Introducing Directiva - An ASR open source platform speaker project

[Rick Sykora]

Not that you want the hastle, but I assume the larger passive radiator version would solve this distortion peak.
Without that peak, you would be well under 0.1% distortion at 95db. What you have is already quite something for a standmount!

Here is the Revel m106 for reference at 96db...View attachment 124053

Big money Revel 328BE floorstander:

View attachment 124056

Without the 400hz peak, your nearer the 328BE than the m106.

By the trend, your design looks like it is likely otherwise capable of clean output at 100db+ without auditable distortion

(debatable around -48db 'ish' for me, but not even slightly 'offensive' until at least -28db from 400hz to 7khz for me personally)

Not that you want the hastle, but I assume the larger passive radiator version would solve this distortion peak.
Without that peak, you would be well under 0.1% distortion at 95db. What you have is already quite something for a standmount!

Here is the Revel m106 for reference at 96db...View attachment 124053

Big money Revel 328BE floorstander:

View attachment 124056

Without the 400hz peak, your nearer the 328BE than the m106.

By the trend, your design looks like it is likely otherwise capable of clean output at 100db+ without auditable distortion

(debatable around -48db 'ish' for me, but not even slightly 'offensive' until at least -28db from 400hz to 7khz for me personally)

I would be very hesitant to compare my distortion measures to Amir's. Will see what results he gets when he measures.

Modeling indicates Directiva should do better than 110 dB.

 

[McFly]

As for that rise in distortion around 400-500 Hz, is likely the the passive radiator. It is primarily 2nd order with some 3rd order mixed in. Realized I posted without the legend, so fixed it this time...

View attachment 123943

If you hadn't said passive radiator Id've said that looks like port resonance, heh

 

[Rick Sykora]

If you hadn't said passive radiator Id've said that looks like port resonance, heh

Frankly, have found the PR results to be underwhelming. The Purifi PRs really call for a different design approach as the output is so high at higher frequencies.

When I get some time, am going to try the SPK5 with a precision port or some other shapes.

 

[McFly]

I've had little success with my purifi PTT6.5's with ports, mind you I didn't try a precision port with massive flares. Yes, monstrous ported bass in the low 30s, but audible port resonance effect every time. I believe it could be sorted by hiding something like THIS (go to port testing section down a bit) in a floor stander, otherwise I'm planning to attempt a transmission line build with it later in the year.

 

[Skeptischism]

Am only only loosely following but am interested to see a proposal. As I often said, expect that there will more than one solution.

Not sure how to answer the power question as it is a function of how loud you listen and how far you are away. I have not measured, but expect sensitivity to be in the upper 80s without any crossover compensation. If woofer is boosted to tweeter level, may be in low 90s.

Of course, there will be hopefully more than one solution. i'm by no means trying to take over the project. Just trying to give it a kick in the bum i'm all for sharing info I garner along the way, so regardless of whether there is any uptake of any board of mine, I would hope to leave the project richer for being involved; just as I would expect to come out of it knowing more than I went in, with the knowledge share typical of this place. If we can get a clearer outline of what is wanted, we'll see how our needs align.

thats achievable with a well laid out chipamp even, maybe even composite LM4780. thats just a tangent though, i'm at the source side at this point. Something to keep in mind though, since we are looking at this thing as an appliance, having built in amps, or at least amps that respond to control signals from the server for power sequencing and DC error protect wouldnt be a bad idea.

sorry just clarifying, as you mention sensitivity as a unit after tweeter compensation, not power needs per driver. So sorry i'm just trying to pull power needs per driver out of that. 2 x 100W per monitor (or 100+50W) would be fairly easy to achieve (and I would think enough for nearfield). i'll have to have a look at the datasheets to see how that would apply to the drivers in you are talking of. jealous of you having these babies to play with there is little chance of my building the speakers, as i'm just finishing off a revelator 2 way design for myself, but call me intrigued for the next coffins I build

of course easily available 2 channel class D modules are plentiful and cheap/cheerful

 

[Juhazi]

I have used Hypex FA123 modules it two 3-way projects, both with sealed bass. They are rated 100/100/50W and I can easily drive the 10" woofer well past 105dB at 1m, up to bottoming. Same with 125W ICEpower amp in another project. Distortion % depends on many factors, it's really difficult (pointless) to compare measurements from different sources.

Hypex has also much more powerul FA models, if the speaker can benefit something from those. The beauty of FA is that it is a ready-to-use package with software and three switchable presets. As well you can feed the board with digital or analog signal, either balanced or unbalanced.

