The 1 Compression Drivers

We will test several 1" exit compression driver in this article:

• 18Sound ND1095N-16: 1.75 inch Nitrided titanium diaphragm
• 18Sound ND1090: 1.75 inch Nitrided titanium diaphragm
• 18Sound ND1TP-16: 1.75 inch titanium diaphragm
• BMS 5530-16: 1.75 Kapton annular diaphragm
• BMS 4552-16: 1.75 Kapton annular diaphragm
• RCF 350ND
• RCF CD350
• Faital HF10AK
• Faital HF108R
• JBL 2426
• Kartesian CMP35 VPA
• Celestion CDX1-1430
• Celestion CDV1-1757
• B&C DE360
• Beyma CP385Nd

For interpreting results about break-up, it is advised to read the breakup article.

In essence: Stiffer materials, like metals, will exhibit a higher but more violent break-up point, while materials with inherent damping, like treated paper or composites (Kapton), will break up at a lower frequency but affect a wider range of frequencies.

It’s all a mater of compromise, the larger the diaphragm is lower he can goes (theoretically) and the sooner the breakup will occur.

The horn used for testing is a X-Shape X33 with his return for doesn’t be impacted with midrange narrowing or beaming.

For the RCF 350ND, as she cannot goes beyond 1250hz a X25 is used.

SPL

Here is the SPL alligned on 4kHz witout EQ of crossover.

Here there is no “more” or “less” of something, the drivers gives an initial energy then the horn distribute as we have explained it here : Energy in horns.

If a driver was flat it will simply mean that the horn is not constant, exactly what we do not want, however the 18S 1090 exibit a very easy to EQ shape.

So the “bell” response is normal, the two 18sound come from the familly even if the generation is not the same, the shape of the BMS is very different.

We can see that the two Faital HF10AK have problems around 2.1kHz, especially one of the two units, that have even worst distortion than the other one.

The main juge here will be the distortion measurement after have EQ it flat.

Distortion

Here is an overlay of all the compression drivers, the measurement is taken at 31.5cm to the horn mouth, every driver is EQ flat, the volume is 95 dB SPL (so the dB at 1m).

A point about audiblity

Our article about audibility and masking effect has explain that a H2 distortion will create a sound at Fr*2, and a H3 at Fr*3.

In conjunction with audibility and masking effects, H2 is likely audible above 2%, and H3 is likely audible above 0.2%.

So some high frequencies rise up in H2 are in fact not audible, we have to put our attention at the low end to the half part of the measurements.

The two 18Sound ND1TP have been let on the graphs for show possible divergence beetween unit, the same divergences occurs with BMS too.

H2

The 18Sound ND1TP are partculary good here, as an improvement of the 1095N, the BMS 5530 a little less.

H3

The HF108R is good but not very low, so in a X-Shape X25 but not with a bigger horn.

For the RCF 350ND I have lost the measurement and I don’t have the compression anymore, it remain a lower measurement one that just show that H3 goes up very fast after 1250/1300hz :

Here we see the an advantage of the ND1TP but in fact all performed very well. It’s now time to talk about divergences on the same unit and diaphragm adjustment.

When I receive brand new ND1TP pair with one producing 2% of THD at 80 dBSPL, she was sounded broken, both unit have been re-tuned about diaphragm height, as 18sound diaphragm are auto-centred it’s easy to tune diaphragm height.

More a compression driver is tiny more it will be sensitive to it, even between my several BMS 5530 there is differences in THD and polar.

18Sound use 3 wedges: 0.2, 0.1 and 0.05 mm, to have these performances in the test I remove 0.05mm on one and 0.1mm on another, a difference of 0.05 completely change THD but also polar map as we will se it just after.

Time-domain measurements

Here we will do Burst Decay at 32cm of the mouth with all driver EQ flat, we will see here the breakup impact on Time-domain measurement.

Nothing audible on 18Sound and BMS, but on the RCF it’s not very good even if it’s also not really in audibility range, the BMS 5530 is a champion on this test.

