BP1048 Based DSP Bluetooth board
Note for the reader: this post is rather vague but I just wanted to publish another project I have done. Sorry if this post is not as concise or detailed as the others
Despite all the features, the main reason I wanted to build this board is its DSP processing. In particular, the virtual bass (missing fundamental effect) is one of the most realistic ones I have ever heard. It sounds like the Waves Maxxbass MX3000CS ICs, but with way better SNR!
The story behind why I wanted to build this board - and why/how I sought after to achieve the missing fundamental effect:
So, I wanted to build a small, loud, and efficient speaker. But unfortunately, as I discovered later, it is impossible to do so due to the iron Hoffman law. The Iron Hoffman law is an Iron law that states that you can only have 2 of 3 desirable traits when one wants to build a speaker, which are small enclosure size, high efficiency (thus sensitivity), and low bass extension. This means that if you want to build a speaker that was small but had a great low bass extension, you would have to sacrifice efficiency as the cone of the driver has a greater moving mass and excursion, thus reducing sensitivity due to more weight on the motor and a wider voice coil gap for a higher maximum excursion capability. This is the most commonly accepted sacrifice the audio industry uses today.
However, look at the JBL SRX812; it has a 12-inch driver and a horn-loaded compression driver. This allows the speaker to play very loud with little power and decent bass response down to 60hz.
Here is another example: Observe the 2 speaker drivers below. The driver on the left is one from a Apple Homepod smart speaker, while the other one is an Atlas Sound FC-104 PA speaker used in ceiling speaker applications (with a back so that the speaker does not work in infinite baffle but works well in either ported or sealed applications). The home pod driver only has a sensitivity of 78db at 1 watt, but the Atlas Sound driver has a sensitivity of 94db at 1 W, a 16 db difference! This is more than twice the volume at the same power input. However, the home pod driver has a free air resonance of 63hz, while the Atlas sound one has one around 125hz due to a lighter cone. The Homepod driver has an excursion limit of 10mm one way, while the Atlas Sound driver has an excursion limit of around 3mm, leading the Homepod driver to have a higher power handling of 50w compared to the Atlas driver of 15w.
These 2 drivers illustrates the concept of the iron hoffman law.
However, I believe otherwise. I wanted a loud and small speaker, but I wanted something that still had a decent bass extension, which is what I sacrificed according to the iron law. But then I discovered a unique workaround for this limitation called the missing fundamental effect.
The missing fundamental effect is a unique psychoacoustics effect that tricks your brain into receiving notes lower than the speaker produces. It creates 2nd, 3rd, and 4th harmonics of the fundamental note, in which the fundamental note is filtered out. When those harmonics are played together, your brain still hears the original note as our brain fills the note in, just like how optical illusions work.
Unlike typical bass-boost effects, the missing fundamental effect does not require more headroom power to accommodate for the positive gain of EQ filters, increased excursion of the loudspeaker driver, or more motor strength (BL) from the speaker voice coil and magnet. Therefore, this effect can be used to my advantage so that I can use high-efficiency speaker drivers of small size (4-inch drivers) that have trouble reproducing frequencies below 100hz in a 10L ported enclosure and use the effect to give me bass extension down to around 40hz, allowing me to technically break the iron Hoffman law
Therefore, I looked at the internet for solutions. I first heard about this effect through a MaxxBass article in EE Times. I wanted to find out how I could obtain this effect. I stumbled upon the MX3000CS IC, which was an ASIC consisting of a DSP that synthesizes the harmonics. However, at that time, it was impossible to get that IC, and I also lacked the ability to design circuit boards. I looked into DSP solutions like the Analog Devices ADAU1701, which has an effect called super bass, which uses the missing fundamental effect. However, the effect started to distort at low volume. While it does increase the speaker's bandwidth, I faced fatigue after some time due to the muddiness of the sound inherited from the distortion from the DSP clipping.
After that, I used a MvSillicon-based DSP amplifier that used the TPA3255 class D chip amp. At 48V, it can output 2x300W into 4 ohms, some pretty significant power. However, I didn't need all that power so I used a 19V 90W laptop power brick that could provide around 40W per channel into a 4 ohm load since that is around what my speakers are rated at. And wow, I was impressed. The effect had no distortion and sounded extremely convincing, to the point I felt there was a hidden subwoofer somewhere in the room.
Now I can listen to the effect for hours, enjoying my music without fatigue!
However, I wanted to find a single-board solution, something that I can just bring and plug in to any system. I stumbled on an online store where they were selling old stock amplifier boards with the MX3000CS chips on them so that they could get rid of their inventory. Since the power of the boards was very underwhelming for me (2x8W at 12V, meh), they still contained the chip that I had been looking for for a long time, so I decided to buy 8 of them in order to take the IC and other components to use them in my board design. These chips are so rare that I couldn't find a credible seller selling genuine Maxxbass chips. I believe that those listings were some chip of some other use (maybe some multiplexer chip or OTP microcontroller chips obtained at low cost and had the marking removed and re-engraved with some fake markings on it), since fake chips are ubiquitous where I live. Even standard chips like real NE5532 opamps were hard to come by and were likely re-badged LM358s, LM741s, or 4558s, some low slew rate opamps with poor audio characteristics. So the best idea is still to get some chips off some scrap board since production model boards will be very unlikely since they directly ordered the chips from the chip-supplying companies or creditable distributors.
Thus, I built a couple boards using the Maxxbass MX3000CS IC.
I really like its simplicity, kind of like how it takes little to implement other analog audio processor ICs such as the PT2258 or the other PT series EQ/tone control ICs. The PCB has gone through a couple revisions, but I stalled this project as even though the bass effect sounded impressive, there was way too much hiss. So much so that it is impossible to get a usable level out of this board without introducing a significant noise floor on an amplifier (although the datasheet states 90 SNR, it felt like less than 60 DB SNR, with an output of 0.4V compared to the 1.5V indicated on the datasheet). This is even worse than an ADAU1701 Wondom board, with no gain staging, boosted the signal by 10db, and the signal output amplified by 2x by SGM4917 headphone amplifiers. The hiss from that is already unbearably connected to an amplifier, and this is even worse than that. Thus, I looked for another solution, especially about having one IC that can handle EQ and other functions. I thought about the BP1048 board I repurchased a few months, and there came the idea of making a board to solve all the problems I faced with this board that caused me to not want to use it back then.
To solve the problem, below shows the baseboard I built for this module. Unfortunately, I cannot build a whole board from scratch as this IC is quite a proprietary one, even it's datasheet is not easily found and the programming tools (including the compiler and hardware writing tools) and SDK required to program are above my ability and something that the company will not give to hobbyists. So I could only buy pre-fabricated boards that had a program already written on them. Despite that, the program works extremely well and the boards cost only 10 USD each, exceptionally cost-effective for such functionality!
This board in particular provides a hardware interface for mute control to prevent loud pops and transients from source changing as well as startup/shutdown (very unbearable especially when connected to high power systems), and to amplify the signal for a 2V output (useful for pro audio). Despite my difficulties in this project, such as grounding and RF issues, the final design works extremely well!
Schematic:
Layout (notice the separate analog and power ground)
PCB rendering:
Final outcome:
Tuning this board in ACPworkbench:
Comments
Post a Comment