I know, I know, I promised I’d publish this post months ago…but I’ve been very busy with my thesis (Oh, by the way, I got my master’s degree last month! Anyone looking for a Qt/QML developer? :D) so I had to postpone this until now.
But better late than never, right?
If you haven’t read the previous post, let me sum up what I’m trying to accomplish here:
basically, Jolla’s loudspeaker SUCKS, it does, really
I thought it could be fun to buy a few genuine loudspeakers from phones which are known to have loud buzzers and try implanting them in my Jolla, just to see if it would actually make a big difference.
I used GsmArena’s loudness tests (they do it for all phone reviews) to select 9 loudspeakers and then bought them from online shops.
What I did was:
1) Open the phone
2) Replace the original loudspeaker with one from the test batch
3) Record a video of my Jolla playing Jolla’s ringtone via the new loudspeaker
4) Use Audacity to analyze the audio stream and collect the raw spectrum data
5) Import the spectrum data on plot.ly to generate an interactive plot!
6) Analyze/Enjoy the results
DISCLAIMER: I am no electronics engineer, I am a software developer who loves tinkering with whatever happens to be under his hands. This test doesn’t take impedance or resonance into account, it’s just plug-and-play fun. For example, the amplifier on the Jolla is probably another bottleneck!
The results are a video and two interactive plots.
The video contains all the audio tests, so you can listen to it to form your own idea about the performance of those speakers on the Jolla
The interactive plots provide an easy way to compare the loudness of Jolla’s original loudspeaker to the ones in the testbed. You can pan/zoom and show/hide some of the loudspeakers in order to make a clean comparison between only 2 datasets, for instance.
PLEASE HIDE SOME OF THE CURVES TO BE ABLE TO ANALYZE THE DATA, OR YOU’LL JUST SEE A MESS IT’S ENOUGH TO CLICK ON THE LEGEND ITEMS TO HIDE THE CORRESPONDING CURVE
IFRAME WARNING (sorry)
Due to the fact that this plot uses a logarithmic scale (because of the dB values), it may not be easy to understand how one speaker compares to another one (i.e. -50dB vs -25dB doesn’t mean that the latter is 2x louder!).
For this reason, in the next step I converted the data to a more easily readable format.
Taking into account that the results are measured in dB using dBFS, we can give a rough estimate of how Jolla’s loudspeaker compares to the other ones by using the formula:
this will give us the amplitude ratio, i.e. how much higher/lower the amplitude of the signal created by the competitor buzzer is, using Jolla’s loudspeaker as a reference (0%). A value of 100% means that the signal amplitude was 2x than Jolla’s one on that particular frequency. Using the toggles to hide some of the curves is particularly useful to be able to get a better visual over the results
Try deselecting all the results but the ones from Jolla and Xperia E1 by tapping on the corresponding legend entries, for example
The amplitude is related to acoustic energy, so a higher amplitude corresponds to a louder sound. (more info here)
As we can see, Jolla’s loudspeaker is pretty weak…there are some buzzers which manage to get below its level at some frequencies, but some loudspeakers are a lot louder at other ones. Xperia E1’s produces a signal whose amplitude is a massive 600% higher than Jolla’s, at 8441 Hertz frequency!
Now, the next steps…
- Choose the best one
- Fit it inside Jolla’s housing
- Find a way to connect the (different) loudspeaker pins to the motherboard audio out pins
Anyone who feels like helping out?