Small 2-way speakers with linear on-axis and power response characteristics (Scan Speak and SB Acoustics drivers). H&V off-axis measurements included

[XMechanik]

I've been trying to make design of a small, simple-to-build and inexpensive speaker that would not need a sub at least for music listening at moderate spl levels. No need for high power. This is third approach, code name "Mechano23". I'm sharing the results and VituixCad project file (LINK) with off-axis measurements in both planes so anyone could play and experiment.

Drivers used in the design:

• Scan Speak H2606/920000 LINK
• SB Acoustics SB13PFCR25-4 LINK

The box is really simple: rectangular panels (18mm birch plywood) glued with overlaps. No detachable panel, the filters are mounted to the rear panel through the woofer mounting hole.
External box dimensions HxWxD: 290x174x263mm, volume approx. 8 liters. Rear bass-reflex, tuned around 45Hz (fi=35mm d=145mm).
Panel dimensions:

• Front and rear: 174x290mm.
• Side: 227x290mm.
• Top and bottom: 138x227mm.

There is a reinforcement inside that connects the centers of the side panels (piece of plywood 135x40x18mm).

The box is filled with polyester fiber. To prevent it getting into the speaker basket, I put a thin mesh on the filling.

I made measurements -180..+180 degrees (V orbit) and 0..180 degrees (H orbit) with a step of 10 degrees (VituixCad vxp file with full set of measurements: LINK). I used a manual turntable.

Filter design. The software apparently needed a 2uF capacitor which is not on the E12 list. It was assembled connecting two values of E12 in series: 2.7uF and 6.8uF.

Spl on-axis and power response characteristics with the target lines (simulation).

Filters transferred to VituixCad and slightly modified (series-connected 2.7uF and 6.8uF capacitors were replaced with a single 2uF cap.).

L and R filters assembled on plywood scraps. Wiring diagram.

L and R spl measurement (f6~=48Hz with reference to the average level)

L and R impedance measurement (minimum 4.1R at 220Hz)

List of components:

Inductors:

• L1 330 uH 0.258 R
• L2 560 uH 0.51 R
• L3 1.5 mH 0.94 R
• L4 390 uH 0.42 R

Capacitors:

• C1 6.8 uF
• C2 27 uF
• C3 18 uF
• C4 820 nF
• C5 3.3 uF
• C6 2 uF

Resistors:

• R1 12 R
• R2 8.2 R
• R3 15 R

Bass-reflex tube: Monacor MBR-35 (145mm)

 

[pma]

Nice job! At what frequency do you merge near field and far field SPL measurements?

 

[abdo123]

Hey! Could you please provide how much polyfil did you exactly put in their (weight) and provide things like baffle thickness, material thickness and more overall dimensions? Thank you!

 

[XMechanik]

Nice job! At what frequency do you merge near field and far field SPL measurements?

When I was merging off-axis measurements with near field the frequency was 300Hz and on the system measurement it was 200Hz (not much difference). I've neglected difraction response on the low frequency nearfield measurement as I assumed that for such a small enclosure it shouldn't matter much. I'd expect an error of a single dB.

 

[XMechanik]

Hey! Could you please provide how much polyfil did you exactly put in their (weight) and provide things like baffle thickness, material thickness and more overall dimensions? Thank you!

I didn't weigh it directly but according to my estimations 60-70g is needed to fill 8 liters moderately. Area next to the end of the BR pipe is not filled.
All panels are 18mm thick and the external dimensions are 290x174x260mm (HxWxD).

 

[abdo123]

I didn't weigh it directly but according to my estimations 60-70g is needed to fill 8 liters moderately. Area next to the end of the BR pipe is not filled.
All panels are 18mm thick and the external dimensions are 290x174x260mm (HxWxD).

If you could edit the baffle layout to include detailed cutout dimensions I would greatly appreciate it.

 

[Raxumit]

Looks like it measures well. Is this a sealed design?

 

[XMechanik]

Looks like it measures well. Is this a sealed design?