The BMS 4552 has a little problem around 10kHz and we will see later that that has an impact on the polar map, as always.

The Faital HF108R is not very good and we will see that this will not help for his polar measurement.

Polar Plot

Why show polar on different driver?

A horn starts from the principle that the compression driver is a perfect plane wave radiation on his exit.

But the breakup put the driver out of plane wave radiation as we have seen in breakup and his distortion.

By analysing the moment when breakup occurs on a perfect wave propagation horn as the X-Shape X33 is we can detect when and how breakup affect wave-front propagation in the horn, and by deduction the quality of this propagation and the respect of plane wave radiation.

Polar on X-Shape X33

Polar conclusion

We can see a huge difference is very high frequencies if the diaphragm is not at the right height, after tuned diaphragm height on the ND1, his polars are better than 1095N, the 5530 and 4552 unit show a very low variation between unit so I let just one the both polar for each.

The RCF 350ND is not the better in this test, it’s a driver that goes out of plane wave radiation sooner than others, but on an X25 she goes out of plane wave radiation very high.

The BMS 4552 has a very little Time-domain problem around 10kHz as we see above, we can see that the new BMS 1" driver generation (the 5530) is a little bit better and that is linked to the Time-domain behavior so the plane wave behavior.

Diaphragm tuning impact

Polar plots offer the most precise method for detecting tuning issues, with an accuracy of 0.05mm (the height of the finest tuning ring). However, a simple sweep test can also provide valuable information.

If you observe all distortion harmonic orders centered at the same frequency, forming a distinct peak or ‘spade’ shape, this often indicates a diaphragm centering issue.

I may have encountered an unfortunate instance, as other customers may not have experienced the same problem out of the box. Furthermore, this type of issue can occur with any brand.

Dust, metal shavings, or particles in the voice coil can also exhibit similar symptoms. However, we will observe the distortion changing at each measurement, particularly at the low end, as the responsible element moves within the voice coil and changes position during each sweep.

Conclusion

The 1090 is a clear winner as the 1095N and ND1-TP. The BMS 5530 will also perform well as the RCF CD350 (that we will prefer over the RCF 350ND).

The 1090 is particularly suitable for passive filtering or reducing the need for EQ. It is also less expensive than the 1095N, while its polar response remains good and representative of that of a titanium diaphragm.

If a very high SPL is needed and the system needs to cut very low, the 1095N and the ND1-TP will have a slight edge over the BMS 5530.

The surprise is the RCF CD350—this cheap driver combines high-frequency breakup with some spreading. Its plastic diaphragm reaches high frequencies like a hard diaphragm but with a bit of the spreading typical of plastic ones, though much less than usual.

This highlights a plastic diaphragm that is stiffer than usual, offering an excellent balance between rigidity and damping. It has a maximum crossover point at 1150Hz with a steep slope. Its “partially” spread breakup is visible in the polar response around 11kHz but remains very controlled. The temporal measurements also show some movement, but it’s surprisingly well-managed for a full 1.75" plastic diaphragm.

The Faital HF108R is fine even if there is better options.

The 4552 is good and is an older generation of the 5530, we can see the improvement on the plane wave radiation behavior and Time-domain measurements.

BMS 5530 annular diaphragm does a good jobs in terme of breakup and plane wave radiation (seen in polar), as always when annular is used, but we paid it with a bit more H2 at 1khz due to the tinier surface when the diaphragm is annular comparing to regular ones. If very high SPL is required (e.g., in a nightclub), it is recommended not to use an annular diaphragm.

The Celestion CDV1 exhibits very good distortion performance and extends relatively low in frequency. However, its Time-domain behavior is significantly degraded: This driver benefits from the advantages of its plain plastic diaphragm, which allows for good low-frequency extension, but typically results in less ideal Time-domain measurements due to the inherent characteristics of non-annular plastic diaphragms.

The best 8/9.5" woofers for be used with these compression drivers are here: 8/9.5 inch woofer test.

