Any interest in an ASR community speaker project?

[dwkdnvr]

Interesting - will have to do some more analysis. The spl_max spreadsheet shows 105dB for the Scan 10f @300Hz, even though the comment in the spreadsheet suggests it's free space. Definitely doesn't agree with your numbers.

 

[Lbstyling]

I think you are 100% on the money here. That 200hz region is where the 'beef' of the sound is, and asking a 3-4" driver to do the job is just going to cut it. This leads to one of two approaches. The first is to use a big midrange, which makes directivity matching to a tweeter difficult. The second is to use a woofer to reproduce the midrange up to 500hz or so, and using a small and agile midrange above. Maintaining good headroom in the 100-400hz region is critical to making a dynamic speaker, and the numbers above bear this out.

These are similar trade off choices of the floor standing Revels. If we don't go for max bass reach as possible via driver design, but instead aim to extend the bottom end with EQ, and expect the owner to invest in a sub to really go super low, this should be very possible with a low budget.

Can always tune a port extra low with a dip between this and the driver that is EQ'ed back up. This reduces the driver excursion, and gives you extension.

My concern is always the 700-7khz zone, so I would steer towards a 3 inch over a 4 inch driver and aim to cover as much of this region as possible with good off axis behaviour.

 

[dwkdnvr]

Keep in mind power compression when dealing with low efficiency drivers. Especially cheap ones with no vent on the pole.

Fair point. I generally assume that I'll be listening 20+ dB lower than these peak numbers, so am generally not concerned about power compression. Probably worth thinking that through, though.

 

[617]

Interesting - will have to do some more analysis. The spl_max spreadsheet shows 105dB for the Scan 10f @300Hz, even though the comment in the spreadsheet suggests it's free space. Definitely doesn't agree with your numbers.

There are any number of sources of error here, I was just trying to get an idea of the relationships involved. Xmax values can be reported in different ways as well.

Now, regarding collaboration tools, I would like to share some spreadsheets, product cutsheets and vituixcad data. I'm going to make a dropbox; if you want to be added to it, please message me your email address.

Regarding software, one of you guys touched on this before, but I'm using excel for all the tabulated data, WinISD for woofer modelling and VituixCAD for everything else. In addition I use Rhinoceros for 3D CAD, but if it comes to the point where we're sending drawings around, DWG/DXF or whatever standard is desired is fine (STL/STEP for 3d printing for example.)

 

[Lbstyling]

Fair point. I generally assume that I'll be listening 20+ dB lower than these peak numbers, so am generally not concerned about power compression. Probably worth thinking that through, though.

If the driver is 84db/1w + 5 inch and you apply 100w, increasing this to 200w usually just gives you a expensive space heater.

I often forget that I'm the weird one playing at 110db on the regular.

 

[dwkdnvr]

If the driver is 84db/1w + 5 inch and you apply 100w, increasing this to 200w usually just gives you a expensive space heater.

I often forget that I'm the weird one playing at 110db on the regular.

Drivers sound better after you let the smoke out - it's part of the break-in process. Really loosens them up.

But, yeah - I don't listen at high volumes for extended periods for the most part. There is definitely a reason there is a segment of the population that gravitates to pro audio setups inspired, though. And, having had the Yorkville U15 Unity horns in a teeny-tiny room for a while, there certainly is something about having effectively unlimited headroom.

 

[Mashcky]

A 3/4 + 3 inch + 6/7 inch would be cheap. Toole puts little importance on distortion profiles of drivers, so this can be sacrificed. + C to C spacing can be kept low and the cabinet can be made aesthetically pleasing.

+1 on the idea of using a small midrange and 3/4 done tweeter. The small tweeter paired with smaller midrange is likely easier to match, as there won’t be any directivity matching in the crossover region. I’m not sure how this would change if a waveguide is used on the tweeter.

 

[dwkdnvr]

Okay, off the cuff driver lineup recommendations using $160/side as a hard limit, and assuming a sealed alignment so limiting the woofers to the previous list.

