how to connect speaker to a sine wave generator

[giorgik]

Hello everyone.
I need to connect a speaker 8 ohm 0.25W to a sine wave generator that has an output impedance of 600 ohms and output value (adjustable) between 0.6V and 2.8V.
Below I attach scheme.
What kind of circuit do I connect these two elements ?
I see you could make a circuit that allows this link and if possible explain?
Thank you for your attention

[IanB]

The box between the sine wave generator and the speaker is an audio amplifier.

[giorgik]

OK IanB, up to here I know myself.
The amplifier design must be such that the sinusoidal signal reported from the speaker has the lowest possible distortion. This circuit I need to generate the sine waves in the range of audio frequencies (20Hz - 20kHz) to be read without distortion by an electret microphone capsule. I must therefore detect which frequencies the microphone detects me without distortion.
Hence the problem of designing an amplifier BF you send the sinusoids from the generator to the speaker without distortion in the range 20Hz - 20kHz.
Can you help me to make it happen ?

[xrunner]

Hello everyone.
I need to connect a speaker 8 ohm 0.25W to a sine wave generator that has an output impedance of 600 ohms and output value (adjustable) between 0.6V and 2.8V.

You need an audio transformer -

[giorgik]

xrunner, as I get this audio transformer?

[TimFox]

Any "normal" audio amplifier can accept the voltage from the 600 ohm generator and feed an 8 ohm speaker.
If you use a high-power amplifier, you risk blowing up the speaker accidentally.  Also, running a high-power amplifier at very low output power will suffer from noise and crossover distortion.
You need to specify the problem correctly:  how much power do you need and how much distortion can you accept?
If you use the 600 ohm to 8 ohm transformer, you can obtain about 13 mW mean power into the 8 ohm load, with 2.8 V rms into 600 ohms.  If the 2.8 V rms you quote is into an open circuit, but the generator has 600 ohms output impedance, then the power will drop by 4:1 to about 3.3 mW.  Is that sufficient for your experiment?

[xrunner]

xrunner, as I get this audio transformer?

Where do you obtain one? Oh lots of places - like Ebay. But As stated by TimFox, we really don't know enough to help much more.

[giorgik]

OK , This is the circuit that uses my electret microphone capsule (see attachment).
I must be able to capture all the sounds in the range 20Hz-20kHz.
Therefore I must first know what frequencies the microphone is capable of capturing.
So I thought about using a sine wave generator of these frequencies by connecting it to a speaker, and then using an oscilloscope connected to the microphone capsule to read the frequency range that is able to capture.

[Richard Crowley]

Virtually ANY amplifier will have orders of magnitude LOWER distortion than ANY kind of speaker. You would have to be pushing it to find an amplifier with > 0.1% distortion.
And it is virtually impossible at ANY price to find a speaker even approaching 1% distortion. So "distortion" is not a problem here, especially with sine waves which is a "best-case" audio signal.

If you attempt to measure the "frequency response" of a typical small electret microphone capsule in the way you described, you will actually be measuring the frequency response of the speaker and the immediate acoustical surroundings.  The frequency response of a typical inexpensive electret microphone capsule is FAR BETTER than even an expensive studio speaker. 

You did not disclose what is your "8 ohm 0.25W" speaker, but I would bet that it has very poor frequency response and quite high distortion as well.  So it will not be even remotely suitable for attempting to "measure" the frequency response of a typical small inexpensive electret microphone capsule.  In fact, it is typical to do exactly the opposite. To use a small, inexpensive electret condenser microphone capsule to measure the frequency response of a speaker.

[Zero999]

Yes it's true, even small inexpensive electrect microphones have good frequency response and low distortion, compared to a cheap speaker which will be horrible.

You could just connect the speaker directly to the output of the signal generator but it will be attenuated by a factor of ⁸/₆₀₈.

If you use an amplifier, it doesn't need to have any voltage gain, just current gain. A unity gain buffer, with a gain of 1, a low impedance output and capable of driving 8 Ohm would do.