The 1.4 Compression Drivers

We will test several 1.4" exit compression driver in this article:

• 18Sound ND3ST-16: 3 inch titanium diaphragm
• 18Sound ND3SN-16: 3 inch Nitrided titanium diaphragm
• 18Sound ND3A-16: 3 inch Aluminium diaphragm
• 18Sound 1480N: 3 inch Nitrided titanium diaphragm
• 18Sound 1460: 3 inch titanium diaphragm
• FaitalPro HF1440-8: 3.4 inch Ketone annuar diaphragm
• FaitalPro HF1460-8: 3.4 inch TeXtreme (woven carbon) diaphragm
  

For interpreting results about break-up, it is advised to read the breakup article.

In essence: Stiffer materials, like metals, will exhibit a higher but more violent break-up point, while materials with inherent damping, like treated paper or composites (Ketone), will break up at a lower frequency but affect a wider range of frequencies.

It’s all a mater of compromise, the larger the diaphragm is lower he can goes (theoretically) and the sooner the breakup will occur.

The horn used for testing is a X-Shape X34 with his return for doesn’t be impacted with midrange narrowing or beaming.

SPL

Here is the SPL alligned on 1kHz witout EQ of crossover

On X-Shape X34:

On M-Shape 50cm, note: The HF1440 have been measured at 1m and not 32cm:

Here there is no “more” or “less” of something, the drivers gives an initial energy then the horn distribute as we have explained it here : Energy in horns.

If a driver was flat it will simply mean that the horn is not constant, exactly what we do not want.

So the “bell” response is normal, the only strange things is the end of the response of the HF1440 (Keton), but we will see that will give she a huge advantage on another type of mesurement.

The low end of the HF1460 (Carbon) show that she don’t want to go down but for a passive crossover it can be an advantage.

In addition, the HF1440 (Keton) have show almost the exactly same response on a RCF 950, as we have said upper, if the coverage is constant and similar beetween severals horns (X-Shape X34, M-Shape, RCF950…) the on-axis response will always remain the same.

The compression driver adapted to an horn is more a myth than otherthings, however we can adapt compression flare to the horn (all compressions are slow flare ones here).

But as energy will always remains the same, no changes should be expected.

The main juge here will be the distortion measurement after have EQ it flat.

Time-domain measurements

Here we will do Burst Decay at 32cm of the mouth with all driver EQ flat, we will see here the breakup impact on Time-domain measurement.

There is pro and cons here, compromise, as always in audio, The two 18Sound perform well (3" diaphragm), the HF1440 breakup soon as it’s a Ketone 3.4 diaphragm, I was excepted better from the HF1460 and his TeXtreme Carbon diaphragm.

The 3.4" ketone of the HF1440 have his breakup at 12khz, but nothing else and we will see that it will have an impact on other measurements.

Distortion

Here is an overlay of all the compression drivers, the measurement is taken at 31.5cm to the horn mouth, every driver is EQ flat, the volume is 95 dB SPL (so the dB at 1m).

Initially the ND3SN has a dust in the voice coil that I removed, this kinds of things degrade performances, I keep the measurement in black for comparaison purpose.

A point about audiblity

Our article about audibility and masking effect has explain that a H2 distortion will create a sound at Fr*2, and a H3 at Fr*3.

So some high frequencies rise up in H2 are in fact not audible, we have to put our attention at the low end to the half part of the measurements.

H2

The 18Sound are partculary good here, the two Faital a little less but everybody is bellow 2%, so almost not audible.

H3

Here we see the huge advantage of the 3.4" Ketone of the Faital HF1440, is goes low without forcing, at the reverse, as the 1460 don’t want to go down, we can see in comparaison more H3 at low end.

The two 18Sound ND3 are almost the same, they pass the 0.2% of H3 that begin to be audible, it’s the limit but it logic for a 3" diaphragm.

Some differene can just be the distance with phase plug, we will seen in others test that can change a lot of things and it’s not necessary well tuned at the out of the factory. moreover there is manufacturing tolerance that can explain some differences here, even with two identicals drivers we will not have the exact same result.

