my tiniest speaker so far

[agnostic1er]

Hello,

Tiny passive 2-way speaker, intended for nearfield listening (up to approx. 7 feet)

1 Woofer: SB acoustics SB12PFCR25-8

1 Tweeter Monacor HT22/8

2 passive radiators Peerless SDS P830878



I named it “aGnome”, a mix of my nickname and “gnome” (french for “dwarf”).

My goal was to create the most affordable standed/bookshelf speaker with an enclosure of H10.12” x W5.75” x D7.24”, so a size smaller than the well known LS3/5A.

With 0.59” panel’s thickness (mdf), Vb is 3.7 liters, so a bit less than 1gal us. In such a tiny volume a vented box isn’t wise. With most of presently available small drivers, we don’t get enough bass from closed box. So I decided to go passive radiators.

Building is very easy, no need of flush mounting the selected drivers.

The real challenge was to create an x-over with as few components as possible for a question of price but also because of the available inside space.

Beside its low price, I selected this tweeter because, being a cone tweeter, it helps align its acoustic center with that of the woofer and smoothes directivity mismatch, a bit like a waveguide. And indeed it works in such a way.

Usually I use VituixCAD with each driver off-axis measurements to create the x-over of my speakers. Not on this one because at the time I didn’t work this way. However, I measured the horizontal (0-90°) and (0/90 and 0/-90°) vertical off-axis with ARTA and got a power response that was satisfactory to my ears. Eq-ing the remaining anomalies did not improve the listening result.

Only 4 components make up the filter.

Listening results: Good tone balance despite light bass and treble missing sparkles. Lack sub-bass. Not very detailed, rather sweet sounding but with natural timbers i.e. beautifull male and female voices. Enough dynamics at safe listening levels in small rooms / nearfield listening. Very good soundscape, width and depth. Matched to a subwoofer, even with a 50Hz low-pass, its global qualities could compete with some floorstanding speakers.

 

[RickS]

Welcome to ASR!

Tell us more about yourself and why you became a member?

 

[Zek]

and why you became a member?

Does it matter why someone became a forum member???

 

[RickS]

Does it matter why someone became a forum member???

It might, but just thought it might be interesting to know. Response is optional.

 

[AnalogSteph]

Ha, the trusty HT22/8 strikes again. One of the few cone tweeters that are actually any good (e.g. the Visatons are a bit of a chore to work with). And it's cheap. Granted, it can't take much of a beating and doesn't have the super low H3 of a dome, but it seems to be keeping up with the 4.5" midwoofer just fine. You can see why Technics/Panasonic liked to use an acoustic lens to broaden horizontal dispersion, mind you.

Is there any particular reason why the drivers are spaced so far apart? I can't imagine it's doing the vertical any good, especially close by. I suppose it's helping the lack of flush mounting, but I'd think some creative sculpting with wood filler would help address that as well if need be.

 

[agnostic1er]

It might, but just thought it might be interesting to know. Response is optional.

Welcome to ASR!

Tell us more about yourself and why you became a member?

Thanks for your welcome.

I'm 67 years old, interested by sound reproduction since I was16, particularly hifi loudspeakers. I began diying speakers in a correct way in the early 2000s with the help of Tony Gee (HumbleHomeMade hifi) and learning a lot with sites like Zaph Audio, Troels Gravesen, RJBaudio, AudioXpress and a lot others. In this regard, I particularly appreciate Zaph / John's always very relevant opinions. It's a shame he's no longer in this passion.

I know of SPINorama / Klippel measurements and ASR / Erin's sites only since some time. Sharing is of great importance in my life. As I am an objectivist more than a subjectivist I decided to become a member of this ASR forum. I'm also supervisor of the french DIY forum of the biggest french audio/home theater site : HCFR.

I built about thirty speakers so far, for me and others.
My biggest making is a three-way using 3 bass-drivers Faital 10pr320-16, 1x Eighteensound 8nmb420-8 and 1x SBacoustics tw29dn-4
95dB/2.83V and, even if mostly PA drivers, a real hifi sound, very detailed and dynamical:

 

[agnostic1er]

Is there any particular reason why the drivers are spaced so far apart? I can't imagine it's doing the vertical any good, especially close by. I suppose it's helping the lack of flush mounting, but I'd think some creative sculpting with wood filler would help address that as well if need be.

