Making a Smart Speaker - Part 1

Smart speakers are the "next big thing." You can tell because Apple started to make them, and they have such a good track record with the next big thing (not recently, but whatever). There are DIY solutions popping up all over the place, so I wanted to try and make my own as much from recycled electronics as possible. I have the things to make this happen (for the most part), so let's make this happen!

A year ago, my friend gave me a set of Logitech S-00040 speakers with the right channel not working. I didn't have any computer speakers other than the ones built into my monitor (those always suck), so I said I'd take them. Now I have much better speakers of my own, I have no use for partially functioning speakers. So I figured I would try and make a smart speaker. But we've been through this.

This is the subwoofer of the speaker set, set up for the frequency test. 

For this first part, I want to modify the subwoofer so that it can deliver the full range of frequencies. Just for fun, I did a quick and dirty experiment to find the frequency curve of this sub compared to my computer speakers. So I played a 20Hz - 20kHz sine sweep through both and recorded them with my Shure something-or-other microphone. They're not clean results, but we can see what we knew all along:

The audio actually clipped, the low frequencies boomed so loud. But as you can see, there's clearly a frequency filter here. I think there are in most subwoofers. I realize bypassing this circuit is dumb because that removes all kinds of audio quality because of how much a single speaker can produce at once and the physical construction of the speaker, but we'll see how it does. So let's open it up and find where this cut off takes place.

Here is the inside of the speaker and then the board removed. There isn't much to it as it's basically an amplification circuit. The chip and the chip under the heatsink is a BT2025BH which, if it's the same as the TEA2025 is a stereo amplifier. So where is the lowpass happening?

The answer was on the underside of the board with some SMT passives. The low pass filter was made from R19 and C10. I don't know what the capacitance of C10 was, but R19 was 15 kOhms. When I shorted R19, the low pass went away. So I bridged the resistor and removed the capacitor. The resulting sound wasn't great, especially with bassy songs (my test was "Godspeed" by Tech N9ne), but the speaker was producing more frequencies.

Because my plan wasn't to listen to music from this (maybe the additional speaker could be for this purpose), I tested the standard Google voice and it sounded okay. The final test was to see how the unit sounded once I put it back together.

That test failed. All improvements I had made by removing what seemed to be the low pass filter had been lost. I can only assume that it's because of the construction of the speaker. Most of the inside is empty space and the air movement driven by the speaker is channeled out the front with the horn like shape. Since air only really moves when there's heavy bass, that's the sound that's thrown out the front.

The obvious answer to this problem is to some how incorporate the other working speaker (the left channel) into the main body of the subwoofer (that's the form factor I want to keep). This involves, unfortunately, having to cut open the smaller speaker. I'll cut open the right one just in case I cut too far and break something. I think they're just wires to speakers in this case. We'll open those up once my dremel tool comes in the mail. At least at this point we know what's inside of the speaker and roughly how it works. I'm okay with this expedition being a waste of time as long as it contributes to the end goal, which it did.