Lineup 1 ($150)
Peerless 830668 ($60)
Scan-Speak 10F-8414 ($73)
Peerless XT25SC90 ($17)

Lineup 2 ($160 if on sale)
Dayton RS270p ($100)
SB Acoustics SB10PGC ($17)
SB Acoustics SB21SDAC ($37, though currently $25)
(maybe) Augerpro waveguide ($15 allowance)
(or, XT25 again, but trying to mix it up)

Lineup 3 ($135)
Peerless 830668 ($60)
SB Acoustics SB10PGC ($17)
SB Acoustics SB29RNDC ($57)

Budget buster. Added only since this is what I have on hand to prototype my 'main project' ($260, or $410 including PRs)
Dayton RSS265HO-4 ($140)
SB Acoustics SB15NBAC ($67)
Peerless DA25Tx ($53)
2xRSS265PR ($150)

I'm actually really intrigued by Lineup 1, only possible due to the ridiculously low price of the XT25. I originally intended to have the SB29RNDC in lineup 2 but it exceeded the budget. Not sure Lineup 2 really needs the waveguide, but the curves with the waveguide look very good and it might allow lowering the xover to reduce vertical lobing. Plus, you can bring it in on budget if you buy on sale.

My gut reaction based on the above is that if we want to adhere to the $160/side budget, the Peerless 830668 is very attractive, despite needing an LT to hit the desired box size. The RS270p just doesn't leave much in the budget for many options for the M/T, and basically all the other options were even more expensive than that.

 

[dwkdnvr]

Just realized we could add the 'bargain bin special' as well. Under $100

Bargain ($94)
Peerless 830668 ($60)
SB Acoustics SB10PGC ($17)
Peerless XT25SC90 ($17)

 

[617]

So, I'm doing directivity simulations, and I'm having trouble. I'm simulating a 3 way with the rs270p (exact driver doesn't matter but size does), a 25mm tweeter similar to the Peerless BC25 and 3 midranges - one tiny (SB Acoustics 10F) on medium (vifa tc9) and one large (rs125p).

I've equalized each driver in VituixCAD with active filters, and then set up crossovers with LR4. LR4 is nice because it's sharp enough that you see all the discontinuities between drivers. Then, by using the optimizer, VituixCAD can move the crossover points around and find a crossover point which has flat on axis and smooth DI.

I'm finding that none of these options have really stellar DI, and I can't say why. I'm sure it will come to me as I experiment more. One thing I am noticing is that going from a 10" woofer to a 2" midrange like the 10F looks a lot like going from a 6" midwoofer to a 1" tweeter.

Now, LR4 is sort of a worst case scenario since the directivity characteristics of the drivers are not very well blended, but I'd like to see something pretty good with LR4.

A typical result, seen below, with the Vifa TC9, a medium-small midrange:

This isn't really a bad speaker - DSP gets us ruler flat on axis, of course, but that big wave in the DI between 1-5K I would like to smooth out.

My suspicion is two factors need to be addressed:
1. A 10" woofer will only mate with a 2" midrange if you cross so low that they're both almost omni, which doesn't work of course, since you'd want to cross them around 500 or above. So, a really small midrange is probably not suitable for a 10" woofer
2. A 25mm tweeter can blend with a 3" midrange pretty well if crossover point is high enough, but it's looking a bit more like a small waveguide would be really advantageous in the tweeter.

So what is the solution? There are a few ways forward. One is to reduce woofer size to 8". This would smooth the transition to a 3" midrange. However, the 10" woofer sizes offer real value in terms of SPL, so perhaps another solution is to use a larger midrange (125-150mm) and then a small waveguide on the tweeter, or at least a more robust tweeter around 28mm rather than 25mm.

One of the current combinations might work just fine with a second order crossover; I haven't tried that yet, but I have a feeling the design may have to shift towards one of the above compromises in order to have really excellent directivity characteristics.

The precedent I have cited for this speaker is the Revel Gem2, which is an unusual shallow stand mount 3 way. It is a passive design retailing for 10K a pair, and is a fairly conventional design. Aside from the very rounded baffle edges and shallow waveguide on the tweeter, it is a textbook 3 way, with a 1" tweeter, 4" midrange and 8" woofer, with crossover points at 400 and 2.3khz. They are specified as 'high order' which I assume means fourth order (higher order passive networks are rare).

I think it is telling that Revel chose to do both of the compromises I listed above - they made the woofer a bit smaller (8") and made the midrange a bit bigger (4") than a very traditional design, in addition to a small waveguide on the tweeter.

The use of a waveguide represents a branch in the design direction of this speaker, especially as related to budget. Good quality baffle mounted tweeters are a dime a dozen, but off the shelf waveguide designs are rare; and the number of 3d printable waveguides I have at my disposal is somewhat limited. However, there may be some good options.

A sb21SDC can be put in the 3d printed WG seen here. This is a 25 dollar tweeter, which, let's face it, is pretty cheap as tweeters go, and the 3d printing cost I would estimate at $20 per. Still, this is significantly more than a baffle mounted Peerless BC25 at 14 dollars.

