tweeter directivity. It's not as simple as copying over the Directiva measurements, as far as I know. The boxes are copied directly from the PE site, and the other models are stock/downloaded from the warehouse on sketchup.
View attachment 228049View attachment 228050View attachment 228051View attachment 228052View attachment 228053View attachment 228054View attachment 228055View attachment 228056

The F3 does take a hit, between 5-10 dB depending on how loud you want to be. Weights can help dig deeper, without them the PR will probably cap out in the 50's with a big bump. Higher weights need some kind of highpass filter. I wouldn't go higher than 40g.
No XO
View attachment 228057View attachment 228047
Passive XO
View attachment 228058View attachment 228048View attachment 228043View attachment 228040
A more expensive passive with a lot of filters and a 1.25" baffle extension for the woofer's phase correction.
View attachment 228044View attachment 228041
Active XO
View attachment 228059View attachment 228046View attachment 228045View attachment 228042

With the DXT you'll want a large chamfer around it. It prefers either as little or as much baffle as possible. On a regular rectangle you'll get some uneveness centered around the 3kHz range. Even the venerable Kii3 with its generous roundovers hasn't fully mitigated this effect.
 

Joe BKNY

Member
Joined
Jun 28, 2021
Messages
26
Likes
9
Just an update that both Rick and myself have resumed work on R2. We both had a bit much on our plate over the past few months, and while I can't speak for Rick, I needed to sort out my measurement rig first so that our data is as accurate and comparable as possible.

Having been able to hear and measure quite a variety of speakers over the past months, I can honestly say that R2 will have been worth the wait.
Thank you for your (ongoing) work on the once and future Directiva!
 
Top Bottom