• 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!

Any interest in an ASR community speaker project?

GelbeMusik

Senior Member
Joined
Apr 19, 2020
Messages
445
Likes
290
As more amplification is comparatively expensive Vs changing drivers to higher efficiency, especially in a cubic volume limited design, we need to be shooting for drivers that are this or higher.

Common saying is that the efficiency is determined by volume and targeted low frequency extension. There are estimates for the required (avg/max) level over frequency:

https://www.hifi-selbstbau.de/images/stories/grundlagen/waveana/TrackAllAvg.png

Full page ( German only ):

https://www.hifi-selbstbau.de/grund...6/81-musik-qvergleichenq-mit-dem-waveanalyzer

The low frequency demand for amp pwr could be estimated like a 50Hz/12dB filter applied, which mimics the spectrum of real life content. There are exceptions like the spoken wisdom piece Usher / Yeah ...

Discarding that wisdom, the low frequency isn't that critical anymore when it comes to available watts. So, yeah, go for higher efficiency.
 

Lbstyling

Addicted to Fun and Learning
Forum Donor
Joined
Jan 2, 2019
Messages
500
Likes
461
There are exceptions like the spoken wisdom piece Usher / Yeah .

Am Interested in this exception. Are these specific examples that don't fit the mold? Do we know why?
 

dwkdnvr

Senior Member
Joined
Nov 2, 2018
Messages
418
Likes
698
Not terribly exciting, but I got my AcePc mini-pc set up with Equalizer Apo outputting over HDMI to my spare NAD T748V2, implementing a basic 2-way crossover. Aside from the NAD eating the first ~0.1 sec of the stream, it seems to be OK so far - I'm a bit concerned that this might interfere with measurements, but I haven't tried it yet. Most of my 'real' measurements will be from the laptop and a separate amp, but measuring over the HDMI link will be needed for REW and room correction filters. I think REW can sent a pre-signal chirp which might be useful.

This is my intended prototyping platform for this project and the other speaker measurement/experimentation I have going on. Next step is to look into exporting filters from VituixCad and see how much processing they need to be usable.

The only glitch is that HDMI audio out isn't really designed as a general purpose audio output, so you have to disable the screensaver

The AcePc is a J3455 w/ 6GB and a 120GB SSD, for $220 off amazon. I'd been eyeing something like this for a while, but finally used this as an excuse to pick one up. A bit of fan noise that I might have to figure out, but otherwise seems ok - no crapware which really surprised me. Win 10 Pro as well, although I'm not quite sure why you need it on this type of device- can't use RDP if you're using HDMI out which would be the obvious benefit.
 

HammerSandwich

Major Contributor
Forum Donor
Joined
Nov 22, 2018
Messages
1,137
Likes
1,497
Woofer Comparison, Mk 1

First, I'm not sure about the W250S. Both -4 & -8 versions show ~2dB more sensitivity in Vcad than on the spec sheet or Dibirama's -8 test. I haven't found a reason yet, but $68 says it might be worth looking more. If Vcad's correct, the top end is loud but LF a little low. That's the classic tradeoff between sensitivity & extension, so it could be close. But maybe the low bass level leaves other woofers more interesting?

Speaking of Dibirama, it's a good resource, and I'll refer to it again.

As before, all SPL plots are done by VituixCAD for 2pi, 1M & 16Vrms.

Let's start with the SB23MFCL45-4 (red, 252cc Vd) in 17L (Qtc 0.577) as the baseline from my earlier post. This woofer has fantastic low end for 8" in a small box, but the "HF" SPL is a bit low at almost 104dB.

The HW251N (cyan, 197cc) uses 23L for 0.707. It's 2dB louder up top but weaker below 50Hz. No further discussion here, because...

...the RS270P-4A (black, 208cc) matches it exactly in only 15L and $30 less! "Exactly?" you ask. "But the curves don't match!" That's because the chart shows the 270P's SPL in 23L (Qtc 0.61). At 0.707, you'd see one line, so here's another look at what builders can adjust.

