Any interest in an ASR community speaker project?

[KSTR]

I think the ultimate example of what you're describing is the Duelund xover desig

Yes, the Duelund is one example of where all three bands have the same phase response but it is somewhat impractical as the slopes are only 2nd order. The concept of tracking phases can be expanded, though, to any number of bands and any orders and any phase offsets. I'll try to show this, soon (when time permis) for one example were it really makes a difference: the targets for a 2.5-way speaker. There most designers make the error of just rolling off the midwoofer path to get the function for the woofer and that spoils everything. The solution here is to add an allpass function that mimics the phase of the additional woofer roll-off for the midwoofer and then everthing falls in place, phase-wise. The cost is the larger overall phase response (which is on the order of a similar true 3-way), but when desired, the global phase can always be unrolled by DSP-pre-EQ (the simplest and best way to get a linear phase speaker at effectively zero cost).

 

[Lbstyling]

What about this other BMR driver (Link to Spec Sheet at Parts Express)? It's capable of a few more SPLs, plays a touch deeper and if going active, might be able to handle the wiggle/dip at 1900Hz. I haven't seen any in-depth analysis on this specific model like we have on the other (thanks @hardisj !) but might be easier to integrate into the overall design goals presented thus far.

Unfortunately, because its a dip not a peak, and it shows in the impedance graph, you may not be able EQ it up without causing audible problems.

 

[617]

I know, I know, it feels like just jumping on the bandwagon, but if it measures well, fits well, and is cheap...... this is ASR after all. Measurements *should* come first, no?

If we want the unique factor, has anyone used a BMR in a WG yet? It may allow crossover to a larger bass driver easier.

XT25SC90 is a great driver for the money, no question. Could prob go to a 19mm tweeter with the BMR, as this may give a wider DI up top? Depends how it matches I suppose. Must be careful of output and testing with small tweeters though, I have blown a few in my time.

I can't get over how cheap the mids are. They should be $100+ each IMO.

$100 a side would be quite something all in, but I don't know the anarchy drivers myself. Will check them out.

If the design ended up at this kind of price point, It could be nice if it happened to be possible to make a pair from a single 8x4 sheet of MDF or PLY, including bracing.

I could see a prism cabinet working well, but it will depend on how hard it would be to construct. Video tutorials are the way to go for this stuff IMO.

The issue of constructability from a single sheet of ply is interesting. I think it could be done from a single sheet. When I design speakers I deliberately use dimensions which are a bit less than the length/width of a sheet divided by some number. For example - don't design it 25 inches tall - design it 24. Or 15" wide (15x3 = 45", 3" less than 4'.)

I have no interest in facets, for one thing they have to be huge to have much effect, and also I would like this speaker to be constructable by more people, including the fine folks at speakerhardware.com, who can make custom cabinets at very reasonable cost. Planing facets into plywood sounds like a nightmare to me.

In regards to the xt25 or whatever, there's a lot of rational behind all the driver size choices, and I've discovered that even a big (28mm) tweeter can't mate well with a 125mm midrange (not to the standards we're applying to this project anyway) and we can't use a smaller midrange because it won't in turn mate well with an 8" woofer. So the driver requirements are in some ways restrictive;

The woofer should be 8", and should primarily be chosen based on how much bass it can produce in a sealed box with boost filters and so on. A spreadsheet I developed filters the list of all 8" drivers of moderate cost down to only a few which are suitable.

The mid has the fewest restrictions, but should have as much headroom as possible and a smooth response. The ~120mm size is big enough that it works well with the 8" woofer (not much size difference).

The tweeter is where we are very restricted; there are only a handful of tweeter waveguide designs I would consider reliable enough, and the sb21 in a 4" waveguide is the best option. The sb26 and rst28 are other good options but have bigger waveguides. So, the sb21 is the best thing going, and it's only 25 dollars.

 

[Lbstyling]

The issue of constructability from a single sheet of ply is interesting. I think it could be done from a single sheet. When I design speakers I deliberately use dimensions which are a bit less than the length/width of a sheet divided by some number. For example - don't design it 25 inches tall - design it 24. Or 15" wide (15x3 = 45", 3" less than 4'.)

I have no interest in facets, for one thing they have to be huge to have much effect, and also I would like this speaker to be constructable by more people, including the fine folks at speakerhardware.com, who can make custom cabinets at very reasonable cost. Planing facets into plywood sounds like a nightmare to me.

In regards to the xt25 or whatever, there's a lot of rational behind all the driver size choices, and I've discovered that even a big (28mm) tweeter can't mate well with a 125mm midrange (not to the standards we're applying to this project anyway) and we can't use a smaller midrange because it won't in turn mate well with an 8" woofer. So the driver requirements are in some ways restrictive;

The woofer should be 8", and should primarily be chosen based on how much bass it can produce in a sealed box with boost filters and so on. A spreadsheet I developed filters the list of all 8" drivers of moderate cost down to only a few which are suitable.

The mid has the fewest restrictions, but should have as much headroom as possible and a smooth response. The ~120mm size is big enough that it works well with the 8" woofer (not much size difference).

The tweeter is where we are very restricted; there are only a handful of tweeter waveguide designs I would consider reliable enough, and the sb21 in a 4" waveguide is the best option. The sb26 and rst28 are other good options but have bigger waveguides. So, the sb21 is the best thing going, and it's only 25 dollars.

