My not-so-mini, not-so-blue-tooth speaker

30th October 2014

As I mentioned in previous posts [1, 2]. I recently purchased an S10 blue-tooth speaker to improve the sound-quality of my phone.

S10 speaker
size of a small apple, weight: approx 200g

On receiving it I found it had terrible sound-quality and decided to rebuild it by disassembling it and making it a new enclosure.

Chapter 1, new enclosure

I went to a local shop and found a nice looking piece of wood for 30p. Then I worked out what was the biggest box I could make out of it and made it, with two circular holes cut by a fretsaw for the transducer and a passive radiator.

speaker in new enclosure
width: 10cm, height: 26.7cm, depth: 13.4cm, weight: 492 grams,

There is a 2cm high alcove at the bottom to contain the circuitry. I made the passive radiator by putting a circular cut-out of wood onto the secondary diaphragm I got from the S10 and gluing a stone on the back to give it a weight of approximately 128 grams. No maths went into the design of this, I just designed it off the top of my head.

The result

The speaker was now 18 times as big and more than twice as heavy as before. The sound was much less harsh, a definite improvement. However, I was still not happy with my speaker for three reasons:

  1. Very poor bass response, the passive radiator did not work at all.
  2. High pitched squealing sound when music was playing.
  3. Blue-tooth connection very glitchy, music intermittently cutting was extremely annoying.

I then began analysing the circuitry to find out the cause of the squealing. When found a chip called an: "8002A IDC1422H" it became clear that this circuit uses one of those switching amplifier things. I'll briefly explain the concept of how these things work (although this might not be exactly how the one in the S10 works).

Imagine we have a dual power supply providing +4V and -4V, and we want to produce the following output waveform for the speaker:

a sine wave

Now at the point marked by the red cross (leftmost for the colour-blind), we want an output voltage of 2V, but we have a 4V power supply, so we need to half the voltage. Now, in a traditional amplifier, it will half the voltage by dissipating half of the energy as heat. But in a switching amplifier, instead it switches on and off extremely fast at a 50% rate to create an average voltage of 2V. This happens at an ultra-sonic frequency. The following is a very zoomed-in image of what is happening at the red cross:

outputs +4V at a 50% duty cycle

Now at the green cross we want -1.3V which is a third of the -4V power rail so it switches the negative power rail on and off at a 30% rate (notice how it's off for twice as long as it's on for):

outputs -4V at a 30% duty cycle

You get the idea. Anyway, this explains the high pitched squealing sound. Being a freak I can hear ultra-sonic sounds and the frequency filter on the output of the S10's switching amplifier clearly isn't very hugely effective.

Now some companies claim that these switching amplifiers can achieve the same kind of sound quality as a traditional class AB amplifier. However, I don't think the amp in the S10 was really designed with quality in mind, rather low power consumption and low cost.

Another doubt I had about the amplifier in the S10 was it's ability to cater for the dynamic needs of the transducer. A speaker transducer is not just a black hole that you can just chuck power at and expect it to do it's job, it fights back. Like an electric motor, it will draw more power when it's struggling, and less power when its moving freely. If you look at the data-sheet of a speaker transducer you will see a graph called "impedance response". This describes how the transducer's electrical resistance changes depending on the input frequency.

Chapter 2, new circuitry

I decided to make an entirely new circuit for my speaker that would use a class AB design and connect to my phone with a cable instead of by blue-tooth. This would solve problems 2 and 3 listed above. Click to see the amp and the power supply I designed. I designed it using parts I had as I didn't want to spend a lot of money on it. All of the transistors are BC337s and BC327s. The Op-amps are a dual NE5532P. This circuit should much better cater for the dynamic needs of the transducer since it has a low output impedance.

the new circuitry is a bit bigger than the old
old circuitry left, new circuitry right

My new circuit unfortunately requires a power supply of at least 10V and the best thing I had lying around was a 12V lead-acid starter battery (LOL). Now this thing can't really fit inside the speaker so it has to go outside.

the enormous battery plugged into the speaker with crocodile cables

The result

Inevitably, the high pitched squealing sound and the glitchy connection were gone. However, what I wasn't expecting was the enormous improvement in bass response and over-all sound quality! This new circuit had improved the sound just as much as the enclosure did! So, despite being made from a rather random selection of parts, and badly soldered by a guy out of practice, this circuit produces some pretty awesome sound quality! It would be interesting to see what would happen if I made a miniature low power version and put it in the original enclosure of an S10.

Now, the circuit produces a voltage gain of something around 0.6, so the speaker won't go hugely loud, but that's not a problem for me since I don't need it to. I could easily modify the circuit to provide voltage gain if I did. Also note that the 4Ω transducer doesn't require much voltage.

my new speaker, plugged into my phone, playing

The only part that hasn't been replaced now is the transducer. Here is a comparison between the original and the new:

Original New
Size, including battery ~190㎤ 4511㎤
Weight, including battery ~200g 2923g
Power consumption, standby 0.11W 0.46W
Power consumption, playing 0.28W 0.62W
Sound quality ☺☺☺☺
Novelty value ✩✩✩✩✩

Chapter 3, finishing off

So I'd made a speaker with decent sound quality, now it just needed finishing off.

To protect the wood, I dyed and polished it. Thanks go to my father for letting me use his work-space and chemicals for this.

It then acted as a beer stand for a while as I contemplated how to satisfy it's preposterous voltage requirements without an external battery.

speaker on floor with a beer on top

Eventually I got some AA battery holders to hold 9 AA batteries. In order to fit them inside the circuitry cavity I had to cut some wood away from the sides of the chamber and trim the edges of the circuit board. I also added a power switch and an LED to indicate when it's on.

The circuitry cavity. The batteries take up 3/4 of the space.
This lot will provide at least 10V even when the batteries start to run out.

So, the project is finally finished :). The sound really is quite nice. It's not small-enough to take out and about but I can easily move it around the house, which is what I wanted it for.

a view of the finished speaker