Finally for this round, the 830668 (blue, 288c) in 28L with a Qtc of 0.9(!). This woofer wants a bigger box and really needs more amp here. This woofer could move more air than any of the others, as shown by the Xmax limit (straight line). If we used it correctly. However, you can easily see that the amp/box comb can't use the last 4dB of excursion. Good woofer for a different project, but the SB23's a better pick for a small box and/or amp.

IMO, the RS270P looks like the winner here (good call, @617!). Look to the SB23 if you need a smaller box.

Woofer Comparison, Mk 1.png
 

GelbeMusik

Senior Member
Joined
Apr 19, 2020
Messages
445
Likes
290
Woofer Comparison, Mk 1

Do You really keep the box size variable? Better make it fix. It will determine, if the speaker fits ones (and her) needs. For an 8"er 25..30 liters are in order; neither to small, dB per watt wise, nor bites too much of precious volume out of Your Manhattan loft or less humble living space.

Bassreflex would spare about half of the excursion for same level and extension. A rule of thumb again. Or the other way round: closed would need twice the volume due to doubled cone surface area, or double the excursion, which latter hardly can be achieved, though. I think that more than +/-4mm is a dream still to come true ( really?! ).

Then one might go through the lists and select the most efficient drivers. As said, a nominal +/- 5..6mm of x-max is the best to expect. A driver with strong motor would work in mostly all volumes without getting out of hand. It would make the driver appear bass-shy on paper, sure.

But: the absolute bass level per watt is the same as with the weaker driver, only that the midrange is stronger! That's why the bass of strong drivers seems underrepresented in the usual simulations, aka transfer function.

But: the higher efficiency in the midrange would spare, say 3/4 of amp power. The EQ doesn't harm. because the weaker and the stronger driver have the same efficiency down below. The latter is determined by the volume alone (nearly). It is not so, that the stronger needs to be pushed. It is toned down in the mids instead!

Since usual music program shows kind of a cut off at around 50Hz @ 12dB, the bass becomes less important. So the requirement for "wattage" is roughly halved with the more efficient driver. Depends on how You weigh the different frequencies in.
 
Last edited:

GelbeMusik

Senior Member
Joined
Apr 19, 2020
Messages
445
Likes
290
Do You really keep the box size variable?

I recently realized a bass guitar speaker using a "midrange" speaker--for bass. This exemplifies on the theory above.

The signal of bass guitar is quite similar to what a hifi-speaker would eat ( why ;-). The maximum level goes down towards 50..40Hz by around 6dB. With a ported design the strongest demand of cone excursion lies around 60Hz.

Three different drivers were simulated:
- original RCF 8MBG200
- the same but with softer suspension,
and
- again with softer suspension and some 70% weaker magnet.

The T/S parameters were adopted accordingly.

The first is that typical "midbass" from which nobody expects much of bass. The second specimen, how it would be, if the first was softer; the resonance frequency was changed from about 75Hz to below 35Hz. But Qts landed at some 0,17 (o-one-seven). Lets give it a rise, so that we eventually, with the third have the typical hifi-ish bass/mid driver with fs = 35Hz / Qt = 0,35.

Transfer function in 25 liter box, tuned to 35Hz:

1591799686449.png

You take the red one, sure. Green and blue are -6dB @ 100Hz, what a shame!

Now EQ'ed for (nearly, +/-0,3dB) identical transfer functions, and the input wattage adjusted, that all would deliver the same sound level, plus some filter applied, which reflects the actual program content (peak values); here's the "apparent load power":

1591799903222.png

The red one is actually an overeater. It takes double the power for the same output. Of course that's the hifi-ish driver. The discussion of green versus blue becomes a little complicated. The instruments attack comes into play. When the string is plugged, the overtones are much higher in level, than the base note. On records this dynamic 'moment' is not preserved … no accords on bass guitar … . So, in the context of bass guitar blue wins hands down. That was the original "midrange". They shake the house now ..
 