Hi,

The lack of flush mounting isn't of great importance since the face plate is only 0.23" thick, so almost no effect on the fr response. The reason of being so close to the top is to minimize diffraction and to shape the fr response as to be able to make the high-pass with the fewest components possible.
I could have made the woofer closer to the tweeter but I prefered this aesthetics. I listen to them at approx. 2/2.30M (7/8 feet), so not a problem about directivity, at least to my ears.

 

[AnalogSteph]

I listen to them at approx. 2/2.30M (7/8 feet), so not a problem about directivity, at least to my ears.

Not an issue at that kind of distance, of course.

I just figured a speaker like this would commonly be used in nearfield (0.5-1 m, e.g. PC desk). It's clearly not a super happy camper at your measured 92 dB / 1 m equiv., with midrange distortion exceeding 3% here and there and bass distortion crossing the 10% line around 130 Hz. If I interpolate between Erin's Kali LP-UNF measurements, I'd say the woofer performance of that is comparable or somewhat better (I won't be surprised if both parts are quite similar in price under the same circumstances), the bass is holding it together about an octave lower in an only marginally bigger enclosure, and there's obviously no contest in the tweeter department (waveguided 1" dome vs. 2.5" cone, doesn't take a rocket scientist to figure out what to expect).

 

[agnostic1er]

@AnalogSteph
I completely agree with the fact that aGnome's woofer isn't close enough to the tweeter for nierfield PC desk listening.
But for the rest, I regard aGnome as a speaker to be compared to a passive equivalent speaker in terms of size and cost, not to an active/DSP speaker. We all know that speaker DIYing isn't able to compete with present Asia made low cost commercial speakers offering. aGnome si more a making which only purpose was to hear if this kind of dwarf is able to give a listening satisfactory.

 

[Heinrich]

... interpolate between Erin's Kali LP-UNF measurements, ...

The devil is in the details. That Kali (vicious Indian goddess type?) is 20dB down at 30Hz, 8dB down at 40Hz. What some won't notice as easily is that the distortion figures relate to way subdued levels compared to the mids from where the reference level is taken. In other words, if, before the distortion test, the frequency response was equalized to flat as it should, those figures would be terrifying. Such small speakers are way way worse than such tests make you believe. Midrange distortion is, taken with a grain of salt, also quite mediocre to even bad.

Also look at multitone distortion. I'm not a fan, because the stress especially in bass doesn't reflect the actual need the least. MD needs to use pre-filtered signals with the lower two octaves raised by 20(!!)dB. That would mimic the spectral content of modern pop/rock music. For EDM add another 10dB. That is not done here, and still, the Kali exhibits like 3..8% of intermodulation at just 87dB even if the bass is cut off at 80Hz. This renders it, for me, unusable. The problem seems to be the small cone area (Doppler) and cheap motor (AM).

This proves exemplary that especially smaller speaker should be designed as three-way. But there is more. The x/over is designed in a way that the lobing suck-out is oriented upwards. Especially with DSP at hand rather the better side could have been tilted upwards, facilitating to stand up without detrimental effects for the tonal balance. Was it just neglected? I think so.

 

[Arindal]

2 passive radiators Peerless SDS P830878

Am I reading it correctly that the passive radiators´ tuning frequency is in the region of 47 Hz? Why so low, was this intentionally? Looks like their output level is pretty low.

Maybe the enclosure is a bit too small to really get the passive radiators into their comfort zone. Due to the high Vas of the neat little SB, a standard reflex design would require significantly more volume to achieve lower bass.

 

[Heinrich]

Am I reading it correctly that the passive radiators´ tuning frequency is in the region of 47 Hz? Why so low, was this intentionally? Looks like their output level is pretty low.

Maybe the enclosure is a bit too small to really get the passive radiators into their comfort zone. Due to the high Vas of the neat little SB, a standard reflex design would require significantly more volume to achieve lower bass.

