Making of M2

By Man-Sun Huh
21 Feb 2000

Printed from SpeakerBuilding.com, 12 Mar 2010 11:39
URL: http:///content/diy/1022/

System Introduction

This time I will show the procedure to make M2 and I hope It will be helpful to who want to build one's own Hi-End loudspeaker.

1st step: System concept & Unit Selection

Revelator has no ferrofluid in the gap between voice coil and magnet. Ferrofluid is used to dampen resonance at natural frequency and to cool the heat of voice coil. So, In general, tweeter with ferrofluid has higher power handling and requires simple X-over network. And almost every hi-fi tweeter has ferrofluid in the magnet gap. But It can be harmful to sound quality. Scan-speak knows it well and make their flag ship tweeter without ferrofluid. Scan-speak suggests second order or higher filter at no lower than 2.5kHz as well as an LRC resonance conjugate network for Revelator. If M2 were a commercial design, I might have followed their suggestion. But I used second order filter around 1.65kHz without an LRC resonance conjugate. (I already remarked the benefits of low crossover frequency at an article of M1) Up to now, two pairs of M2 are working well more than half year without breaking a tweeter.

Woofer is a SEAS Excel W17E002. Of course It's a one of highly regarded 6.5inch woofer. I chose this unit because It seems to produce very detailed sound. And It looks great!! It uses magnesium as a diaphragm. It's material is quite different with Scan-speak 18W8545 woofer which uses pulp as a diaphragm. Sound of SEAS is somewhat different with Scan-speak. SEAS has more detailed sound in mid-high, Whereas Scan-speak has more open sound in mid-bass.

2nd step: Measurement of Unit characteristics

Serial #	40/96-029	40/96-030	40/96-031	28/96-137	28/96-138	Meas. AVG.	Spec. sheet
Fo(Hz)	33.33	32.44	35.05	32.98	33.11	33.41	34
Qms	2.73	2.50	2.77	2.78	2.72	2.65	2.4
Qes	0.39	0.36	0.39	0.40	0.40	0.39	0.37
Qts	0.34	0.31	0.34	0.35	0.35	0.34	0.32
Vas(Liter)	30.9					30.9	29.5

On the contrary, measurement of thiele-small parameter of tweeter unit is not so critical for the design of loudspeaker.

3rd step: Cabinet design

I simulated many kind of bass alignment and decided to go to bass reflex system. I prefer acoustic suspension system but SEAS woofer can't provide enough bass with acoustic suspension design. I selected Bessel alignment and It requires enclosure volume of 12.57 liter and duct tuning frequency of 36.4Hz.(QB3 Alignment has almost the same result.) Fig.2 shows the enclosure design Fig. 2. Enclosure design plan.

4th step: Measurement

So, good measurement of each unit installed in cabinet is very important. One have to measure both Impedance and Sound pressure level. Measurement of Impedance is quite simple for measurement program such as LMS, Clio, MLSSA... Fig.3 and Fig.4 shows measurement result of woofer and tweeter Fig. 3. Impedance measurement of woofer unit Fig. 4. Impedance measurement of tweeter unit.

Impedance can show us many information about enclosure and unit. It's like X-ray for man.

In Fig.3 we can see that duct tuning frequency is 38Hz (dip point between two peak at low frequency). Besides duct tuning frequency, one can see defeats of system. We can see some small ripples at 212Hz, 732Hz, 1367Hz, and 3960Hz. After some other experiments, I found the reason of each ripple.

212Hz ripple:

It's a resonance frequency of woofer frame. It doesn't happen when I measured a unit alone. It happens when installed on cabinet and It even can be felt by touching fingers to the frame. Excel woofer have only 4 screw holes on frame and frame is not well damped. I think SEAS have to consider to make more holes on the frame. To dampen that resonance, I attached damping materials to the back frame(See strips of Dynamat in Fig. 5. Treatment for woofer frame resonance) After that treatment it disappeared clearly.

732Hz ripple:

It is the cavity resonance of 230mm(depth of cabinet) and clearly disappeared after I stuffed sound absorbing material(Dacron). Almost any inner cavity resonance can be eliminated with sound absorbing material.

