I'm trying to reconcile this speaker's performance with the datasheet. The speaker is 0.5W and about 40mm in diameter.
Product link: https://www.taydaelectronics.com/mylar-speaker-8-ohm-0-5w.html

Using my signal generator, the speaker is *significantly* louder at 3500Hz than any other frequency, but the datasheet says it has a resonant frequency of just 380Hz. Does this make sense? And I get similar results when generating sound from a microcontroller.

Pertinent info from datasheet:

 

The response graph is with the speaker mounted in a sealed box or "baffle" as they call it - not free air or sitting on your desk.
Loudspeakers need an enclosure to prevent the backwave from interacting with the frontwave.
This interaction cancels almost all low frequency, bass response so it will naturally sound tinny without the back side muffled or in an enclosure.
The self-resonant frequency will also shift, which I think is what you are observing.
Even a cardboard box helps. I cut fabric from old socks to absorb the backwave. You can try experiment.

 

The response graph is horrible with frequencies below 300Hz being attenuated. I suspect that they cheated and the flat level shown below 100Hz is noise but the graph should show the diagonal line between 100Hz and 200Hz as a diagonal line continuing down to 20Hz at 20dB (inaudible).

The graph shows a peak at its resonance of 380Hz and another peak at 3.8kHz. The maximum of 10% distortion is very high.

 

Thanks for the feedback. I made a quick cardboard enclosure and it did drop the resonant peaks.

 

Thanks for the feedback. I made a quick cardboard enclosure and it did drop the resonant peaks.

An enclosure increases the frequency of the lowest resonance and does not affect the high frequency peak.
The enclosure must be solid and rigid so its sides do not vibrate. NOT cardboard.
The enclosure should ne large enough that it does not increase the free air resonant frequency too much.
EDIT: Here is a graph showing the affect of speaker enclosure size on a normal fairly large speaker:

 

Hmmm. I must have been hasty in the testing.

 

Why are you using such a cheap tiny poor quality speaker?
A speaker size of 3" can make fairly good sounds but does not cost much.
Here is a quality 3" speaker with a free air resonance of 135Hz so it could be in a fairly small enclosure and produce good sounds down to about 150Hz.

 

I was suggesting cardboard just for a quick test, you can hear the effects just listening to music to better understand what the issue is, as far as low end bass response. I have some of these mylar speakers and they are basically terrible without an enclosure or baffle that is well sealed around the speaker's diameter. The mylar plastic cone is pretty peaky.

Another aggravating factor for the HF self-resonance is how good the speaker's driver circuit is - you need low impedance push-pull drive to not allow the resonance to take over. Some signal generators are 600ohms output impedance and aren't great for a frequency sweep of an 8 ohm speaker. If you are driving it with PWM from a MCU, a single transistor does not really work well.

 

The small speaker is all I have right now. I have one these, but I'll have to look for it.

https://www.radioshack.com/products/radioshack-mini-audio-amplifier

 

I also have one of those awful-sounding Radio Shack amplified speakers.
It will play your buzzer-piano but not loudly.

 

The Parts Express speaker looks good. For power amp, I'll pick up these, because they're suitable for uC voltages:

https://www.taydaelectronics.com/ncp2890-ncp2890dmr2g-audio-power-amplifier-ic.html
https://www.taydaelectronics.com/te...lass-ab-4-7w-x-1-at-8-ohm-3v-12v-pdip-16.html
https://www.taydaelectronics.com/td...el-class-ab-1wx1at32-ohm-1-8v-15v-pdip-8.html

 

The Parts Express 3" speaker is 4 ohms. Some of the very old amplifiers you list cannot drive 4 ohms properly and are obsolete or are now made by a Chinese company in Taiwan. they produce very low output power.

 

Using my signal generator, the speaker is *significantly* louder at 3500Hz

Am I seeing things, or does the response curve show about a 12db peak right around the 3500 Hz that you say it the sound peaks at?

Bob

 

Am I seeing things, or does the response curve show about a 12db peak right around the 3500 Hz that you say it the sound peaks at?

Yes, I said it in post #3. +12dB is a very high and obvious peak for a speaker.