The comfort zone of a passive radiator is a peculiar formulation. It is part of a machinery. I'm foremost interested in the reasoning on why the tuning is too low. I think it should be 30Hz, in order to allow modern content to be played without unloading the driver below the tuning frequency.

But regarding the enclosure volume you're right. 10..15 liters would be more appropriate, giving a whoopy 10dB more output for the same, limiting, cone excursion (@30Hz).

 

[kyuu]

Does it matter why someone became a forum member???

Is there a problem with asking???

 

[AnalogSteph]

The devil is in the details. That Kali (vicious Indian goddess type?) is 20dB down at 30Hz, 8dB down at 40Hz.

Which, mind you, is still better than the OP's passive speaker.

Midrange distortion is, taken with a grain of salt, also quite mediocre to even bad.

At least it's still average for the size. Which is more than their 6.5" driver can say for itself (or the 8"). The cheap Edifier MR4 (with a reputation for rather meh level handling) pretty much falls apart at 96 dB entirely... but that's a $100 powered speaker set for ya, those parts can't cost much of anything. (Interestingly enough, at 86 dB it still does pretty OK, and arguably better than the LP-UNF outside the bass region.)

The problem seems to be the small cone area (Doppler) and cheap motor (AM).

I don't think Doppler has much to do with it. That tends to be pretty much a non-issue in practice with standard excursion drivers unless your driver is a coax and the cone is doubling as a tweeter waveguide. (Coaxes of the stick-out tweeter variety tend to have better level handling, but come with a whole host of other issues in return. The other approach you'll commonly see is using a coax as a mid-tweeter only.) Cheap motor will definitely be a major factor, maybe with an Al voice coil former thrown in for extra eddy currents. (Which might be why the no-frills MR4 driver looks better in the midrange at 86 dB.) These will be likely to tick the boxes for every single kind of distortion that Purifi drivers try to minimize.

A Neumann KH80 woofer may be a better example of what a decent 4.5" can do, even if that one still has a fair bit of H2 for my liking and clearly isn't up to the same level of refinement as their KH120 woofer. (I think standards and competition in this size class just aren't the same as in the 5.25" class.) Seeing H3 dipping below -60 dB by 1 kHz at 86 dB certainly is more confidence-inspiring than seeing it only be clearing the -40 dB mark. I take it that suspension and L(x) may not be the most linear but electrical nonlinearity seems to be good, so my guess would be Kapton voice coil former and maybe a copper shorting ring.

 

[Arindal]

A Neumann KH80 woofer may be a better example of what a decent 4.5" can do,

Do you have an idea who produces their 4.5" and 5.25"? They claim to use in-house modeled ones, but I think it might be based on something that is freely available.

10..15 liters would be more appropriate,

That is what I would have estimated looking at the woofer´s parameters. Unfortunately it is difficult to find a 4" really capable of producing bass in just 3.7 l, even with the help of two passive radiators. It is technically doable. If you find one with really low Vas. Maybe they are rare because the proper parameters inevitably come with very low efficiency.

I have by chance been listening to a battery-powered Klipsch Detroit bluetooth speaker. Would estimate a 3" or 3.5" max is used in less than 1.2l with two passive radiators per side. That ain't much, but the bass is amazing.

The comfort zone of a passive radiator is a peculiar formulation. It is part of a machinery. I'm foremost interested in the reasoning on why the tuning is too low. I think it should be 30Hz, in order to allow modern content to be played without unloading the driver below the tuning frequency.

Finding a 4" allowing an effective 30Hz reflex tuning is close to impossible. Even if such a driver existed and you were ready to live with low sensitivity, it would still have to make too much of excursion in the 50-100Hz region which is where the energy is in most of recordings.

In my opinion, solely the opposite strategy might work: finding a 4" with low fs and Vas suitable for a small vented enclosure, and tuning the passive radiators to 55...65Hz so they take over the heavy load for the whole band. Such design is challenging without ending up with a boomy ´one-note bass´. I regard it to be high art of developing speakers.

By chance I have some drivers which fit into this scheme, maybe I will try something like that, so the OP can rename his ´aGnome´ to ´Goliath´.