No, it's a bass reflex design. The port is on the rear panel.

 

[XMechanik]

If you could edit the baffle layout to include detailed cutout dimensions I would greatly appreciate it.

front and rear panel: 174x290mm
side panels: 227x290mm
top and bottom panel: 138x227mm

tweeter mounting hole diameter fi=72mm (limited by the magnet size, see datasheet: LINK))
woofer mounting hole diameter fi=122mm (see datasheet: LINK)
(add few mm free space for both)

flush depth is 4.9mm for the woofer and 4mm for the tweeter.
tweeter flush radius r=52mm
woofer flush radius r=75.25mm
(as on the front panel layout)

 

[Raxumit]

No, it's a bass reflex design. The port is on the rear panel.

Thanks.

I had a look at Mechano22, and you don't document the port size and length there either as far as I can tell.

If it's something off-the-shelf, easy enough to document, yeah?

 

[XMechanik]

The enclosures for Mechao22 and Mechano23 are basically the same, and BR tuning is the same too (could be few mm port length difference but that makes almost no change). The port used is Monacor MBR-35 adjusted to 140mm. It wouldn't be bad to use a bit wider port for Mechano23 (keeping 45Hz tuning). The driver SB13PFCR25-4 has 13 in name but it's actually a 15cm driver.

 

[tktran303]

When I was merging off-axis measurements with near field the frequency was 300Hz and on the system measurement it was 200Hz (not much difference). I've neglected difraction response on the low frequency nearfield measurement as I assumed that for such a small enclosure it shouldn't matter much. I'd expect an error of a single dB.

Did you combine the nearfield of the woofer and port and use the Merger tool to calculate baffle diffraction loss from the rectangular baffle?

A smaller baffle causes MORE diffraction loss, not less.

This is a common blind spot amongst designers and reviewers alike.

Happy to help if needed. Just start a conversation and I shall advise.

 

[XMechanik]

Basing on the general assumption that diffraction effects can be neglected if the object dimensions are much smaller comparing to the wavelength, the diffraction response on the NF was omitted. The baffle dimensions are 17.5cm and 29cm and for the far field measurements gating frequency was c.a. 200Hz which is equivalent to 170cm wavelength. Was the assumption justified in this case? On the Far Field measurement side the directivity next to the transition frequency is 1.2dB, so the response is not purely omnidirectional here but it's pretty close to this.
I am thinking of a new design with a slightly larger enclosure and this time the diffraction response is going to be taken into account with the near-field LF . I would certainly appreciate any comments/hints/suggestions on this topic.

 

[jaakkopetteri]

Nice build, wouldn't have expected directivity that smooth with such a small waveguide. Thanks for the VCAD files too!

 

[XMechanik]

I am trying to determine the distortion characteristics for Mechano23.
I've never measured distortions before, the measurement system I use (CLIO Pocket that is) does not measure distortions with a sweep. I'm trying to deal with the task using FFT analysis.

First, I made some standard spl measurements to adjust the volume knob to get the desired level (95dB).

Then I ran a frequency-by-frequency FFT analysis geting chart like this for every step.

I read the harmonic components up to and including the fifth (if they were in the 20 kHz range), below is a summary.

Now I'm considering two options of calculating THD. In the first case I take the fundamental signal component from the FFT analysis, and here's the result:

In the second case I put 95dB everywhere and I get the following characteristic:

I'm not sure which option gives lower error. I think it's the second because in the first case, reading of the fundamental and harmonic components is overlapped with room modes. In the second case, it only applies to the harmonics as the level fundamental signal is well approximated with anechoic spl sweep (95dB +-1.25dB). What do you think?

 

[tktran303]

1. What a nuisance that Clio Pocket doesn't come with automated distortion testing with MLS or sine sweeps. Have you considered switching to ARTA or REW?

2. 95dB may be close to the limit of that 5.25” midwoofer, hence the distortion high. How does the 85 or 90dB look?