1,367Hz ripple:

The cause of this ripple is not clear but also disappeared after I stuffed sound absorbing material.

3,960Hz ripple:

It's the cone-break up resonance frequency of woofer diaphragm. If one want to use metal diaphragm unit, this always happens. It's because metal has low damping. To dampen that resonance I designed parallel LRC filter in Woofer crossover network.

Measurement of sound pressure level is not so easy. One have to measure in anechoic condition. I guess few of person have seen anechoic room. It's quite expensive room. Furthermore all anechoic room have limitation at low frequency because of the size of wedge.(Around 50Hz)

Then how do I measure anechoic response? I have a perfect anechoic room with free of charge. It's a large automobile parking lot near my house. And I use ground plane method for measurement. Fig.6 shows the method Fig. 6. Ground plane method. It measures 6dB higher because sound projected area is half the sphere. So, you can simply subtract -6dB to get exact value. But to measure outdoor measurement, you should have battery powered measurement system. I used notebook and battery powered amplifier to drive loudspeaker.

Fig. 7. SPL of woofer (W/O X-over) and Fig. 8. SPL of tweeter (W/O X-over) show sound pressure level of woofer and tweeter. In Woofer response, you can see so called 6dB loss at low frequency(around 300Hz). It happens because speaker baffle has limited area. Low frequency sound wave project to 4PI area while high frequency sound wave project to 2PI area. We have to consider this effect when designing a low pass crossover network and this is why almost every textbook formula of low pass filter doesn't work in real loudspeaker design. So, If you want to make real hi-fi loudspeaker, you have to use program which can import measured sound pressure and Impedance data (LEAP, CLIO, CALSOD, ...).

5th step: Crossover network design

At first, I selected acoustically 4th order Linkwitz-Riley X-over around 1,800Hz. But during optimization I changed X-over frequency to 1,650Hz.

Then what is good x-over? Listed below is my criterion of good x-over.

• Flat sound pressure response over full frequency range.
  
• Steep dip at x-over frequency when tweeter polarity is reversed.(It means that phase matching at x-over point is good.)
  
• Minimum Impedance phase angle difference over full frequency range.(less than +-45 degree is recommended)
  
• Too low Impedance must be avoided.(You can break your amp!!)
  
• The fewer x-over element, the better
  
• High quality capacitor for tweeter network.
  
• 1% tolerance resistor is recommended.
  

To meet all those criterion, I simulated more than 1 month for optimization. Fig. 9. The X-over network is the final schematic diagram of X-over. And Fig. 10. Picture of the X-over network shows photos of x-over.

6th step: Final test & fine tune

Till now, I didn't feel any fine tuning for x-over. Instead, It's sound seems to be quite dependent on the sound absorbing material. I have used simple polyester(used for pillow) and Dacron. I think Dacron is much better. Some people have recommended me to use poly wools but still I didn't tested those material yet. I recommend you to adjust kinds of sound absorbing material and quantity to fine tune bass quality.

7th. step: Happy listening!

M2 is a best speaker for near field listening in a small room. And It doesn't make me think to make another small 2-way loudspeaker.

Part List

Units

Network

Resistor
Ohmite 10.0 Ohm 10Watt: 2 ea.
Ohmite 2.0 Ohm 10Watt: 2 ea.

Coil
Air-core 3.3mH 14AWG: 2 ea. (Woo! How big it is!!)
Air-core 0.8mH 14AWG: 2 ea.
Air-core 0.16mH 14AWG: 2 ea.
Air-core 1.2mH 18AWG: 2 ea.

Miscellaneous

Specification

Man-Sun Huh

Figures

Fig. 1. Picture of the finished M2

Fig. 2. Enclosure design plan

Fig. 3. Impedance measurement of woofer unit

Fig. 4. Impedance measurement of tweeter unit

Fig. 5. Treatment for woofer frame resonance

Fig. 6. Ground plane method

Fig. 7. SPL of woofer (W/O X-over)

Fig. 8. SPL of tweeter (W/O X-over)

Fig. 9. The X-over network

Fig. 10. Picture of the X-over network
Picture of the X-over network (Upper:Woofer network, Lower: Tweeter network)

Fig. 11. SPL result of M2

Fig. 12. Impedance result of M2