4 ohm 18" LF driver recommendations

quiet motors
I heard loud chuffing noise from a driver for the first time recently. Testing a Nero-15sw800 driver with it upside down the with cone firing into a 3cf sonotube, with high power (not sure just high, but with 300W capable amp), chuffing & other noises emerged from the ports at the back with transients, drum whacks etc. Surprising how loud it was; at first I thought it was hitting the limits of driver excursion. Thankfully, in the sealed box, none of this noise is audible.
 
@profiguy: I am not sure about the noise solution in this case. What I see is a spider issue. Now with Mms in the ballpark of 30-60g depending on the driver size, it is obvious that even the spider takes a hit from weight reduction efforts as a compromise for sensitivity. They now do couple of high power dual spider (possibly silicone one) drivers that should behave much better, but at that point these are 2-3dB down, so additional 50% power input capabilities go almost nowhere. Sometimes it is even a loss. Anyways, endof OT on my side.
 
I agree that most published port recommendations tend to be on the small side.
If they weren't, then they'd tend to be very long ports. I suppose many systems won't run close to maximum power very often, so the problems associated with a smaller port won't show up all that often.
I'm a firm believer on running the largest port you can for a given tuning, but that also means larger enclosures to obtain less port length (if the driver allows it).
The NERO-18SW1900D driver provides a triple whammy when it comes to working out a suitable port length. Being 18-inch, the driver has a large radiating area, and a very high Xmax, so it displaces a lot of air when it is operated at high power levels. That requires a very large port area, resulting in a very long port.
QB3 alignments usually call for large enclosures close to actual driver Vas with average loss factored in. In these instances, you can use very short, large area ports, consequently almost eliminating port resonances.
I'd tend think that these QB3 alignments can use shorter ports, rather than "very short" ports. The QB3 alignment tends to have a ratio of fb/fs > 1, which helps keep port lengths shorter. However, for a driver with a Qts=0.42, such as the NERO-18SW1900D, we are getting into territory where a low-frequency alignment will have fb/fs ≤ 1, so the port length requirement starts going up.
 
  • Like
Reactions: profiguy
Most motor noise stems from excessively restricted airflow through the VC gap with elevated excursion and/or higher output levels at midbass frequencies. Sometimes suspension rocking modes and physical cone shape distortion can aggravate VC alignment issues. The VC former itself not transferring the drive forces evenly to the cone can be the result of going out of round from radial resonance modes. This is why light weight, unreinforced VC formers aren't a good design for midbass drivers, despite requiring lower moving mass to keep efficiency and output sensitivity high.

Cheaper drivers don't usually have under spider venting, which combined with the above mentioned VC gap restrictions causes chuffing and odd order harmonic type clipped sounding reproduction of the fundamental frequencies.

Another common cause of motor noise isn't directly related to the VC gap itself, but rather a resonance of the air trapped behind the dustcap through the main rear pole vent. Some drivers have a foam insert to dampen this but larger LF drivers run into an airflow restriction at certain frequencies due to these resonances maxing out the airflow capacity of the main rear vent.

Using pro drivers for hifi applications is a craps shoot because you never know if the driver is capable of quiet excursion. These speakers are mainly designed to be durable under hard use so priority is given to cooling capacity instead of having the quietest possible motor. Using any PA type speaker in an inverted arrangement with basket facing out usually won't produce desired results. In the reasonably priced range, the Eminence Lab series drivers are very quiet for applications using inverted arrangements. If a driver is of mediocre design, it will be noisy at higher output levels, even in 4th order bandpass enclosures. Sometimes baffles can help, but usually they don't get rid of all the noise despite having the high output levels mask it.

For midbass purposes, the best B&C drivers I know of are the 12MH32 and 12PE32. I primarily look for units with perforated VC formers and softer spiders coupled with narrow surrounds. Lowering the sub's upper LP cutoff and having the range above it carried by a separate dedicated midbass is usually the only way to control motor noise induced distortion. Many of the lower Qts drivers designed for FLH uses will have elevated motor noise. This i due to the tighter clearances and VC gap needed to transfer heat away from the VC when mounted in very small sealed boxes of FLHs.
 
Last edited:
  • Like
Reactions: witwald
@witwald --

You calculated the response of a NERO-18SW1900D in a 160 liter sealed box earlier in this thread. Could you please try scaling the box down in steps down to about half that size? That would be most helpful. 🙏

I was too busy to build proper enclosures for the two NERO-18SW1900D drivers, and just had them running naked OB for a while. With enough power and EQ, two of these subs reach nicely below 30Hz, but it's not ideal. Neither are 2 160 liter boxes, to the boss of the household. Admittedly, I would not like such big boxes in my LR either. 😉

1000013094.jpg
 
OTOH, that's asking more than I need. Could you please do your magic for 3 & 4 cf sealed boxes?
Sure...happy to run the simulations for the NERO-18SW1900D. Below are the results, with QA=QL=30, a 4th-order Linkwitz–Riley low-pass filter set to 58Hz, and the enclosure volumes set at 5.65cuft, 4.00cuft, and 3.0cuft (equivalent to 160litres, 113 litres, and 85.0litres). At 30Hz, reducing the enclosure volume from 160litres to 113litres drops the output by about 0.8dB, and when the going from 160litres to 85.0litres the SPL output drops by about 1.3dB. These certainly aren't huge differences.