Last edited:

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,690
Likes
6,013
Location
Berlin, Germany
But: the absolute bass level per watt is the same as with the weaker driver, only that the midrange is stronger! That's why the bass of strong drivers seems underrepresented in the usual simulations, aka transfer function.
Finally someone that got the strong-drivers-have-no-bass fallacy. Per-dissipated-watt output always is higher with the stronger driver, but voltage-based transfer seems to be bass-shy. EQ it, done. There is of course the fact that a very strong driver produces much more microphonic voltage ("back EMF") which the amp has to overcome, sort of, thats why amp with strong drivers need some more voltage headroom compared to the midrange, but that headroom directly relates to more SPL output as well.
 

GelbeMusik

Senior Member
Joined
Apr 19, 2020
Messages
445
Likes
290
... but voltage-based transfer seems to be bass-shy. EQ it, done. There is of course the fact that a very strong driver produces much more microphonic voltage ("back EMF") which the amp has to overcome, ...

I think the EMF is represented alone by the complex impedance, the phase angles and such. So the question may boil down to: could the amp deliver its max voltage even into the complex impedance around base resonance? I never tried … :) But since it is all low frequency, I trust the feedback mechanisms. Maybe there is a current limit, though … at the capacitive part, I guess, to the right left side(s) of impedance peaks. That could spoil my fun with the blue line, because that alone has some "power" needs. But again these occur at the saddle in the impedance curve, hence these feed into a mostly "real" load.

If my point of view is screwed up, please let me know.
 
Last edited:
OP
617

617

Major Contributor
Forum Donor
Joined
Mar 18, 2019
Messages
2,404
Likes
5,296
Location
Somerville, MA
Woofer Comparison, Mk 1

First, I'm not sure about the W250S. Both -4 & -8 versions show ~2dB more sensitivity in Vcad than on the spec sheet or Dibirama's -8 test. I haven't found a reason yet, but $68 says it might be worth looking more. If Vcad's correct, the top end is loud but LF a little low. That's the classic tradeoff between sensitivity & extension, so it could be close. But maybe the low bass level leaves other woofers more interesting?

Speaking of Dibirama, it's a good resource, and I'll refer to it again.

As before, all SPL plots are done by VituixCAD for 2pi, 1M & 16Vrms.

Let's start with the SB23MFCL45-4 (red, 252cc Vd) in 17L (Qtc 0.577) as the baseline from my earlier post. This woofer has fantastic low end for 8" in a small box, but the "HF" SPL is a bit low at almost 104dB.

The HW251N (cyan, 197cc) uses 23L for 0.707. It's 2dB louder up top but weaker below 50Hz. No further discussion here, because...

...the RS270P-4A (black, 208cc) matches it exactly in only 15L and $30 less! "Exactly?" you ask. "But the curves don't match!" That's because the chart shows the 270P's SPL in 23L (Qtc 0.61). At 0.707, you'd see one line, so here's another look at what builders can adjust.

Finally for this round, the 830668 (blue, 288c) in 28L with a Qtc of 0.9(!). This woofer wants a bigger box and really needs more amp here. This woofer could move more air than any of the others, as shown by the Xmax limit (straight line). If we used it correctly. However, you can easily see that the amp/box comb can't use the last 4dB of excursion. Good woofer for a different project, but the SB23's a better pick for a small box and/or amp.

IMO, the RS270P looks like the winner here (good call, @617!). Look to the SB23 if you need a smaller box.

View attachment 68133

For the purposes of directivity matching, I want to use an 8 inch speaker. Can you do a similar evaluation of the rs225p, sb23 and so on?
 
OP
617

617

Major Contributor
Forum Donor
Joined
Mar 18, 2019
Messages
2,404
Likes
5,296
Location
Somerville, MA
In progress.

I haven't done all the simulations but a 10 inch woofer wants a 5 inch mid which in turn wants a pretty biggish waveguide. The 8 works with a 4 inch mid and a smaller waveguide.
 