Polar Plot

Why show polar on different driver?

A horn starts from the principle that the compression driver is a perfect plane wave radiation on his exit.

But the breakup put the driver out of plane wave radiation as we have seen in breakup and his distortion.

By analysing the moment when breakup occurs on a perfect wave propagation horn as the X-Shape X34 is we can detect when and how breakup affect wave-front propagation in the horn, and by deduction the quality of this propagation and the respect of plane wave radiation.

Polar on X-Shape X34

Polar conclusion and breakup impact

Here the logic is well respected, more the diaphragm is Stiffer (“hard”) more the breakup will bring the wavefront too far from plane wave radiation and enter in an impredictible chaotic behavior.

The HF1440 (Ketone) is incredible for a 3.4" diaphragm, she acte like a good 1" one and his breakup almost doesn’t affect it’s plane wave expansion at 12 kHz.

Conclusion

As always it’s hard to give a winner as we have taken the most advanced 1.4" drivers of the market.

While I was hoping for even more impressive results from the HF1460 and its TeXtreme diaphragm. It will be interesting to see how TeXtreme (or other Carbon) diaphragms compare to Beryllium options, particularly TAD’s vaporized Beryllium, in the future.

But we can give some orientations:

The Faital HF1440 (Ketone) is interesting, the only downside is the H2, due to annular diaphragm the emissive surface is lower than others, it’s very classic with annnular diaphragms, they are very good in H3/5/7/9 but not in H2 due to the need to moove more for the same SPL level.

18Sound 1480 and 1460 are good in high frequency Time-domain behavior and high SPL, for high demanding SPL the 1480 is recommended (Night-Club).

For 1" test see our another test: 1" compression driver test.

The 12" Mid-woofer

For now we have one 12" in this article:

• Sb Audience NERO 12mwn700d

We will also measure distortion with and without a 1.2mH air core inductor, to reduce the driver breakup impact on distortions when possible, as seen in speaker breakup article.

In essence: We use Back Electromotive Force principle with an air core inductor that will increase gradually the impedance (affecting the voice coil) and replace some EQ to linearize the woofer in the same time, it works very well if the woofer is already rising up in his final box.

It will reduce a lot H3 and H5 distortion related to breakup if there is.

Distortion measurements are done at 31.5cm 95dB SPL, all drivers are EQ flat to be compared : The SPL is the same on all the bandwidth, for all drivers, with and without air core inductor, the difference with and without air core inductor is not due to SPL but to Back Electromotive Force principle, as explained in our dedicated article.

SPL

Distortion Without inductor

H3

H5

Distortion with inductor

H3

H5

Time-domain measurement

As always, there’s nothing to see in the time domain if we stay away from breakup:

Conclusion

The driver has a well-damped breakup, so naturally there’s no rise in distortion. The increase in SPL makes it suitable for operation with an air-core inductor. We replaced some EQ with an air-core inductor effect.

The SPL remained the same, but we changed the impedance at breakup, so the driver performs even better in H3 and H5.

It’s a very good alternative to the 12LW1400.

The driver allow a crossover not limited by distortion, we can cut it when directivity match with an X-Shape X33 or X34 at 1150hz or even higher, without the need for an air core inductor.

The 6/7" Mid-woofer

We will test several 6/7.5" Mid-woofer in this article:

• SbAcoustics SB17NBAC-4
• SbAcoustics SB17CAC-4
• SbAcoustics MW16-TX-8
• SbAcoustics MW19-TX-8
• ScanSpeak 18W4434G00
• Sica 6.5H1.5CP
• STX W.18.200.MCX_v2
• STX W.18.140.8.MCX

All drivers are EQ flat to be compared, at 95 dB SPL.