1743755637039.png


Now that I've completed my "homework assignment" 🙂, I'm going to do some further exploration of the available parameter space.

Sticking with the 85-litre enclosure, I've tried adding a series capacitor to see what it can achieve in terms of boosting the low-frequency response. This is a method that I believe Arendal use in some of their speaker systems, judging by the published impedance curves. The solid black curve below is the frequency response obtained when a series 1000µF capacitor is added, but the power input has also been reduced from 240W re 8Ω to 130W re 8Ω. Note that a 4th-order Linkwitz–Riley low-pass filter set to 68Hz is also in the system.

1743756777150.png


For the same enclosure settings as above, but with the power level increased to 660W re 8Ω, the simulated frequency response, power and driver excursion curves are shown below. Note that the capacitor is providing a 1st-order high-pass filter roll-off below about 25Hz.

1743758499810.png
 
Last edited:
For the BMS 18S450, we get the following response curve for a 85-litre enclosure. The input power level is 280W re 8Ω, there is a series 1000µF capacitor, and the low-pass Linkwitz–Riley low-pass filter is set to 65Hz. The low-frequency extension is very similar to that of the NERO-18SW1900D, apart from the reduction in maximum SPL.

1743758991075.png
 
Sticking with the 85-litre enclosure, I've tried adding a series capacitor to see what it can achieve in terms of boosting the low-frequency response. This is a method that I believe Arendal use in some of their speaker systems, judging by the published impedance curves. The solid black curve below is the frequency response obtained when a series 1000µF capacitor is added, but the power input has also been reduced from 240W re 8Ω to 130W re 8Ω. Note that a 4th-order Linkwitz–Riley low-pass filter set to 68Hz is also in the system.
Thanks for the sims! Interesting how little the response changes with Vb halved!

WRT the series capacitor -- it changes the rolloff slope, similar to increasing Vb? Not sure if that marginal stretch lower is worth the 1000uf in the output path from the amp. What effect does this have on the amp?
 
Yes, the series capacitor increases the low-frequency roll-off slope by 6dB/octave. In general, I would say that increasing Vb does not really alter the roll-off slope, but rather it modifies the response around the knee of the curve. This effect will also be dependent on the nature of the low-pass filtering that is being applied. That's because all of these filter functions are interracting together to produce the overall response.

The series capacitor also provides some boost in the band-pass region of the subwoofer. This is illustrated by the plot shown below. The high-frequency roll-off is due to the action of the active low-pass filter that was added to the circuit. The low-frequency roll-off includes a boost region in the 23–44Hz frequency range, with the maximum boost being about 2.5dB. That is simply passive gain that has resulted from the interaction of the series capacitor and the driver's in-box impedance peak. The use of this simple technique was described by Thiele in his 2010 paper "Closed-Box Loudspeaker with a Series Capacitor".

1743822150566.png


As to what effect it has on the amplifier, that will, of course, depend on the amplifier. It seems worthwhile studying how the series capacitor affects the impedance curve of the subwoofer. In the plot below, the black curves are for the subwoofer alone, while the green curves are for the subwoofer with a series 1000µF capacitor added to the circuit. The solid curves correspond to the relevant magnitude response, while the dashed curves correspond to the phase response. In the frequency above the impedance peak, the impedance is a little more capacitive in nature, as the phase is a bit more negative. Below the impedance peak, in the frequency range from 30Hz to 55Hz, the impedance is more resistive (less inductive). In the frequency range from 10Hz to about 30Hz, the impedance becomes capacitive in nature, whereas previously it was inductive. However, the phase at 10Hz, which is about −70°, is not much less than the phase at 67Hz, which is −64°. Hence, the question seems to be whether or not a typical solid-state amplifier will have an issue driving a capacitive impedance load at frequencies around 20Hz. Maybe some amplifier designers can provide some advice on that topic?

1743822794422.png
 
Last edited:
For the BMS 18S450, we get the following response curve for a 85-litre enclosure. The input power level is 280W re 8Ω, there is a series 1000µF capacitor, and the low-pass Linkwitz–Riley low-pass filter is set to 65Hz. The low-frequency extension is very similar to that of the NERO-18SW1900D, apart from the reduction in maximum SPL.

View attachment 1444264

X-max figures are not comparable for the two drivers.
 
@-CGL- Thanks for the clarification.

For the BMS 18S450, Xmax = (Hvc−Hg)/2 = (36−12)/2 = 12 mm.

For the SB Audience NERO-18SW1900D, Xmax = (Hvc−Hg)/2 = (39.5−14)/2 = 12.75 mm.

Hence, using this more conservative definition of Xmax, it's apparent that the two drivers have a very similar Xmax, differing by 0.75mm. Re-visiting the simulation for the NERO-18SW1900D using the new lower value of Xmax, the results for an input power level of 350W re 8Ω.

1743842127514.png