Designing an amplifier for a 500mW, 8Ohm Speaker

[mritchey99]

Hey all, this is my first time on this forum and I really need the help! I'm currently designing an amplifier for a 400mVpk signal (wav file of the human voice coming out of a PCM channel of an STM32 MCU.) I'm using the Taoglas - SPKM.17.8.A - 17mm Mini Speaker.

This speaker is quite small (17mm in diameter) and after hooking it into my LM386-driven amplifier circuit, it's still very quiet. In fact, using a signal generator, even with a high voltage and 1kHz sine wave, the speaker is too quiet. The speaker needs to be loud enough to be heard when worn on the wrist.

I'm having trouble understanding why the speaker is still so quiet even though it's larger than many speakers I find in say a phone or computer monitor. I've started to wonder if this is more of a mechanical and physics issue. Do I need a baffle for the sound it's producing to be more audible or is this speaker not meant to be used in the scenario?

I couldn't figure out how to upload a picture but I'm using the LM386 amplifier from TI's datasheet (page 10):

www.ti.com/lit/ds/symlink/lm386.pdf

 

[crutschow]

Have you measured the signal levels into and out of the LM386?

 

[mritchey99]

Have you measured the signal levels into and out of the LM386?

Yes, I get a roughly 1.8V peak on the load. The input is the 400mV signal I was describing. Is that what you're referring to?

 

[KlausST]

Hi,

I'm not sure where the contradictions come from.
0.4V peak, 8 Ohms and 500mW does not match.

Thus I don't know how to help. Do you need assistance in learning formulas, physics, math .... or more in circuit design?

If the speaker is too quiet, then:
* maybe the amplifier is not suitable
* the amplifier input level is too low
* the speaker's efficiency is too low
* or your expectation is wrong

From simple mathematics (you should be able to do on your own), it's clear the 400mV can not result in 500mW.
Now we don't know where and how you measured the 400mV.
If you talk about an 8 Ohms speaker for 500mW, you also need to measure the voltage (directly) at the speaker.
Often the speaker values are specified at pure sine waveform. But human voice is not pure sine. Thus for a 0.5W average speaker power on voice I recommend that the amplifier is able to produce at least 3W (continous sine, or 6W peak) to get low distortion result.
Mind: Don't test the speaker with more than the specified 0.5W average.
Use a bandpass filter in front of the amplifier to adjust the (amplified) frequency range to the speaker.

I guess you need to give use more informations. A sketch about your circuit, and where you measured / expect the values is very very helpful.

Klaus

 

[FvM]

1.7 Vpk (about 180 mW into 8 ohm for sine) can produce a lot of noise with a good speaker. Either your speaker has low efficiency or the drive signal has most spectral components outside the speakers frequency range. What kind of signal do you want to transmit? What's the available amplifier supply voltage?

 

[mritchey99]

Hi,

I'm not sure where the contradictions come from.
0.4V peak, 8 Ohms and 500mW does not match.

Thus I don't know how to help. Do you need assistance in learning formulas, physics, math .... or more in circuit design?

If the speaker is too quiet, then:
* maybe the amplifier is not suitable
* the amplifier input level is too low
* the speaker's efficiency is too low
* or your expectation is wrong

From simple mathematics (you should be able to do on your own), it's clear the 400mV can not result in 500mW.
Now we don't know where and how you measured the 400mV.
If you talk about an 8 Ohms speaker for 500mW, you also need to measure the voltage (directly) at the speaker.
Often the speaker values are specified at pure sine waveform. But human voice is not pure sine. Thus for a 0.5W average speaker power on voice I recommend that the amplifier is able to produce at least 3W (continous sine, or 6W peak) to get low distortion result.
Mind: Don't test the speaker with more than the specified 0.5W average.
Use a bandpass filter in front of the amplifier to adjust the (amplified) frequency range to the speaker.

I guess you need to give use more informations. A sketch about your circuit, and where you measured / expect the values is very very helpful.

Klaus

Sorry, I'll be more specific. The 500mW is the rated input power of the speaker. I looked at it again on the data sheet and see that it is RMS power.

The signal ACROSS the load is reading on an oscilloscope as a 1.8V peak. I will measure this again when I get the chance but it was definitely around 2V, slightly lower, hence why I say 1.8. The 400mV was measured at an output pin on my MCU (the pin associated with the PCM.

What I am most curious about is when I plugged the speaker into a wavegen and tested it at higher voltages with a 2kHz sine wave and the volume still wasn't loud enough for what I needed. It's definitely louder than the wav files I'm playing (short snippets of my voice), but I'm not sure if I need to come up with some sort of physical housing for the speaker to vibrate to help produce more sound pressure.

Another question I have is, how can I reduce the clipping at the top and bottom of my output signal? The supply voltage changes nothing other than overall gain (clipping still exists).

Note: The actual quality of the audio doesn't need to be that great.

 

[mritchey99]

1.7 Vpk (about 180 mW into 8 ohm for sine) can produce a lot of noise with a good speaker. Either your speaker has low efficiency or the drive signal has most spectral components outside the speakers frequency range. What kind of signal do you want to transmit? What's the available amplifier supply voltage?

The goal is to play short audio files of human speech. I'm using a 5V supply voltage. That's the most I have available to me for this design.

Edit: The frequency response of the speaker is described in the data sheet as 100 Hz - 10kHz

 

[KlausST]

Hi,

but I'm not sure if I need to come up with some sort of physical housing for the speaker to vibrate to help produce more sound pressure.

Generally a speaker without housing produces a lot fo "short circuit" for the sound waves. The high pressure in the fron gets compensated by the low pressure in the back.

Another question I have is, how can I reduce the clipping at the top and bottom of my output signal? The supply voltage changes nothing other than overall gain (clipping still exists).

I still don´t know at which level you need help.
Do you know what "gain" means? What gain you need? What gain your amplifer has?

Can you do "power calculatons" on your own, or do you need help with this?

I mean: The datasheet figure 6. tells you that even with a 6V power supply you get less than 400mW ... accompanied with a lot of distortion. You sould be aware that if you reduce the power supply to 5V then the output power will be less.

Are you aware that the scope diagram shows a sinewave input but a square wave output?

Klaus

 

[Audioguru]

1) With a supply that is only 5V then the datasheet for the LM386 shows that output power is only 0.14W into 8 ohms at clipping. With a 9VDC supply, the output power is 0.56W at clipping.
2) The 10uF capacitor between pin 1 and pin 8 causes the amplifier voltage gain to be 200 times!
Then severe distortion occurs if the input is higher than 7.5mV peak. Your 400mV peak input is 53.3 times too high.
3) The tiny "speaker" is actually a squeaker since its frequency response does not say the level loss at 100Hz. The frequency response of most speakers is usually rated at -3dB. The squeaker's output level at 100Hz might be -40dB which is almost nothing.
4) The sensitivity of the squeaker is 96dB at 1kHz at 500mW at a distance of 0.1m with a baffle. Most speakers are spec'd at 10 times that distance. Does the squeaker have a baffle to prevent low frequencies from its rear from cancelling low frequencies from its front?
5) The input to the volume control must not have any DC on it. A 1uF input coupling capacitor should feed the 10k ohms volume control to block any DC voltage.