Discalified mid-woofers

There is too many woofers to have readable charts in overlay so severals has been disqualified, there is the reason why:

• STX: Both STX have a hollow about 6 or 8dB in plain midrange, even without EQ we see abnormal distortion then after have EQ it flat, it’s of course worse, letting this hollow will not allow to use theses mid-woofer, so they are in fact not usable.

• Sica 6.5H1.5CP: It’s not a bad driver but other are excellents (without air inductor), it was a good Q/P ratio driver before, now his price has risen so he lost his interest, However it’s a driver that rise up in frequency and love to be linearize with a 1.7mH air inductor, resulting a very good driver in active filtering thanks to this trick (as seen in speaker breakup article).

• ScanSpeak 18W4434G00: Not a bad driver either, the best of the four, it rises a little bit too much in distortion after 800/900hz but very good in low medium.

Top 4

We will take a look of order that show a difference between each drivers, some order are “far” but they are outside of mask effect and in audible range (depending of their %) as seen in masking effect and audibility article.

Due to their surface and directivity match none of this driver should be crossed above 2/2.5khz, almost all have their breakup begining at 4.5/5kHz.

IMD

Here we take the two better and try to find a difference in IMD, with a multitone 4 IMD test at 95 dB SPL.

Here we see IMD but THD is also present in the chart so it’s hard to separate the two, we can just say that the MW19-TX looks like better but as he is better in THD too.

Time-domain measurement on midwoofer

When we do it with a compression driver and a horn we measure in fact the couple, for the woofer we don’t want to have the box and the way the woofer is integrated with it.

If we measure CSD or Burst Decay at 25/32cm we measure the woofer with his very close environment (even if we gate the measurement).

If we measure the woofer at 2cm we will only have the woofer alone, and we will see that at breakup his Time-domain measurement is not good. The rest will be perfect, for all.

So avoid being close to breakup, anyway distortion will dissuade you to go near to it, all drivers perform perfectly in Time-domain measurement in usable range here.

Conclusion

STX does a great performance in the 8inch driver test, but here we can see that they have both the same issue between 700 hz and 1.2khz, without EQ there is a huge hollow, and EQ we excite the conception problem.

The MW19TX is a perfect companion for a 2 voices speaker, cross arround 1600hz with a tweeter waveguide like ReShape, not too big, and without the need of a midrange voice, at the cost of his huge price of course.

The SB17CAC under perform compared to NBAC/NAC.

The MW16TX is a little bit stuck, stuck between the MW19TX and the SB17NBAC and by its price, moreover the frequency raw response of the MW16TX is a little bit accidented after 1khz when the MW19TX is not.

There are two winners, the SB17NBAC and the MW19-TX, both have linear response in bonus.

The 8" Mid

We will test several 8/9.5" midrange drivers in this article:

• 18Sound 8NMB420
• B&C 8MBX51
• BMS 8S219

We will measure distortion with and without a 1.2mH air core inductor, to reduce the driver breakup impact on distortions when possible, as seen in speaker breakup article.

In essence: We use Back Electromotive Force principle with an air core inductor that will increase gradually the impedance, affecting the voice coil, and replace some EQ to linearize the woofer in the same time, it works very well if the woofer is already rising up in his final box.

It will reduce a lot H3 and H5 distortion related to breakup.

Distortion measurements are still done in the same condition, all drivers are EQ flat to be compared : The SPL is the same on all the bandwidth, for all drivers, with and without air core inductor, the difference with and without air core inductor is not due to SPL but to Back Electromotive Force principle, as explained in our dedicated article.

Raw Response 2.84V/1m

We can see that the breakup is around 2750 Hz for both

Distortion

All measurements have been done at strictly the same SPL, 95 dB SPL, and with EQ flat across all the measured bandwidth.

For these cone drivers with similar emissive surfaces (Sd), the main source of distortion will be the breakup effect. Therefore, we should carefully examine H3 and H5.

H3

H5

All perform well in H3 but should no be cut so high even if they are midrange drivers. There is a strange peak in H3 for the BMS 8S219.