 

[Heinrich]

... tuning the passive radiators to 55...65Hz so they take over ...

You know for sure, that a reflex enclosure unloads below the tuning frequency. Please consider, that the unload will happen many times with contemporary pop music - and let's be honest, that is the program to be played, not the 'classical music' stuff. I've seen relevant, means -6dBFS, peaks a lot recently @30Hz, even @20Hz with more elaborated music.

One might be tempted to adjust the drivers' suspension, the passives' suspension, that won't help either, a port no way. DSP / electronic filtering is mostly out of reach, especially when meant to be specifically tailored for an individual speaker - it would otherwise happily serve as an e/q too ...

Not the least, what for? Such speakers are 'low interest' (very) to the conoissieur due to the inevitable technical limitations. Who else is likely to spend money for a well thought-out design? A single sub is too boomy mostly, two are bothersome.

I experimented in that field myself (https://www.audiosciencereview.com/forum/index.php?threads/sealed-mini-speaker-3-way.61896/)

As you say "for its size" it is fairly successfull. But considering the effort, well, it's an experiment in its own right, which justifies more than just a bit more expenses and work. I was eager to outperform Purify, and, until proven wrong, I assume I did a good job (subjectively, in-shelf mounting, wide dispersion). The industry could help with a proprietary plate amp as to circumvent the hassle with external DSP and amplification.

Finally I think that two aspects need more attention:
- bass extension down below, 30Hz is a mark to reach in-room regardless of size, w/o too much distortion, let alone unloading
- isolate intermodulation (Doppler, AM) from the mids, long excursion is *not* the way to go (Purify, Neumann, Genelec), but 3-way

(see also: https://www.linkwitzlab.com/frontiers.htm#J)

 

[Heinrich]

I don't think Doppler has much to do with it. That tends to be pretty much a non-issue in practice with standard excursion drivers

See the link above, froniers #J. Just 5mm excursion @50Hz generates 10%IM (-20dB) in-room @2kHz. Since Doppler is 'linear', 1%IM (-40dB) would be had with only 0,5mm excursion, shocking. Me thinks that Erin's IM (aka multitone, which has its issues in stressing the bass too little, same with Neuman's MD) is overestimated a bit. Microphones show IM for themselves. Would like to see a cross check. When taking this into account, the IM / Doppler of 3-way is always a lot better than with every 2-way, regardless of the 'quality'.

From Erin's site
Neumann KH80, active 2-way: 3% IM at 86dB in the mids, even with cut-off at 80Hz, hence a subwoofer won't help at all (https://dl.dropboxusercontent.com/s...n-80.png?rlkey=y81nq7yygf8uxp8a6hhfikifx&dl=0)

Wharfdale Super Linton, passive 3-way: 0,3% at 86dB in the mids driven fullrange (20Hz!) with actually good bass extension https://dl.dropboxusercontent.com/s...ey=rtyimk4l1hqjf25vqqevdutga&st=4nx7rhyl&dl=0

My personal threshold for IM is like 0,3% when listening focussed in search for defects, but I would accept 1% when listening to music.

 

[Arindal]

I've seen relevant, means -6dBFS, peaks a lot recently @30Hz, even @20Hz with more elaborated music.

I am not doubting the existence of such frequencies, but I doubt it would be possible to reproduce them substantially at sufficient SPL with any 4"/3.7l passive speaker. So anyways, a compromise is needed. I personally would go for a bass driver suitable for such a small enclosure plus capable passive radiators tuned a bit higher, in the 55-60hz region. There is some very low bass cut off, surely, but the rest might sound pretty convincing as many modern bluetooth speakers of this weight class are proving.

