I know, right? Ready-made modules usually don't have enough filtering caps and have no mounting holes.

On the inside with the laptop's current speakers - probably not worth it. But it can be used for an external sound device powered from USB

Would a relay work in your case?

Hi! Not yet. I'm not sure how to measure it yet :)

@@LeftyMaker Room EQ Wizard can do it. It is set up for using with USB measurement mics but I'm sure there is a way to do it using any decent mic. Your response curves won't be calibrated without a measurement mic, but the harmonic distortion is more likely to come from your amp and speaker than from the mic (unless you use a really cheap or beat-up one.) You can compare results with other speakers which you like the sound of.

This is not what one might call a low distortion amplifier, at THD 0.1% typical as per datasheet and recommended test circuit, it's a little up there. But it's also below what are often considered audible levels and foremost the nonlinearity of a tennis ball speaker is probably going to be a lot higher than possible nonlinearity from such amplifier (excluding clipping), in particular if you want it to be somewhat loud you'll be running into driver nonlinearity due to issues with voice coil magnet coupling (xmax related) that will be easily several % plus you have intermodulation from the driver, plus compliance related distortion from the speaker cabinet as it were. The main THD contributor in this amplifier to the mediocre figure (as long as you don't hit clipping) is basically overlayed broad spectrum switching noise, vaguely blue noise shaped, visible on the rising THD up to 2-3% at very low volumes, which i mean, you sort of expect it don't you that if you put your ear to it, you'll hear the distinctive noise, the percentage figure isn't all too meaningful if the resulting noise is near or below absolute hearing threshold at intended listening distance. It's a little different from pure added noise since when there's no signal at all, there's not necessarily noise either, it's triggered by useful signal. Similar to quantisation noise, and it sort of is a variant of that, since though the drive circuit isn't time discrete, it's state-discrete and also has a minimum possible state transition times to observe. I don't think MAX4410 is going to maintain high linearity when used in this way as you're doing so now either, say with a 4 Ohm speaker, i think it's in pain. It is however probably going to stay relatively low noise, much lower noise than any speaker purpose chip amps. Of course there's all sorts of THD sources and it's an endless question which of them are audible and objectionable under which circumstances - various overlayed noises, zero crossing distortion, decay in linearity near peaks at increased amplitude, symmetrical vs asymmetrical. Cute project though for sure

@@SianaGearz I failed to mention that this was for an outdoor art project: the tennis ball speaker goes at the focus of a 1m parabolic reflector. Out in the woods there is so much background/wildlife noise, it wasn't that critical. I would like to move the project to larger spaces, indoors or out. I've gathered what I need is an inverting op-amp on one of the 'stereo' channel inputs, then I could run the MAX in bridged mode. I have crudely measured near-field response using REW. The HD is there but it is pretty low. The frequency response is about as good as you might expect -- not flat but intelligible. In the reflector, the 'bass' at 150 Hz is not very directional. The highs are very much so. I need to get motivated to measure in the far-field.

@@mog068 MAX will still be in pain and even more so in BTL configuration, since the effective load impedance is lower! The current draw correspondingly rises dramatically. Consider that it was only specified for 16 Ohm impedance load not lower. This NS4150 and any number of small low cost chip-amps are already internally BTL, this also goes for various Diodes Inc (PAM) and Fuman speaker amplifier chips, and they master 4 Ohm speakers in BTL configuration (2 Ohm effective) with absolute ease.