There is a peak in the H5 measurement for the 18Sound unit, indicating an unwanted sound frequencies around 2500/2700Hz produced. It’s related to breakup, this H5 peak is a little bit too high.

Distortion with Air core inductor

H3

H5

As these speakers exhibit a significant rise in their frequency response, they demonstrate a substantial improvement in H3, making them very suitable for ’true midrange’ applications. However, the H5 peak remains present in the 18S model as well of the H3 spade for the BMS.

Time-domain measurement

As always, there’s nothing to see in the time domain if we stay away from breakup:

18Sound 8NMB420

B&C 8MBX51

BMS 8S219

Conclusion

The 8NMB420 exhibits some counter-performance due to its significant cone breakup, but this will likely be noticeable at very high SPL.

The BMS 8S219 has a small spade in H3, hard to explain, not very noticeable. The BMS motor looks very evolved, but the speaker basket uses only four screws. That is a little bit optimistic for this kind of driver (8") design to play very loud, when it comes to high SPL usage, where vibration and airtightness become important.

Considering its price the B&C is very interesting, the B&C 8PE21 was already the best 8" medium for years and it seems B&C still have the lead in 8" medium, even if all drivers perform very well.

Important Note: All these drivers are considered as midrange-capable only. Even for the tests conducted, substantial low-end EQ was necessary to achieve a flat response. 200Hz already represents a very low frequency for these drivers.

Generally speaking, we can observe that this speaker allows for a high crossover frequency, enabling a good directivity match around 1400/1500Hz. However, in the midrange at 105 dB SPL, their performance is comparable to that of good 8" Hi-Fi speakers.

These filters, implemented using linear-phase (IIR filters), will influence both Group Delay and phase, allowing for their linearization in the low-frequency range.

The 8/9.5" Mid-woofer

We will test several 8/9.5" Mid-woofer in this article:

• Wavecor WF223
• SbAcoustics SB23NBAC
• SbAcoustics WO24P
• ScanSpeak 22W4534G00
• Tymphany Peerless HDS-P830869
• Sica 8H2CP
• Visaton TIW 200 XS
• STX W22.200.8.MCX_V2
• STX W22.200.8.MCX_V3
• PURIFI PTT 8.0X 4 NAB 2

We will measure distortion with and without a 1.2mH air core inductor, to reduce the driver breakup impact on distortions when possible, as seen in speaker breakup article.

In essence: We use Back Electromotive Force principle with an air core inductor that will increase gradually the impedance, affecting the voice coil, and replace some EQ to linearize the woofer in the same time, it works very well if the woofer is already rising up in his final box.

It will reduce a lot H3 and H5 distortion related to breakup.

Distortion measurements are done at 25cm 98dB, all drivers are EQ flat to be compared : The SPL is the same on all the bandwidth, for all drivers, with and without air core inductor, the difference with and without air core inductor is not due to SPL but to Back Electromotive Force principle, as explained in our dedicated article.

Time-domain measurement

As always, there’s nothing to see in the time domain if we stay away from breakup:

WO-24P

PURIFI PTT8.0

STX W22.200.8.MCX_V3

Conclusion

Without the little rise up on H3 at 200/300hz the STX V2 and V3 will be ex-aequo on distortion with the WO24P.

The self transform both WO24P and STX in low-medium woofer for a 2 way usage cut at 1250/1300 hz with an X-Shape 25.

With a Back Electromotive Force system, such as a notch filter or an air-core inductor, the STX matches the best-in-class woofers in terms of distortio, at a fraction of the price.

The Purifi is an expensive killer and doesn’t need an air-core inductor to perform extremely well compared to the WO24P. The last advantage of the WO24P is its 1.5" more larger size (more emissive surface), which will make a difference in the very low end (not visible on charts) and also in directivity control when we want to control directivity as low as possible with the HF horn at crossover.

The best compression drivers usable with them are tested here: 1" compression driver test.

There is a huge difference in chassis between the WO24P and this STX that is an stamped sheet metal making the STX harder to integrate, but well, we have seen that it doesn’t impact performances.