SAL451 - a small 2-way loudspeaker

[gornir]

Hi everyone,

After having wandered around many audio forums and websites for years, I thought it would be my turn to contribute and share one of my loudspeaker designs to the DIY community

Who am I?
My name is Göran, the Swedish version of the name “George”. To earn my living I work as an IT-engineer. As you can imagine my main hobby is loudspeaker building. In fact it has been my hobby for more than 25 years now…… oops that didn’t sound right, I’m getting old

For the last 2-3 years I have been more of an “active” than “passive” loudspeaker designer and have been using the DEQX HDP-3 for my active/hybrid designs (DEQX High Definition Audio). The DEQX is a wonderful tool to use for “active” designs. However this approach isn’t the same challenge as a “passive” loudspeaker design. That’s what triggers me to continue to build “passive” loudspeakers and there are in my opinion something’s that the DEQX doesn’t handle to well e.g. BSC (Baffle Step Compensation) etc.

Okay, enough of me and down to business……

The design background:
SAL 451 (SEAS Aluminium 4.5” woofer and 1” tweeter) is a small 2-way loudspeaker I’m building for use in our bedroom (I haven’t told my wife about the placement yet ). I will let you follow the design and construction process and you can expect a lot of pictures and measurement details. I will not give you a lot of the woodworking details, which I think are the less interesting part of the construction. I will however give the details of the box enclosure itself.

Tools for construction and design:
I will use the following tools for my design:

• Dayton WT3 tester for impedance and T/S parameter measurements. (Dayton Audio - WT3 Woofer Tester)
• HolmImpulse with M-Audio Audiophile 24/96 soundcard for frequency measurements. (HOLM Acoustics)
• LspCAD 6 Pro/5.25 for crossover design and simulation. (www.ijdata.com)
• EMM-8cal and MP-1r calibrated mic and pre-amp. (www.ibf-akustik.de)

Deliverables:
I intend to deliver and share the following:

• Pictures of the finished loudspeaker
• Loudspeaker enclosure design schematics.
• Loudspeaker unit T/S and impedance measurements.
• Frequency measurements at 0, 15, 30 and 45 degrees of individual driver’s response in box.
• Simulated frequency measurements at 0, 15, 30 and 45 degrees for the system.
• Real world frequency measurements at 0, 15, 30 and 45 degrees for the completed system.
• Cross-over schematics.
• Listening impressions of the finished design and comparison to a very well known DIY design ( I will not reveal which one yet )
• Probably a lot more and of course my thoughts and comments during the design process.

Since English obviously isn’t my native language, I ask you not to come down too hard on me during the design process. I fully respect that there are many different design philosophies in the DIY area, but hey… I’m designing this loudspeaker for me and sharing it with you.

I hope you will enjoy the trip……..

The selection of driver units:
This one is easy, since I already had the drivers lying around from an earlier active loudspeaker experiment. The Tweeter is well known, have good reputation and used in many DIY projects. The small woofer isn’t that common in DIY designs, but is also very well built and is about 75% of the performance of the SEAS Excel version, but at 25% of the price. These are not high-end drivers, but they deliver respectable performance for its price. I have used a lot of different SEAS drivers during the years and have always appreciated their build quality and consistency. SEAS are in my opinion also one of the most honest manufacturer, regarding spec sheets etc.

Woofer:
SEAS H1207 L12RCY/P - build week/year 20/06
Tweeter:
SEAS H1212 27TBFC/G - build week/year 16/07

Design goals:
Initially I’m aiming for the following design goals:

• Bassreflex tuning around 60-70Hz with the option for closed.
• BSC around 4-5db.
• Crossover point around 2000Hz.
• Crossover slopes 24db/octave, acoustically.
• Optimized frequency response at a 15-30 degree angle.
• Keep the crossover as simple as possible.


To be continued…………

[gornir]

T/S and impedance measurements:
OK, time for some T/S parameter measurements on the SEAS L12. I’m using the WT3 tester for this. Before I have used an older Clio measurement tool for this, but I’ve found that the WT3 is faster, easier and more reliable to use.