 

[mcdn]

Going to an 8” plus 3” plus baffle mounted tweeter without waveguide might get us close. It would also allow a slightly smaller front baffle which might broaden appeal.

 

[dwkdnvr]

I"ll have to start doing something similar. My initial instinct is that there may be some cabinet-diffraction related problems in what is shown. For example, there is no reason I can see for the drop in DI between 1k and 2k - the driver itself should still be basically omni, or close to it a that point. (although most measurements only show response out to 60 degrees, so that might be part of it.)

But, yes - this is why I suspect anything we call an 'ASR' design is almost certainly going to need a waveguide. There is a reason I suggested the XT25 and SB29RDNC in my driver lineups - as ring radiators they generally have higher directivity and should match the mid better which is about the best you can do without a waveguide.

The SB21 w/ augerpros waveguide (or SB26ADC) are well worth looking at, although I couldn't make the SB26ADC work in the $160/side budget. He designed them with an oval profile as a compromise between directivity match and C-to-C spacing.

Oh, and I know you're aiming for a sealed design, but going vented with a 6/7/8" might be something to consider as well - much easier to get the desired output in that arrangement.

 

[617]

I"ll have to start doing something similar. My initial instinct is that there may be some cabinet-diffraction related problems in what is shown. For example, there is no reason I can see for the drop in DI between 1k and 2k - the driver itself should still be basically omni, or close to it a that point. (although most measurements only show response out to 60 degrees, so that might be part of it.)

But, yes - this is why I suspect anything we call an 'ASR' design is almost certainly going to need a waveguide. There is a reason I suggested the XT25 and SB29RDNC in my driver lineups - as ring radiators they generally have higher directivity and should match the mid better which is about the best you can do without a waveguide.

The SB21 w/ augerpros waveguide (or SB26ADC) are well worth looking at, although I couldn't make the SB26ADC work in the $160/side budget. He designed them with an oval profile as a compromise between directivity match and C-to-C spacing.

Oh, and I know you're aiming for a sealed design, but going vented with a 6/7/8" might be something to consider as well - much easier to get the desired output in that arrangement.

I really, really liked the idea of a sealed box, but going 8" vented looks sensible. I'm simulating the 8" woofers I left out of the last comparison to see what they look like in terms of output.

 

[briskly]

I can't do math so I'm not sure if these are equivalent,

The formula I gave corresponds to the peak SPL in half space. Linkwitz's spreadsheet is for the RMS SPL of a sinusoid in full space, Bagby's formula is for the RMS value in half space. Not sure what is going on with WinISD, but it appears to be the RMS value in half space.

 

[617]

OK, some data on 8" drivers. There are probably some good drivers I am missing, but I'm fairly confident that this is a good sampling of high performance, reasonably priced 8" drivers currently on the market. Results surprised me.

First, I didn't disqualify any drivers for requiring too much or being able to handle too little power. What matters ultimately is how much sound they produce, not how much electricity they use. Using a vented alignment, all but one driver was able to meet the SPL threshold at 30hz that I set for the 10" drivers (94db at 2M I think). The levels at 200hz weren't too far off either, with all of them being able to do at least 103db@2M, and most being able to do much more. Additionally, these drivers are generally cheaper, with only one being disqualified for price (the Hivi Lk8*). So, how to pick?

I'd first look at enclosure size; but the range of enclosure sizes is not that wide. Generally when you make a vented box smaller than ideal, you're giving up some output potential, and causing some low end boost above the tuning frequency. However, I changed the tuning frequency of each of these designs so peaking was around 2db, and the box sizes are just not that bad. Depending on the size of the mid enclosure and vent, you could probably get that kind of volume out of a box which is 11.5" W x 21.5" H x 13.55 D (or less).

*The T/S info for this driver is a little weird and lacks some variables. It's clear it is a low efficiency weird driver, and it seems to like a big box. Relatively new on the market as well.

The vented enclosure brings some complexity to the design, but not too much. I might look into a slot port rather than a conventional tube.

As far as which drivers here look best to me, it's hard to say. The ubiquitous RS225 is a solid choice, as is the RS225P. The Peerless Nomex is quite attractive, but has a an odd baffle shape which is a little tricky to recess using DIY tools. The ScanSpeak really wants a big box, but has high efficiency - it would be a great choice in a passive design. The sb23 has a wonderful extended response with a very low Fs, but is a bit expensive compared to the other options.