It will come down to off axis measurements I think. The BMR may be limited to crossing to a 7" bass unit, (unless used in a WG). Tweeter quality matters less with a BMR, more with a 120mm mid. BMR speaker would be less likely to need a WG on the tweeter. 7" vented bass would likely play lower with lower distortion than a 8" sealed. Pick your poison I suppose!

 

[617]

Waveguide in question. The original had an eliptical shape which is difficult for a normal person with a router to recess. Flange is quite thick at 1/4". Holes are for #8 pan head wood screws.

 

[dwkdnvr]

Waveguide in question. The original had an eliptical shape which is difficult for a normal person with a router to recess. Flange is quite thick at 1/4". Holes are for #8 pan head wood screws.

NICE! Is that in Fusion360?

 

[617]

NICE! Is that in Fusion360?

Rhinoceros. I could never figure out fusion360.

 

[dwkdnvr]

Rhinoceros. I could never figure out fusion360.

Cool. Once you get to a 'final' STL I can try to machine it.

 

[617]

Cool. Once you get to a 'final' STL I can try to machine it.

If you want to test the tweeters/mids together I can send you a 3D print (along with any mids you want to test). Right now I can make a crude box with a circular saw and track, but I don't have my tablesaw.

One great feature of VCad:

It has a little calculator to calculate the reflection free time in your measurement setup. I'm going to do some measurements on my other speaker project outside to get my technique dialed in, and I think when I stick my two platforms together it's 3175mm off the ground (my mic stand will go on top of a table and hopefully get high enough.) According to this I will be able to get gated measurements down to 55hz, although as Erin points out resolution is quite poor there (1 data point per octave or something). Still that is good enough to accurately splice near fields in, which is prone to error if you are measuring indoors and are trying to splice the nearfield measurements at like 200 hz.

Hopefully it won't rain.

 

[dwkdnvr]

If you want to test the tweeters/mids together I can send you a 3D print (along with any mids you want to test). Right now I can make a crude box with a circular saw and track, but I don't have my tablesaw.

.

Just need the STL - can either print it directly, or else CNC it.

 

[617]

Just need the STL - can either print it directly, or else CNC it.

Well, the backside has some little nubs to align the tweeter properly, so 3d printing is probably more practical. I'll send some stl files tonight.

What sort of cnc do you have?

 

[TimW]

I find SLS printing to be the only process to give satisfactorily smooth results. You can smooth ABS prints with vapor but if I want something 3D printed to be accurate and good looking I just pay for an SLS print. https://craftcloud3d.com/ is a good site for finding printing services.

 

[HooStat]

FWIW, there is a review of the NHT C3 Carbon Fiber that seems to be built in the spirit of this discussion (3 way, sealed monitor). There is some discussion elsewhere on ASR, but the review is here (with measurements). https://www.soundstageaccess.com/index.php/equipment-reviews/1035-nht-c3-carbon-fiber-loudspeakers

 

[HooStat]

It even has the angled baffle that was briefly discussed. But smaller woofer.

 

[617]

I find SLS printing to be the only process to give satisfactorily smooth results. You can smooth ABS prints with vapor but if I want something 3D printed to be accurate and good looking I just pay for an SLS print. https://craftcloud3d.com/ is a good site for finding printing services.

I'm willing to make a friendly bet that a PLA waveguide measures the same as a SLS print. I won't argue it will look as nice though.

 

[somebodyelse]

I find SLS printing to be the only process to give satisfactorily smooth results. You can smooth ABS prints with vapor but if I want something 3D printed to be accurate and good looking I just pay for an SLS print. https://craftcloud3d.com/ is a good site for finding printing services.

A couple of friends have Prusa i3s, and the finish is quite presentable, even close up. The variable z step in the slicer certainly helps the appearance of radiused parts, and can be used with other printers too.

 

[Lbstyling]

A couple of friends have Prusa i3s, and the finish is quite presentable, even close up. The variable z step in the slicer certainly helps the appearance of radiused parts, and can be used with other printers too.

I run a farm of various printers. The Prusa is good enough to just give a 2 light coat blow over in acrylic paint and it looks like a cast part. Print in a close shade to the final colour you want though.

Design the part hollow and upside down, with 2mm wall and you can fill it with polyester resin. This is better than setting the fill higher.


Works better than a vapour bath

 

[617]

I run a farm of various printers. The Prusa is good enough to just give a 2 light coat blow over in acrylic paint and it looks like a cast part. Print in a close shade to the final colour you want though.

Design the part hollow and upside down, with 2mm wall and you can fill it with polyester resin. This is better than setting the fill higher.


Works better than a vapour bath

So you would recommend the waveguide surface be printed facing down rather than up?

 

[Certainkindoffool]

I would donate some coin to development of this project.

 

[dwkdnvr]

So you would recommend the waveguide surface be printed facing down rather than up?

Ha, I literally just posted this question to Augerpro's thread on diyaudio. The construction of his waveguides with the alignment nubs being the only points of contact if printed mouth-up seems problematic - either you need significant support structure to help, or you're asking your printer to print an unsupported horizontal surface. I'm guessing that's just asking for trouble.