The woofers are burnt in for 24h before I measure them and I use the “added mass method” for calculating the Vas and cone weight etc. The added weights are in fact “tape rolls” of different sizes that have calculated the weight of on a precision scale. This trick has always worked for me, but I’m sure the other smart ways of doing it.

As can been seen in the first picture the measurement are in very close agreement with the manufacturer data sheet. Usually this is not the case, but as sad before I think SEAS is one of the more reliable manufacturers regarding to spec sheets.

The impedance charts shows the free-air measurement of the two driver samples. There is nothing strange going on here except a small wrinkle at 9000 Hz, which is expected due to cone resonances at that frequency (this will be described later on in the frequency measurements section).

Feel free to comment

To be continued……

[gornir]

T/S and impedance measurements:
What about the Tweeters? I haven’t done any T/S measurements on them, but of course some impedance measurements. The SEAS H1212 - 27TBFC/G is a well known tweeter and others like John K at Zaph|Audio and Mark K at Mark K's Speaker Pages has done frequency and distortion testing on them. They think the tweeter is a performance leader in its price range, as do I. I realize that SEAS drivers aren’t as cheap as before in the US due to the dollar drop, but here in Europe they aren’t so expensive. I bought the tweeters for about 35$ each and the woofers for about 58$ each ex. Tax. In my opinion that’s not expensive for high quality drivers.

Since I trust the measurements done by John and Mark, I haven’t done any distortion test on the tweeters, but my goal of a crossover frequency around 2000 Hz shouldn’t be any problem, regarding the distortion aspect.

As can be seen on the impedance charts the impedance peak around Fs isn’t’ that sharp and high because of the use magnetic fluid. There is also a second smaller bump between 1-2 kHz, but that’s not likely to cause any problems. The Fs is around 550 Hz as indicated by the impedance plots and conforms well to the SEAS data sheet. I don’t think I will have to deal with any impendence peaks in the cross-over design. Time will tell.

Comments appreciated

To be continued……..

[gornir]

Box tuning:
A textbook design would suggest a box size of around 1.5 litres bass-reflex and about 1 litre for a closed box. This is obviously too small to get any bass extension. I choose a box size of 4.5 litres which is a bit larger than the textbook design, but a size I know from previous experiments work quite well with the SEAS L12 driver.

The first picture shows a box tuning of 65 Hz in a 4.5 litre enclosure. The second picture shows a 4.5 litre closed enclosure with an Fb of 75 Hz.

[gornir]

Enclosure design and construction:
When designing an enclosure there are several factors to consider like box volume, baffle width, height and depth, shape and material thickness etc. In my case I choose to slaughter an existing commercial loudspeaker and refit it with a new baffle. With this approach I was stuck with the only option to decide the placement of the driver units on the baffle. As noted on the construction pictures there are no round-over on the baffle which will affect the frequency response due to baffle diffraction more severely. This baffle diffraction will be considered for in the cross-over design.

The baffle size is 310x150 mm and the depth of the box is 208 mm. The enclosure is made of 19 mm (3/4”) MDF with 4 mm bitumen pads on all internal walls. This gives a total wall thickness of 23 mm and it passes the “knuckle” test with flying colours. Sheep wool is placed behind the woofer, some egg foam on the top wall above the tweeter and the bass-reflex tunnel is placed on the backside behind the tweeter. It’s very important to chamfer the inside baffle hole for the woofer to let it “breath” properly. The tweeter is aligned 20 mm from the top of the baffle and both drivers are flush mounted with 10 mm distance between them.

Picture 1: Box drawing.
Picture 2: The attached baffle with filler applied.
Picture 3: Primed with auto primer and ready for painting.
Picture 4: The finished result.

To be continued……..

[gornir]

Frequency measurements:
Frequency measurements are made under following conditions:

• The enclosure is elevated from the floor by 1.25 meters at tweeter height (halfway between floor and ceiling).
• Both left and right loudspeakers are measured.
• Measurements are done at 1 meter on tweeter axis with a 2.0 Volt MLS signal.
• Measurements are done with drivers connected simultaneous, tweeter only and woofer only for a total of three measurements for each axis.
• Measurements are done at 0, 15, 30 and 45 degree angle.
• Near field measurements are done at 25cm and 1cm on woofer axis for the woofer.
• Measurements are gated and frequency and phase response are exported to a LspCAD readable format.
• The gated 1m measurements are valid down to approximately 270 Hz.