 

[617]

The formula I gave corresponds to the peak SPL in half space. Linkwitz's spreadsheet is for the RMS SPL of a sinusoid in full space, Bagby's formula is for the RMS value in half space. Not sure what is going on with WinISD, but it appears to be the RMS value in half space.

I've looked at some forums and I think you're right. That's silly, bass drivers should really be considered in full space. So the output is really more like 6db less, but of course at these frequencies the room dominates. The drivers' beef relative to each other is still accurate but the midrange requirements are a lot less stringent in terms of SPL, which is great news.

 

[dwkdnvr]

Looks like you're modeling the RS225-8, I think. The 225-4 models slightly flatter in 1.2 cu ft - higher f3 but similar max spl. My model shows the -4 likes 1.6-1.8 as an optimal volume, but 1.2 works.

Alternate ($157)
RS225-4/8 $60
SB10PGC ($17)
SB26ADC ($60)
waveguide $20

Of course, you could just do a DIY Revel Performa 3 - SB23, SB15NBAC, SB26ADC+Waveguide. Over the $160 budget, but kinda a no-brainer for an 'ASR' project, in a way.

 

[617]

Looks like you're modeling the RS225-8, I think. The 225-4 models slightly flatter in 1.2 cu ft - higher f3 but similar max spl. My model shows the -4 likes 1.6-1.8 as an optimal volume, but 1.2 works.

Alternate ($157)
RS225-4/8 $60
SB10PGC ($17)
SB26ADC ($60)
waveguide $20

Of course, you could just do a DIY Revel Performa 3 - SB23, SB15NBAC, SB26ADC+Waveguide. Over the $160 budget, but kinda a no-brainer for an 'ASR' project, in a way.

I think the SB10 is going to be too small. Revel is using a 4" midrange and I think something like the RS125p, Vifa NE123 or something is going to end up working better. Let's have a look for bargain 4" fullranges, I'm sure there are tons I'm missing.

So maybe one 'high end' option is ($234)
SB23 - $110
NE123 - $63
SB26STAC - $41 (I think the soft dome performs better in the WG, we can discuss this)
waveguide - $20

But a better value would be ($139):
RS225/p/4/8 - $60
RS125- $34
SB21RDC -$25
Waveguide - $20

I honestly don't think the more expensive option would be any better here.

The midrange is the biggest unknown. There are a very limited number of tested waveguide designs, for under $50 we have the RST28 (WG too big) and the sb26/sb21. The CSS LD22 is a great unit but overpriced at $75. 8" woofers we've narrowed down pretty well - and the old medley's musing's page ranked the sb23 and rs225 as being very close to each other in distortion, so their performance differences have more to do with bandwidth at both ends. I strongly suspect the 3.5-4.5" class of midranges is going to match best with this waveguide, and so called 'fullrange' drivers are very plentiful these days, so this would be the best place for more research.

 

[zermak]

Looking to see were this project goes and maybe one day make the speakers myself.

About the midranger/full ranger (even woofers), may I suggest Faital/Faital Pro products?

 

[617]

Final driver ranking of the night. Here are 8 and 4 ohm variants of the most viable 8 inchers. Each was modeled in an 1.2 cubic foot enclosure and tuned for maximum output at 30hz. 1.2 cubic feet is about right for some, a bit small for a bunch, and a bit big for the Tang Band.

Note the differences here are much smaller. The SB23 NBAC is really made for deep extension in a biggish box rather than high output in a small one. It doesn't do badly but it's not impressive for the price. The RS224p-4 ohm variant and the sb23 NRXS paper cone have a good combination of output and price. The 225-8 and 225-4 demonstrate why they are used in like 20 DIY designs, lagging only a bit in output at mid-band. The SB23 MFC are 8 inch subwoofers which do pretty well - the 8 ohm variant is probably the best best for SPL, but at 130 you're paying over twice the cost of a Dayton Reference for 5db.

I would say the best value option is the RS225p-4, which I give additional points due to its nice treble extension. If you want to spend the money, the sb23 MFCL is a little beast and justifies its cost better than any other option here. The Tang Band looks seriously impressive, with a magnet the size of a mini cooper, and would be a good direction to go in if cabinet volume is to be minimized.

I also tested, for fun, the Dayton "Epique" ECF220 or whatever it's called. It's a funny driver, giving you huge output in a small box but with very little bass extension. It's an unusual combination of features and would be an interesting driver to use, but the price on sale is $300, which is a lot to pay for that performance.