3. THD, which is most commonly dominated by H2 or H3, is poorly correlated with subjective sound quality. Makes for an simple graph, I feel there’s little to be learnt by measuring it. I suspect this has to do with interchanging levels of H3 H4 and H5 ((grey, yellow, blue lines wiggling up and down crossing) depending on the non-linearities of the transducer and H2 and H3 taking turns dominating the THD.

A very good midrange driver looks like this:

I would use it between 200Hz and 2KHz.

(2.83V drive level; measured at 50cm (closer distance than 1m to increase signal to noise floor level)

 

[Gringoaudio1]

Amazing in every way! The response is outstanding. The wood work is lovely. What drove the crossover design choices? The tank circuit C4 and L2 especially in the Vituix image? And the RLC ? A complex design to say the least. All I have ever done is first and second order with simple voltage dividers to attenuate. Series caps to the tweeter and maybe an inductor to ground. The opposite for woofer: series inductor and maybe a capacitor to ground. Your thinking and use of circuit elements is way beyond me.
And even with all these parts your inductance curve goes pretty high. No way to flatten the curve?

 

[tktran303]

He's the designer of XMachina, a great automatic passive crossover designer. Does the AI even know why it does what it does?

About XMachina

XMachina is a passive crossover designer. XMachina works unassisted. Just tell XMachina your goals, load your drivers characteristics and ...

xmachina-ai.blogspot.com

 

[XMechanik]

1. What a nuisance that Clio Pocket doesn't come with automated distortion testing with MLS or sine sweeps. Have you considered switching to ARTA or REW?

2. 95dB may be close to the limit of that 5.25” midwoofer, hence the distortion high. How does the 85 or 90dB look?

3. THD, which is most commonly dominated by H2 or H3, is poorly correlated with subjective sound quality. Makes for an simple graph, I feel there’s little to be learnt by measuring it. I suspect this has to do with interchanging levels of H3 H4 and H5 ((grey, yellow, blue lines wiggling up and down crossing) depending on the non-linearities of the transducer and H2 and H3 taking turns dominating the THD.

A very good midrange driver looks like this:

I would use it between 200Hz and 2KHz.
View attachment 286670


(2.83V drive level; measured at 50cm (closer distance than 1m to increase signal to noise floor level)

I haven't been considering it yet. I've heard good opinions about these programs but, correct me if I'm wrong, using them involves things like jigs, sound card compatibility, and an extra microphone (probably my current CLIO mic wouldn't work with a typical sound card as it requires power supply).
Actually one of the goals of this exercise was to let me check how well I can perform the distortion measurement using the FFT and if it doesn't go well, I'll think about what to do next. Well, in fact it didn't work out very well, I have an impression that both characteristics that I presented have a large degree of error.

Regarding the spl level, yes 85dB would be a better reference for typical listening conditions. At 95dB, the sound was so horribly loud that I had to use hearing protectors during the measurement session, as if I was working with a router. The problem with the THD/FFT measurement is that it's not easy to repeat. Depending on the resolution on the frequency axis, several or several dozen separate measurements must be made.

 

[XMechanik]

Amazing in every way! The response is outstanding. The wood work is lovely. What drove the crossover design choices? The tank circuit C4 and L2 especially in the Vituix image? And the RLC ? A complex design to say the least. All I have ever done is first and second order with simple voltage dividers to attenuate. Series caps to the tweeter and maybe an inductor to ground. The opposite for woofer: series inductor and maybe a capacitor to ground. Your thinking and use of circuit elements is way beyond me.
And even with all these parts your inductance curve goes pretty high. No way to flatten the curve?

Thank you, it's reaaly nice to hear your positive feedback.
C4 L2 is a woofer breakup resonance trap above 7kHz.
Next branch (C2, C3, L3, R2) is a series-parallel resonant circuit that helps to equalize the woofer respone in the range from beffle step to the crossover frequency.
There is a way to flatten the impedance but I think there is no need. It would add several more components to the network that is already not so minimalist. My feeling is that this impedance is pretty easy for most of contemporary amps.