Picture 1 shows the effect of the baffle step. The blue frequency plot show a -3db baffle step around 750Hz, which correlates well with the 15cm baffle width. The huge cone break-up at 9000 Hz has to be corrected with a notch filter in the filter design.

Picture 2 shows both tweeter and woofer driven simultaneous. I import and use this reference frequency plot in LspCAD in order to get accurate acoustical phase response between drivers. In this case the woofer is 25,1 mm “behind” the tweeter acoustically, measured at 1m on tweeter axis.

Picture 3 shows both left and right woofers measured at 1m on tweeter axis. They match each other almost perfectly.

Picture 4 shows both left and right tweeter measured at 1m on tweeter axis. The left tweeter “blue” has a bit more rising response above 17000 Hz. Besides that they match each other almost perfectly. This small variation between drivers will not affect the filter design.

Picture 5 shows left tweeter measured at 1m on tweeter axis at 0, 15 and 30 degrees. Here can the baffle diffraction clearly be seen due to the baffle width, the sharp edges of the baffle and the tweeter location on the baffle. There is a dip in the frequency response between 3500-4500 Hz. Note that the diffraction smoothes out at >15 degrees. This will be accounted for in the filter design.


Picture1: Green - Woofer near field at 1cm. Blue - Woofer at 25cm.
Picture2: Both drivers driven simultaneous at 1 meter 0 degrees.
Picture3: Left + right woofer at 1 meter 0 degrees.
Picture4: Left + right tweeter at 1 meter 0 degrees.
Picture5: Tweeter at 1 meter 0, 15 and 30 degrees.

To be continued......

[gornir]

Frequency measurements:
Picture 1 to 4 shows frequency measurements on the woofer at 1m on tweeter axis at 0, 15, 30 and 45 degrees.

Picture1: 0 degrees.
Picture2: 15 degrees.
Picture3: 30 degrees.
Picture4: 45 degrees.

To be continued…….

[gornir]

Frequency measurements:
Picture 1 to 4 shows frequency measurements on the tweeter at 1m on tweeter axis at 0, 15, 30 and 45 degrees.

Picture1: 0 degrees.
Picture2: 15 degrees.
Picture3: 30 degrees.
Picture4: 45 degrees.

Next, filter design and simulation…….

[jcake5]

Hi Gornir,

Thank you for the build. I have a pair of L12s waiting for a tweeter and crossover and look forward to your results.

Kind regards,
Eric

[gornir]

Simulated woofer cross-over and notch filter:
Ok, there has been a long time since the last post, but other stuff and a terrible cold came in between.

First thing I want to do in the filter design is to tame the woofers terrible cone breakup at 9 kHz and above. It’s absolutely essential do this; otherwise it will ring like church bell. Try to hook-up the woofer to a signal generator and tune in at around 9 kHz and your ears will melt

Fortunately there is a solution for this, a notch filter. The notch filter consists of a simple cap and an inductor in series of each other (see picture1). The inductors resistance is a part of the notch filter. It was quite hard to find the best component values to get rid of the sharp breakup and it’s absolutely critical to use the exact component values shown, otherwise the notch will not have the desired effect.

Picture2 shows the simulated effect of the notch. Observe that a 12db/oct electrical filter is in place to achieve the desired 24db/oct acoustical roll-off at the targeted cross-over frequency of around 2000 Hz (see picture3). I have verified with a signal generator that the notch works as planned. No more church bell…..

Picture2 also shows the simulated “Baffle Step Compensation” for the design. Here the large inductor (2.0 mH) is the major part of the BSC and the compensation is around 4-5db.

Picture1: Notch filter.
Picture2: Simulated effect of the notch filter, woofer cross-over and BSC
Picture3: Simulated woofer cross-over 0 degrees.
Picture4: Simulated woofer cross-over 15 degrees.
Picture5: Simulated woofer cross-over 30 degrees.
Picture6: Simulated woofer cross-over 45 degrees.


To be continued……..