Here one measurement done in-room (mode dip at 33Hz)

 

[Skeptischism]

Oh there's no arguing with the quality of the Hypex, but options options options and IMO camilla DSP is a far more powerful DSP solution.

 

[Lbstyling]

Oh there's no arguing with the quality of the Hypex, but options options options and IMO camilla DSP is a far more powerful DSP solution.

Would you run it on a raspberry Pi then?
The convolution filters are an advantage, but it's a lot of hastle just to get them.

Perhaps someone should look at it as a possible add on modification, but I would imagine it would end up as expensive as the fusion or more, but without the 105db SINAD.

 

[Skeptischism]

yeah nah, I wont be putting anything together with less than 105dB sinad .

you can run it on whatever you wish. there are mac, win and linux versions of it, with headless/remote configuration with mobile websocket.

then you just need multichannel i2s or spdif*xx channels, with i2s preferred.

Either

A. from MAC/PC/Linux->USB->xmos->i2s*xx on bespoke USB input, Minidsp MCHStreamer or okto DAC8, or similar mutichannel dac

or
B. Multichannel i2s directly on ROCK64 or similar Single board computers (rpi does not make multiple sdata pns available, only stereo).

We are looking over other SBCs with i2s facility, like the Rock64pro. they can run camilla and/or act as a usb gadget. This appears as a USB input DSP interface and you can optionally run that with whatever linux build is suitable (support for volumio), or run the player side, with perhaps any 2 channel room correction (Bass and room mode FIR for example) before feeding 2 channel to the USB gadget and taking i2s out, if you want access to all of the audio centric builds for rpi plug and play.

All allow you to use any multiple input DAC, or multiple dacs if you wish as well as choice of amp modules, a stereo module or dual mono per speaker.

 

[Skeptischism]

its possible we can make the USB gadget mode work over ethernet as well, which could mean you could just run whatever you want on your mac or PC (crazy, horsepower draining digital filtering and upsampling/oversampling/transcoding if you want, though you may want to switch those around and do XO on your mac and upsampling last, on the rock64) and have the rock64 USB gadget as a network connected DSP node, with i2s output. not 100% that will work, but it has a good chance of it. it is confirmed to work with other USB I/O. its connected via ethernet, but appears as a sort of USB gadget; like a peripheral to the server.

 

[Juhazi]

Skepticism, yes everything can be done differently and even better. But this thread is not for that, I suppose? But since you seem to be well educated and interested, please wait a bit!

From OP - goals in Rick's words (bolded by me)

Towards the end of last year, @amirm and I started discussing a potential audio system project that would include an open source streamer and an active speaker design. I had already been working on a DIY modular speaker concept and planned to start a design based on it, but had some other projects that needed to be completed first. This allowed me to make some adjustments to my plans and reconsider some of my original ones. Here were my goals for the speaker:

1. The speaker would be active and simple to build (for me, this meant no tools beyond a circular saw but for more optimal results is more like a table saw and router).
2. The design tools would be free but flexible enough to allow external data import/export.
3. The design would be openly shared but FOR NON-COMMERCIAL USE ONLY.
4. The cost target for the drivers and crossover would be approximately $1000 US.
5. Keep the cabinet size to around 15 liters or about .5 a cubic foot or smaller (allowing it to be reasonably shipped and Amir to test). The electronics would be external. (EDIT)
6. The design would allow for modest flexibility and scalability so others could (eventually) modify and add value. So, here is where Amir intervened with his ideas/goals...
7. The speaker should have good directivity. Note, this did force me to reconsider some of my (mainly cabinet) plans as we'll see later.
8. The speaker should use the new Purifi 6.5 woofer. As it was not in my original plans, I was hesitant due mainly to cost. However, after working on the SPK5, had them in hand and agreed to use...

I guess there will be time and place for "modest flexibility and scalability" - in a new thread. Now Rick is just trying to find best construction and eq, xo and delay settings. When those are settled, different dsp, dacs and amplification can be foud and implemented easily!

 

[Skeptischism]

Sorry, I didnt mean to threadjack, I did worry when answering that question that it was one tangent too far soz, just sorting through a bunch of options and gauging interest as i'm at a fork in my project where I can make allowances for this one and there seemed a bit of convergence between this and the forum streamer project.

 

[hex168]

Hi Rick,

I have a vague recollection of a technique using a woofer as a PR, with a resonant circuit between the otherwise-unconnected woofer terminals. That approach might be able to reduce that 400Hz resonance (or whatever it might be with a different passive radiator).

 

[Rick Sykora]

Hi Rick,

I have a vague recollection of a technique using a woofer as a PR, with a resonant circuit between the otherwise-unconnected woofer terminals. That approach might be able to reduce that 400Hz resonance (or whatever it might be with a different passive radiator).

sounds like something I saw on my Vandersteen 2c speakers. Think he calls it an active coupler. Would be expensive proposition with the Purifi PRs, but may be consideration for the SB.