HammerSandwich

Major Contributor
Forum Donor
Joined
Nov 22, 2018
Messages
1,137
Likes
1,497
That surprises me a bit. With ~300Hz XO, does woofer size really affect the polars that much?
 

HammerSandwich

Major Contributor
Forum Donor
Joined
Nov 22, 2018
Messages
1,137
Likes
1,497
While I'm still putting together another batch of charts, here are a few comments on drivers that probably won't get plotted. Not that they're poor drivers, but they don't seem like good fits for this project.

22W/8534 - Close to RS225 but wants a bigger box and more money. Compare those 2 plus 830869 at Dibirama. RS has slight advantage in THD, while the Peerless wins on sensitivity.

26W/8534 - Vb & Qtc mirror the 830668, yet the Peerless has 37% more Vd at 1/2 the price. IOW, same as previous entry in 10" size. Dibirama has both and shows Scanspeak ahead in distortion.

BTW, we should lean to 4Ω drivers to best use the 16V amps.

SB29s - Both have lowish sensitivity and lean toward larger boxes.

RS270-4 - Compared to the 270P, the heavier, aluminum cone costs SPL above 100Hz. LF performance is identical, and box sizes are very close. I'd spend the extra $7 for the 270P.

RSS210s - They're very much modern subs for small boxes, which means very low efficiency. Peak SPL is only around 100dB. They do put big Vd in little boxes, though! FWIW, Dibirama shows moderately high THD for RSS210HO-4. In comparison, the 315HFA certainly justifies its larger size.

SB23NBAC - Poor ratio of Vd to cost. Tends to larger boxes.

Comments welcome.
 

GelbeMusik

Senior Member
Joined
Apr 19, 2020
Messages
445
Likes
290
Comments welcome.

So, You are aiming for an as small as can be top-box loaded with an 8 inch driver closed. Plus midrange sized about 2 to 4 inch complemented by maybe an 3/4 inch tweeter loaded by a waveguide of DIY making (3D print).

This again shall be complemented by a subwoofer, that is planned to reside under the small box above, as kind of a stand.

All shall be driven by a low power plate amp alike comprising substantial DSP.

Estimated cost, I think will be about 1200$ (or more, but not less) plus some considerable expenses for esthetics. The range of necessary skills would span from somehow precision, bulky woodworking, over soldering along some basic electrical understanding, doing the paint job, 3D printing, up to programming the DSP along own measurements using some qualified microphone and the programs, which again would ask for a basic, but real understanding of acoustics and signal theory. Some extra space is required for the dirt jobs, while the rest shall be done in the living room, I assume.

If it cannot be finished, the loss is at least half of the initial expenses, the woodworking tools not included.

I tempted to headline this "spoiler", but then I realized that the points given are simply matter of fact.

I first expected this to become a humble undertaking. Eight inch plus 4 inch plus 3/4 inch, all baffle mounted--no waveguide. As to achieve a wide dispersion along the Toole concept of stereo. The 8 is set due to the minimum capability required with bass volume and extension. Reflexed of course, as to aim for a physical, satisfactory result departing from the emotionally sterile "critical listening" attempt of audiophoolery.
The rest falls into its place automatically. Cost would be about a considerable 500 for a pair, again plus esthetics. Driven by a low cost miniDSP or (!) PC / tablet based EQ and some two channel amp for cheap.

If it fails for an individual, at least the DSP would have been worth the effort. No high fidelity without mike and EQ anymore!

If I took the standpoint of a less experiecend, or even first time DIYer, I would not buy the first idea. Alone because of the less convincing subwoofer stand ... . Second, maybe, but it would still need some talking.

Nevertheless, I wish You all the best success with whatever You do.
 

