In this blog, I go through the design and build process of a mini subwoofer that uses the technique of PPSL and DSP. The blog also discusses what went right and wrong in the process of my first build.
This past semester at University of Miami, I took Transducer's Theory under the guidance of Dr. Bennett. For the final project of the course, I decided to—instead of buying a kit—design and build my own mini subwoofer using CAD modeling, 3D printing, and DSP.
I was heavily inspired by the design of Sonos Sub Mini. The way they arranged two drivers to be facing each other in order to reduce vibration and distortion was both clever and gave the product a rather unique look in the commercialized subwoofer market. Although, there is a lack of information from Sonos on how well the PPSL (push pull slot-loading) method works realistically, especially when compared to the usage of DSP for a mini subwoofer. It is to my understanding that drivers placements such as PPSL and Isobaric setups were used to combat the presence of weaker magnets found in older drivers.
Nevertheless, it looks so incredibly cool.
For the drivers of the mini sub, I used two Dayton DS135s. They have a 5in cone diameter. Given the limited cone size and excursion, my hope(without any prior knowledge) was that the PPSL design can help reduce how hard a single driver would have to push in order to achieve the same loudness. The two drivers share the same enclosure, with their cones sharing a short port together. The wiring ensures that when one cone pushes out, the other pulls in. Each driver receives the same mono output that was converted from stereo with Analog Devices' Sigmastudio, which means they share the same signal. One of the drivers is wired in opposite polarity so that when one rarefies, the other compresses. While it reduces mechanical vibration, it can also help neutralize the drastic pressure changes inside the enclosure that would have been present had the two drivers shared the same wiring.
While designing, I naively thought that there would be no real concern for the length of the shared center port because of its relatively short length. The resonance frequency would have probably been in the 200-300 Hz range, and the subwoofer’s crossover frequency would have happened way before that. However, real life testing showed that when volume exceeds over 50%—60%, heavy, audible chuffing occurs. New design with a much bigger center port is already underway, and hopefully can help resolve this issue. As for vibration, real world testing from 30 Hz to 100 Hz demonstrated that the enclosure produced almost none even when pushed to max.
There is a port on either side of the back of the drivers as well. Measured at 18 cm long, 1.0 cm wide, and 5.3cm tall. Since the drivers share the same enclosure, we can treat the two ports as one as well, but first, we need to convert the combined rectangular cross sectional area to a circular one in order to simulate it in the software:
Combined cross sectional area:
1.0*5.3*2 = 10.6 cm^2;
Find equivalent radius: 10.6/3.14 = 3.37 cm^2;
sqrt(3.37) = 1.83 cm;
Diameter = 1.83*2 = 3.6 cm;
As a general rule of thumb that I found online, the minimal diameter of the port should be at least ⅓ of the cone diameter to avoid chuffing, and from there we can calculate the necessary length. The port should have had a minimal diameter of 4 cm, but the corresponding length would have been impossible to fit inside the enclosure. This is why 3.6 cm was chosen. Moreover, the length it needed meant that chuffing is inevitable. However, chuffing from the main port became so overwhelming that it was difficult to tell how much chuffing the two side ports were producing.
The amp I chose was the Dayton KABD 2x50W. It features DSP using Sigmastudio, bluetooth 5.0, 3.5mm line in, 4 pots for customized controls. A lot of processing was done using the software. For example, I added dynamic bass boost so that when the input signal is lower, it boosts the lower frequencies so that it remains present. Some EQ was used to tame the drivers more inside the enclosure. It also featured some compressing and limiting so as to not damage the drivers. The cutoff was also done digitally as opposed to electronically by a simple EQ. It was much easier and the EQ allows for a much higher order cutoff, as well as different types of cutoff frequency: Chebyshev, Butterworth, and Bessel. One pot was assigned to gain control and another was assigned to cutoff adjustment from 80Hz—120Hz.
Hardware wise, the control panel I printed for the subwoofer contains two LED lights: one for Bluetooth connection status, and the other for incoming signal status. It also has the 3.5mm line in option as well as a bluetooth antenna on the backside. A switch can be found that turns on and off the subwoofer, which I found to be quite convenient as opposed to having to unplug the DC power supply each time I was done testing the speaker.
The entire enclosure was printed by a Creality Ender 3 V3 SE, using PLA+ filament. Contrary to popular belief, plastic like PLA and ABS, and certain the higher end ones like PC as well, have a higher density than MDF. The lowest end plastic, PLA, has almost double the density of that of MDF, at around 1.3g/cm^3. PLA+ was developed upon the PLA formula, but companies were able to add modifiers to make it even stronger. On the inside of the enclosure, a large amount of butyl rubber was used to secure wires in place.
For the simulation, I used using this online simulator. It provides a vented box model based on the T/S parameters, a closed box model, and a vented box model based on my port design. There is debate online as to whether PPSL can be considered as one driver similar to isobaric clamshell. For the sake of ease, I will assume they can be.
Basic T/S parameters for one driver:
Impedance | Re | Le | Fs | Qms | Qes | Qts | Mms | Cms | Sd | Vd | BL | Vas | Xmax | RMS Power |
8 ohm | 5.9 ohm | 0.84 mH | 52.5 | 2.02 | 0.48 | 0.39 | 9.7 g | 0.95 mm/N | 75.4 mm^2 | 36.6 mm^2 | 6.23 Tm | 7.27 L | 4.85mm | 50 W |
I understand that a Qts of 0.39 is quite low for an open baffle enclosure, but Parts Express recommends using an open box following the golden ratio for a much lower tuning frequency, and my design was based on those numbers. I am willing to sacrifice the potential distortion and loss of efficiency for a little extra bass extension. For a mini subwoofer, a 5in driver with a Fs around 50 Hz is exactly what I was going for. We need to change some values, however, when we combine the two drivers:
Basic T/S parameters for two drivers:
Impedance | Re | Le | Fs | Qms | Qes | Qts | Mms | Cms | Sd | Vd | BL | Vas | Xmax | RMS Power |
8 ohm | 3 ohm | 0.42 mH | 52.5 | 2.02 | 0.48 | 0.39 | 19.4 g | 0.475 mm/N | 75.4 mm^2 | 36.6 mm^2 | 6.23 Tm | 3.63 L | 4.85mm | 50 W |
Simulated response:
Simulated group delay:
Simulated voice coil impedance:
Test result:
Clearly, the simulation was inaccurate. I am still looking into the cause of it and will upload a new blog once a new version comes out. Note the steep cutoff past 120Hz.
List of material:
Hardware:
Creality Ender 3 V3 SE
PLA+ filament
Dayton DS135
Dayton KABD 2x50W
Meanwell LRS-350-24
High voltage DC adaptors & miscellaneous hardwares
Software:
Analog Devices Sigmastudio
Autodesk Fusion 360
Creality Print
Written 06/21/2024 to 06/25/2024
Comments