DIY 3D printed speakers
Why did I do this project?
I used to use a Logitech z523 speaker system that sounded awful. the bass was very uncontrolled and the mids and highs were almost absent from the sound. It made my ears hurt and it felt like I had tinnitus after. I wanted to get a studio monitor, like the JBL LSR series but it was too expensive. So I had to make a solution myself.
GOALS:
- make a pair of speakers with a flat frequency response up to around 15- 20kHz.
- should have a reasonable bass response for the size
- play reasonably loud
- does not cost too much to make
THE PARTS USED:
Woofers:
Beats Rave SDS-65T19PR01-02 15W 2 Ohm Woofers
These woofers are apparently made by Tymphany, which is a high quality producer of transducers. It was from the Beats Pill XL Bluetooth speaker. There are many sellers offering this for cheap and I used them as it provided a high quality sound at a low price. I am using 4 of them (2 per channel) which is going to be wired in series for a 4 Ohm load. Since T/S parameters are not given, I measured it with REW and a custom cable made with Audiojudgement’s tutorial (T/S parameters of these drivers are at the very bottom of this article)
Tweeters:
EASTECH FIADT1010-0400 4 Ohm 10W dome tweeter
These tweeters are very cheap and cover the response up to 20kHz. The T/S parameters are also provided on the EASTECH website, allowing me to easily make a crossover network.
Crossover network:
Weah D224 2 way crossover
These crossovers will not work out of the box with the speakers. Modification is required for them to work properly, or else the sound will be very weird. The modification of the crossovers will be described in this article. Since the schematic of the crossover is not provided, below is the schematic I made:
As you can see from the schematic, it is a 2 way first order crossover. For the woofer, the crossover output is fed into a inductor (the original inductor is 0.8mh and a DCR of around 4-8). I am not sure of the value as I don’t have inductor measurement tools and I got the information by asking around. Then its feed to the woofer. My guess for the capacitor between the output of the inductor and the negative terminal is because for it to have a steeper roll off frequency. For the tweeter side, it is connected to a 3.3UF capacitor, then to 2 1 Ohm resistors to attenuate the tweeter, followed by a polyfuse for protection to prevent overloading. Given now that I know the schematic and how everything works, it is very easy to modify the values.
Terminals:
WP2-3 (with the backs sealed)
MAKING THE ENCLOSURE: THE FIRST ITERATION
the first iteration of the enclosure used a passive radiator design. It used a passive radiator that was used in Beats Pill XL Bluetooth speaker (came out of the same speaker as the woofers). At that time I thought that it was a good time to match the same drivers to the passive radiator as it was tuned for it (I was wrong, as passive radiators also relied on air volume). The design is a TMM (tweeter midrange midrange), with the passive radiator facing the back. At that time I used a tweeter out of a iMac, with a 5000uf capacitor as a high pass without the crossover. Below are the pictures of the first iteration:
MAKING THE ENCLOSURE: THE SECOND ITERATION
I was enjoying the setup, until the TDA7379 chip burnt. One day when I turned it on, it started to make crackling and popping noises, and had a burning smell. It seems like the output stage of the amplifier was damaged (likely due to high volumes and clipping, so I bought a more powerful amplifier. The new amplifier was a TDA7498E class D amplifier due to how it produced less heat (90% efficient), and had a high output rating of 160W per channel at 4Ohms, allowing me to turn up the volume to loud levels without even coming close to clipping or causing damage to the amplifier.
A design flaw with the speakers is there is no simulated data when designing the speakers. The speakers were designed without thought for passive radiator tuning and I was not able to find the t/s parameters for the passive radiators. So I decided to use a ported enclosure design to extend low frequency response and also even out the tuning.
For the crossover modification, I eventually removed the inductor and capacitor that goes with it as it made the sound very weird, and resulted in a unflat response. I changed the capacitor for the tweeter to a 5.6uF one so it has a crossover frequency of around 5000hz. In the future, I will try a Bi-amped setup where there is separate amplifiers for both the tweeters and woofers and all the crossover stuff is done before being amplified via a active crossover using op-amps (like the ne5532) or using a DSP to do so. This will improve the sound quality as the crossover is not affected by the impedance of the speaker drivers and can be easily tweaked and modified, while also reducing some complexity of the system.
Also, another thing to note is that I inverted the polarity of the tweeter, which fixed the dip you can see from the vitiuxcad simulation, which made the speakers sound more natural and detailed.
RESULTS:
After assembling and setting up, the speakers sounded very good. It had a flat frequency response and creates a highly detailed soundstage. For the first time, the speakers sounded very realistic and could accurately playback music and the speakers can expose the shortcomings of music, rather than covering them up.
Despite the low sensitivity of the speakers, the speakers can provide surprising output from such a small form factor. Using a TDA7498E amplifier running at 19V (90W laptop power supply), the speaker can put out 110 decibels near field (A weighted). I'm guessing that most of the output is coming out of the tweeters as it is more sensitive compared to the woofers, and given that A weighting places an emphasis on mid to high frequencies, it is understandable. Though, these speakers cannot play the same output as large PA or home speakers (or keep up with max SPL measurements at 1M with larger speakers), it is certainly enough for near-field use. In fact, i compared it with a Bluetooth (JBL flip 2) and computer speaker, and they both can only play out 90db for the computer speakers and 95db for the bluetooth one. Interestingly enough, the JBL seems quite loud compared to the SPL reading, as I guess they use heavy compression and also increase HF response in EQ at high volumes to maximize output at higher volumes.
T/S parameters of the Beats Pill XL driver using REW:
| fs 100.1 Hz Qms 4.339 Qes 0.530 Qts 0.472 Fts 212.1 Mms 3.04 g Cms 0.831 mm/N Rms 0.441 kg/s Vas 0.45 litres Bl 2.294 Tm Eta 0.08 % Lp (1W/1m) 81.41 dB Dd 5.00 cm Sd 19.6 cm^2 |
From the parameters above, these drivers have a reasonable resonance frequency of 100hz and sensitivity of 81.5 dB for being 2.5 inch speaker drivers. These drivers have a small Vas, as well as a lower qts of 0.47, making it suitable for bass reflex type enclosures, and the vas shows that the drivers can work in suitably sized enclosures such as the one designed above. Therefore, these drivers are suitable for small bluetooth speakers or desktop speaker projects, but projects requiring more than 90 or so decibels of output at maximum output should look into drivers of larger diameter (such as drivers as large or larger than 4 inches), since such drivers have a higher sensitivity and is easier to drive louder with the same amount of power. However one thing to note is that such drivers would need a comparatively larger enclosure as well, bringing us back to the basic compromises when designing a speaker enclosure (the iron hoffman law).
Comments
Post a Comment