DeruDog

Member
Forum Donor
Joined
May 13, 2020
Messages
68
Likes
65
Location
Philadelphia, PA, USA
If you are saying that the Hypex amps are cheaper than the Daytons, you're looking at the wrong Daytons. We're looking at the little TPA3116 amps - I'm guessing you may be looking at their high-end DSP plate amps. In that case, I tend to agree that the Daytons are a bit expensive for what they bring. The little TPA3116 amps are interesting though, and quite inexpensive. e.g. https://www.parts-express.com/dayto...ass-d-audio-amplifier-board-with-dsp--325-126

The Hypex amps are certainly interesting, and could certainly be used to execute the type of design we're working through here. They'd run about $750 for a pair of the cheapest 3-way units though, which would raise the barrier to entry for this project beyond what we're really trying to target. So, the reason we're looking at the Dayton amps as the 'default' recommendation is simply since they look to be in keeping with the desire to keep the overall project budget somewhere in the $500/pr range.

Having said that, the Hypex Fusion amps really don't seem to be any more expensive than the raw amp modules, so you almost get the DSP capabilty for free. Certainly an option that should be looked at for a 'high end' or 'final' build.

I WAS looking at the wrong Daytons, thanks!

As a matter of fact, a few months ago I was looking for a class D solution with DSP like these Daytons, which is how I found out about Hypex. Not sure if they are new, or if my search for an internal amp was incomplete, but I I am glad to see these now. I am very interested in building some active speakers.
 
OP
617

617

Major Contributor
Forum Donor
Joined
Mar 18, 2019
Messages
2,404
Likes
5,296
Location
Somerville, MA
That surprises me a bit. With ~300Hz XO, does woofer size really affect the polars that much?

For the degree of off axis smoothness I am attempting to achieve, yes. I was surprised too, which is why I went with a 10 inch driver initially. Revel uses an 8 and 4 in their gem speaker and similar close spacing of driver sizes in their other product lines. If you want a truly flat DI with direct radiators, this is the sacrifice.

So one data point is Revel's success (and the fact they publish spinoramas for their speakers) The other is simulations using VituixCad, which I think are fairly accurate for baffle mounted drivers at those lower frequencies. Both bear out the use of an 8 inch.

Having said that, a 10 inch plus 5+ inch speaker is not a horrible idea, it wouldn't be that much bigger if sealed. I will run some simulations today to see if the DI is smooth in that configuration. The problem here is that you need a bigger waveguide, but 5 inches isn't insane or anything.

Designers normally worry about off axis smoothness in the upper frequencies but if you want to flatten out that first hump you need drivers of similar size.

As a side note, it's really amazing how well optimized the Dayton woofers are. Whenever I do simulations for boxes they somehow end up at the top of the list. I couldn't tell you why but they always seem like great options.
 
OP
617

617

Major Contributor
Forum Donor
Joined
Mar 18, 2019
Messages
2,404
Likes
5,296
Location
Somerville, MA
1591883239357.png


This is a simulation with a 270mm and a 150mm driver crossed at 300hz. Not bad, but at 150 hz:
1591883292203.png

That's pretty good (these simulations don't have a tweeter)

If we go to a LR2 instead of LR4:
1591883355415.png


Doesn't make a huge difference, but what if we do LR2 at 300hz?
1591883410664.png

Not a big difference really. How about LR2 at 600hz?

1591883457141.png

Even worse.

This sort of summarizes the issues. From a traditional speaker design perspective, where power handling and flat axial frequency response are most important, the big woofer+small mid works fine, within the limits set by the beef factor of the midrange unit. Traditionally very small midranges have been used with big woofers (JBL L100) and it's not uncommon to see midranges which are basically big dome tweeters (Spendor BC-1). This approach yields 'wide dispersion' as well, but if you want truly smooth DI, you need to make sacrifices.

For truly flat DI in a smallish 3 way speaker, you can examine the sacrifices inherent to each woofer size.

1. A 6-7" woofer with a 3" midrange and a dome tweeter.
2. An 8" woofer with a 4-4.5" midrange and a dome tweeter in a small waveguide
3. A 10" woofer with a 5-6" midrange and a dome tweeter in a big waveguide

#1 is a bit of a silly design; you could just use a big waveguide and eliminate the midrange. This isn't a bad design, but the output capabilities of midwoofers are never very impressive. I've done a speaker similar to this with a 52mm dome and a 17cm woofer. It's quite pleasant but poorly optimized.