[ciccio]

As other member have written above, your objective is unrealistic, because the frequency response range of  ANY electret capsule is wider than that of any small speaker.
Measuring the frequency response of a microphone is not trivial, it requires a wide range speaker whose frequency response is known (surely not the one you are proposing), an anechoic chamber, suitable amplifiers (both for the speaker and the microphone) and a suitable meter or an audio analyzer.
I any case, what you are searching  is a small power amplifier (a good audio transformer could do the job, but it will be an order of magnitude more expensive than the amplifier).
Search e-bay for "audio amplifier board" and add a simple power supply.
You can find a suitable one for some dollars.
Best regards

[sdg]

Therefore I must first know what frequencies the microphone is capable of capturing.

The constructor of the microphone will be able to answer that.

[Mechatrommer]

D2025
TA8225H
PT2313L
TEA2025B
PT2314
search in http://www.datasheetcatalog.com/
otherwise open ebay type "audio amplifier chip" enter

[giorgik]

I thank you all for your answers.
So for the circuit I posted (with amplifier LM358) does not make sense to make these measurements ?
This project was born from the idea of wanting to design a system for hearing my robot rover, which should (through a circuit diagram that I still have to think well) to be able to recognize certain sounds (artificial intelligence) that cause a sense of fear or curiosity.
So I should determine the amplitude levels for certain frequencies output from the preamplifier I posted. Even here I do not know well how to set up the project ... ehmm then if you can give me some idea ... )

[Zero999]

There's nothing wrong with that circuit. The LM358 isn't the best choice for audio but it will work. The slew rate will impose limits on the higher frequencies and it's quite noisy. The NE5532 would be a better choice but you may need a negative supply.

The main issue is with the idea of using a speaker to measure the frequency response of an electrect microphone.

Have you decided what you're going to be using as a processor?

The analogue signal needs to DC biased to the middle of the ADC range and the gain should be selected so it doesn't clip but gives decent resolution.

[Richard Crowley]

The frequency response of the microphone is absolutely the VERY LEAST of your concerns. IMHO, you shouldn't give it another thought. You are concerned about completely the wrong thing here.   

The biggest problem with speech recognition (or even more simple "sound" recognition) is ACOUSTICS, which is a VERY COMPLEX problem. Your ear-brain system automatically handles an unbelievable amount of "signal-processing" separating the SIGNAL of interest from the attendant NOISE of the "Real World".  Even with absolutely perfect pickup of exemplary speech (which is virtually impossible in the Real World), speech recognition is still of quite low quality and reliability.

There are already quite inexpensive board-level products that have an electret mic capsule and the  mic preamp, ready to feed into an analog input of your microcontroller. You are re-inventing a wheel that is already fully developed.  You will have an overwhelming project just devising code that will recognize the sounds without also re-creating a circuit that is already perfected and available dirt-cheap.

For example:  SparkFun Sound Detector SEN-12642  https://www.sparkfun.com/products/12642

[Mechatrommer]

so AI huh? interesting! i thought about this long time ago... to distiguish certain sound to build fear factor... now what i thought was and still is... fpga setup to separate each freq and magnitude in fourier transform, and then pass the parametric to mcu for AI processing, probably genetic or neu-net algorithm whatever et al. by analyzing individual freq we may distiguish its john or marry. but its only a dream because i dont know fpga, you can do all in mcu but its slow and serious programming there. i bought few AI books, i never have time to read and get my brain around it... i did quick read some of them, but i still never get my brain around it...

[SL4P]

you can make the measurements, just not at the speaker...
use an electrical reference, and do your analysis as close to the source being tested as possible ... the mic or amplifier.

even then, you may want to visit a cal lab to profile your circuit's response, and then design in some *fixed* EQ to flatten out the 'known' response, or compensate for that in your readings.

keep in mind that amplitude and non-linear gain will also introduce inter-modulation distortion - especially in slower devices (like the speaker!)

if you change *anything* you need to recalibrate from the start.
that's why studio0/reference gear is so expensive!

good luck learning.
ref to richard's comment... look up psychoacoustics, that is part of what he was referring to.

[giorgik]

  thanks for your advice