Will hunt around and see if I can figure out how it was wired. Thanks!

 

[Adam_M]

I've had little success with my purifi PTT6.5's with ports, mind you I didn't try a precision port with massive flares. Yes, monstrous ported bass in the low 30s, but audible port resonance effect every time. I believe it could be sorted by hiding something like THIS (go to port testing section down a bit) in a floor stander, otherwise I'm planning to attempt a transmission line build with it later in the year.

A "plumbers delight" design with the massive precision port flares, paired with the SB BE tweeter in a waveguide is going to be my next build, but there's a workshop redesign between me and that project - probably will start it in the fall. I don't think it'll need to be in a floor stander - or even an abnormally large bookshelf since the volume requirements of the Purifi are so small.

Also, I ran the idea of a TL past Paul Carmody on PE's board and he said while it works, it wasn't anything special - so I'm back to ports, huge flares, and 1/4 wave traps.

 

[hex168]

sounds like something I saw on my Vandersteen 2c speakers. Think he calls it an active coupler. Would be expensive proposition with the Purifi PRs, but may be consideration for the SB.

Will hunt around and see if I can figure out how it was wired. Thanks!

The Vandersteen active coupler is not what I was semi-recalling, but might work in this situation as well. The active coupler is a driven woofer with a very low crossover that is in the same acoustic space as the woofer, and acts as a passive radiator above the very low crossover frequency. As far as I know, it does not specifically target resonance reduction.

The alignment I was trying to describe has a woofer and another driver, complete with magnet structure and voice coil, used in the same acoustic space as a passive radiator. The "passive radiator" is not connected to an amplifier at all. Instead, a resonant circuit is connected between the terminals and supposedly provides a braking force at a selected frequency using the voltage generated at the "passive radiator" terminals. I have never tried it and cannot vouch for the technique, but it sounds like it may be applicable to your use-case if it actually works.

 

[Rick Sykora]

The Vandersteen active coupler is not what I was semi-recalling, but might work in this situation as well. The active coupler is a driven woofer with a very low crossover that is in the same acoustic space as the woofer, and acts as a passive radiator above the very low crossover frequency. As far as I know, it does not specifically target resonance reduction.

The alignment I was trying to describe has a woofer and another driver, complete with magnet structure and voice coil, used in the same acoustic space as a passive radiator. The "passive radiator" is not connected to an amplifier at all. Instead, a resonant circuit is connected between the terminals and supposedly provides a braking force at a selected frequency using the voltage generated at the "passive radiator" terminals. I have never tried it and cannot vouch for the technique, but it sounds like it may be applicable to your use-case if it actually works.

Found the Vandersteen info and you are correct. It might help somewhat as it would push the pr output around its tuning higher. Involves using a high value inductor but I recall not really liking my 2c’s bass all that much....

As for something more passive, creating higher impedance above some cutoff frequency is essentially a low pass filter. Should be able to try on the dual PR cabinet once I get some more woofers.

 

[dansan69]

The alignment I was trying to describe has a woofer and another driver, complete with magnet structure and voice coil, used in the same acoustic space as a passive radiator. The "passive radiator" is not connected to an amplifier at all. Instead, a resonant circuit is connected between the terminals and supposedly provides a braking force at a selected frequency using the voltage generated at the "passive radiator" terminals. I have never tried it and cannot vouch for the technique, but it sounds like it may be applicable to your use-case if it actually works.

What you are describing is what was used in the legendary Sonus faber Extrema speaker (they used a variable amount of resistors across the terminals of a rear mounted racetrack shaped woofer).
Stereophile review, see page 2

Interestingly the SB acoustics oval shaped passive radiator that was discussed earlier in this thread, has an inexpensive counterpart that is an actual woofer (the SB Acoustics SB15SFCR39-8), with a bit of added mass to the cone it could be an interesting choice.

The extrema was recently modernized and reintroduced as the ex3ma https://www.sonusfaber.com/en/products/ex3ma/ still features a similar bass design. (More on internals and design here: https://www.theabsolutesound.com/ar...elebrates-the-30th-anniversary-of-sonus-faber)

 

[Wolf]

The resistor will change (increase) the amount of damping on the PR than without (open circuit). The LCR suggested will apply damping in a small freq range than over the entire bandwidth, and this could kill a resonance in the PR.

 

[Wolf]

A suggestion awhile back was to place a large cap across the terminals, and then the damping would be applied above the working range of the PR and mute higher harmonics.