#2 is an interesting compromise, still being a smallish speaker, especially if sealed, and with 3d printing we have good options for a small waveguide.

#3 is viable if sealed, but a 10" speaker becomes pretty big if vented, and a waveguide which is 5-6"+ is bigger, and top end dispersion becomes a challenge as well. Augerpro's waveguides do give us some options here.

Option #2 seems like the most logical to me given the modest constraints of the Amp/DSP modules which make this project fun and cheap, but #3 is not a bad idea either. Another option is to scrap the midrange and make an 8" two way, which is sort of a lifelong dream of mine.
 

Attachments

  • 1591883392506.png
    1591883392506.png
    36.8 KB · Views: 78
  • 1591883398652.png
    1591883398652.png
    36.8 KB · Views: 82

dwkdnvr

Senior Member
Joined
Nov 2, 2018
Messages
418
Likes
698
As a side note, it's really amazing how well optimized the Dayton woofers are. Whenever I do simulations for boxes they somehow end up at the top of the list. I couldn't tell you why but they always seem like great options.

Yeah, there's a reason the Dayton Reference series is in a bazillion designs out there. Not the absolute best drivers, but they are very good, relatively inexpensive, and their parameters work really well. Plus, they came out at the right time.

I think SB Acoustics has finally started to provide an alternative in that space - even better measurements, not much more expensive, and available worldwide - that was maybe the one downside of the Dayton drivers - they never set up any international distribution, and so everything had to be shipped from Ohio.

Interestingly, I have been looking at the Grimm LS1 lately. My ultimate project is to do something like a Sonus Faber Ellipsa clone, but the LS1 may be an opportunity to do a much simpler wide-baffle setup for some testing/prototyping. As far as I can tell, they're using the W22EX sealed and as far as I can tell it is rather comparable to the RS225 - mid-60's F3 in ~0.6 cu ft, 5/6mm xmax. These limitations don't prevent Grimm from claiming '105dB' output. If a sealed 8" driver is good enough for a $10+ speaker, hopefully it'll be good enough for the first cut of this design.
 
OP
617

617

Major Contributor
Forum Donor
Joined
Mar 18, 2019
Messages
2,404
Likes
5,296
Location
Somerville, MA
Yeah, there's a reason the Dayton Reference series is in a bazillion designs out there. Not the absolute best drivers, but they are very good, relatively inexpensive, and their parameters work really well. Plus, they came out at the right time.

I think SB Acoustics has finally started to provide an alternative in that space - even better measurements, not much more expensive, and available worldwide - that was maybe the one downside of the Dayton drivers - they never set up any international distribution, and so everything had to be shipped from Ohio.

Interestingly, I have been looking at the Grimm LS1 lately. My ultimate project is to do something like a Sonus Faber Ellipsa clone, but the LS1 may be an opportunity to do a much simpler wide-baffle setup for some testing/prototyping. As far as I can tell, they're using the W22EX sealed and as far as I can tell it is rather comparable to the RS225 - mid-60's F3 in ~0.6 cu ft, 5/6mm xmax. These limitations don't prevent Grimm from claiming '105dB' output. If a sealed 8" driver is good enough for a $10+ speaker, hopefully it'll be good enough for the first cut of this design.

I think that's the crux of it - an 8" woofer is just a great size. Sd makes a huge difference. There's probably some overlap with the absolute best 7" and the absolute worst 8", but in my testing, even a monster 7" like the Anarchy 708 couldn't compete with any 8" woofer in my lineup.

The wide baffle design of the Ellipsa or Stradivari is very interesting and I would recommend looking at Troels Gravesen's writings on that approach. The desire for the woofer/mid crossover point be at baffle step is probably the main motivator for such designs, and perhaps less relevant in the world of DSP, but it's still worth investigating.
 
Top Bottom