I experimented in that field myself

I remember the thread, and again my thought was: this bass driver requires much more volume. There was another similar project mentioned, employing a much more capable driver:

Mini 3-way (7l). f6 ~40Hz. Low distortion in midrange. Recommended PR PIR slopes. Inexpensive drivers and components. (Mechano325)

The drivers: DSA135-8 link PC83-4 link ND28F-6 link Filter schematics and and charts: THD -47dB or 0.5% (at 86dB) Assembled crossover (3 filters): Component costs: c.a. 280EUR/pair (including drivers) Box dimensions: 290x174x263mm. Only the tweeter is flush-mounted. Midrange driver...

www.audiosciencereview.com

This Dayton DSA135-8 is a good example for an excellent compromise, pretty capable of handling two passive radiators in a 5l enclosure tuned to 50...55Hz.

bass extension down below, 30Hz is a mark to reach in-room regardless of size, w/o too much distortion, let alone unloading

Would rather think this is senseless with such a small speaker, as the combination of driver + volume + passive radiators will inevitably lead to shortcomings below 50Hz. It is nice to see some 30Hz on the graph, but I do not think anyone would really enjoy music typically containing such low beats on that one.

isolate intermodulation (Doppler, AM) from the mids, long excursion is *not* the way to go (Purify, Neumann, Genelec), but 3-way

Nothing against 3-way, but in such a compact design it is rather useless in my opinion. The importance of Doppler and AM is vastly overstated at usual SPL, and midrange IM, being the real problem with a compact, broadband midbass, will kick in suddenly in any driver of this class known to me. A classic 3-way design with higher x-over freq between bass and midrange would not help here, as it requires the woofer to retain clarity below acceptable distortion threshold almost as much as in a 2-way concept.

So the only reasonable concept increasing the usable max SPL in a very compact design w/o the risk of audible distortion is a true ´subwoofer plus 2-way´ concept allowing the subwoofer to run in the red zone while the mid-bass is staying clear above 120Hz. Which in a very compact design will eat up sorely-missed space for the midrange volume.

And I doubt you would find many capable subwoofer drivers for 2.5l and 30Hz. Which midrange and woofer would you suggest fitting into 3.7l combined?

When taking this into account, the IM / Doppler of 3-way is always a lot better than with every 2-way, regardless of the 'quality'.

You are comparing apples and oranges here. The Neumann is a 4" design comparable in size to the ´gnome´ concept, while the Super Linton is listed as a 40l net volume model, hence 10x the cabinet size, employing a true midrange which is bigger than the bass of the Neumann and the Gnome.

I agree that with 40l net volume a true 3-way design makes sense and you can expect a lower level of IM. If you want to have a fair comparison, check the IM of a 40l, 10" 2-way concept.

The industry could help with a proprietary plate amp as to circumvent the hassle with external DSP and amplification.

Hypex or miniDSP could be your friends.

 

[Heinrich]

The importance of Doppler and AM is vastly overstated at usual SPL, and midrange IM, being the real problem with a compact, broadband midbass, will kick in suddenly ...

Overstated because nobody can do something about it? I can

You are comparing apples and oranges here. The Neumann is a 4" design comparable in size to the ´gnome´ concept, while the Super Linton is ...

It is obviously, I thought so, not a matter of size. Bandwidth is the issue. See the formula

As / AH = 0.5 DfH / fL = pi * fH * Xp / c = relative second order distortion

from the link given above (Frontiers #9). It is the relative amplitude of the first sideband-frequencies, the distortion that is, for a given excursion Xp of the modulating freq/ (unspecified), modulated freq/ fH, c = speed of sound. I leave it to you to evaluate the topic besides calculating some numbers.

Only one hint: We halve the size, we halve the cone area, which doubles the excursion, and then we say, whooey, we saved cone size, now we can double the bandwidth of the bass and make it a 2-way?!

So, should we build ever smaller because the market asks for it, losing every quality aspect on the way, or should we solve the accompanied problems on the go?

 

[agnostic1er]

Hello,


Selected passive radiators are adapted to the mid-woofer in this tiny load.
The tuning frequency is 52Hz, not 47, see the attached impedance curve in post1.
Bass is really good sounding and of appropriate level.
Bellow is a 3.3 feet unwindowed measurement , speaker elevated (so no bass reinforcement), 1/6 oct smoothing aGnome vs another vented speaker equipped with a SEAS U18 in 9.3 gal us, Fb 40Hz. Levels are scalled. You can see that the bass range is of a similar level. Evidently aGnome doesn't have the same low-bass level
.