4

I want to try something audio related and found the following circuit from technoblogy

enter image description here

I built it and loaded a sound clip. Everything works fine, but I found it a bit too quiet and I want to amplify it, but not entirely sure how.

The biggest problem is that I do not have access to an oscilloscope so while this does play sound clips just fine, I have zero idea what the waveform output looks like at the speaker, or even its voltage level (I'm assuming the peak to peak voltage is less than 3.7V) and because of that, I'm not sure how to amplify the output to make it more audible. Or is this even possible to amplify it due to the 3.7V constrain?

cheunste
  • 176
  • 6
  • 1
    One-word: "H-Bridge". – brhans Mar 21 '19 at 17:48
  • 8
    @brhans: Have another look at the circuit. It's already a H-bridge (two hyphenated words) so I think you'll need more words than one. – Transistor Mar 21 '19 at 17:54
  • @Transistor - I meant a 'real' H-bridge which can deliver more than a few mA into a low-ohm load. – brhans Mar 21 '19 at 17:55
  • 2
    @brhans: And that requires some understanding of how to connect a H-bridge output to an unbalanced input. My point is that your comment was useless to the OP given that s/he's a novice. You're usually quite helpful though. – Transistor Mar 21 '19 at 17:59
  • @Transistor - `connect a H-bridge output to an unbalanced input`?? You've lost me there - the H-bridge output goes to the speaker. The 'unbalanced input' (of the H-bridge) is perfect since all you need to do is hook each half of the bridge to one of the GPIO pins... – brhans Mar 21 '19 at 18:03
  • @bhrans: That could work with the right type of H-bridge rather than a H-bridge audio amplifier (which I was thinking of). OP would also need a higher voltage PSU and would need to check logic levels between the two devices. – Transistor Mar 21 '19 at 18:10
  • 1
    What impedance speaker are you you using? The output characteristics of the Attiny's IO mean you need about a 100 ohm speaker, or the outputs may not be able to develop the full voltage. – Phil G Mar 21 '19 at 18:10
  • @Transister - yes, I wasn't referring to an audio H-bridge, just (in it's simplest form) 4 transistors hooked up as a 'raw' bridge, driven directly by the ATtiny's GPIOs. – brhans Mar 21 '19 at 18:45
  • @brhans I have to agree with Transistor. I find these "one word" answers very frustrating. You clearly have a great deal of knowledge and experience to share, I hope you will be able to write a sentence or two. – Elliot Alderson Mar 21 '19 at 20:01
  • 2
    Is your speaker in a box? If not, try putting it in one. Even a cardboard box with a hole for the speaker will improve sound a lot. – jpa Mar 22 '19 at 05:54

2 Answers2

16

Actually, the peak-to-peak voltage is going to be close to 7.4 V. This is because the speaker is being used in a differential mode, connected to two separate MCU outputs, with neither terminal grounded.

Since the firmware is simply driving the two pins with complementary waveforms, you can use just one of them (with a suitable filter, as JRE shows) to drive an external amplifier.

Of course, the speaker will also get somewhat louder if you simply raise the supply voltage to the full 5.5V that the ATtiny85 allows.

However, the primary limitation is not voltage, but rather the current capability of the MCU pins — ±20 mA nominal max, ±40 mA absolute max. If you were to simply add external buffers to those pins, you'd get a lot more sound out of the speaker:

schematic

simulate this circuit – Schematic created using CircuitLab

Note that the buffers are configured as emitter-followers. There's no voltage gain here (in fact, it loses about 1.3V in peak voltage), but a lot more current is available to the speaker. Since it's a digital signal, we don't need to worry about the horrible crossover distortion of this configuration. But it should be able to deliver about 4.8 VPP @ several hundred mA (1.7 VRMS @ 200 mARMS into 8 Ω), or about 360 mW of audio power!

(BTW, this is the solution brhans was alluding to in his very first comment.)

Dave Tweed
  • 168,369
  • 17
  • 228
  • 393
  • The OP's link answers his question in the comments, at least to the extent we have here. – TimWescott Mar 21 '19 at 18:19
  • 3
    I prefer this solution to JRE's because it adds things incrementally to the current circuit without hiding the complexity or changing the general mode of operation. – pipe Mar 22 '19 at 08:53
  • umm, i might be totally mistaken here, but how is a totempole driver (either the one inside the attiny, or the one you just built there to amplify the former) between 0V0 and +3V7 rails supposed to provide >3V7pp? we're not talking AC switches or massive inductances here after all? or did you simply draw the circuit example with 3V7 because that's what OP asked, but write up the 5V5 version to show what the tiny can deliver? – nonchip Jan 18 '23 at 06:48
  • @nonchip: Note that this is an H-bridge driver. It can apply +2.4V or -2.4V to the speaker. In other words, the peak-to-peak voltage across the speaker could be as high as 2x the supply voltage if there were no losses in the drivers. – Dave Tweed Jan 18 '23 at 12:13
  • @DaveTweed but your battery only supplies +3V7 and -GND? oooh sorry i think i see, the peak-to-peak voltage accounts for literally swapping the polarity, yeah that makes total sense sorry. – nonchip Jan 18 '23 at 15:58
4

The site you got the circuit and code from tells you what it is doing.

From the description, you can find out what the output looks like. It says it is using pulse width modulation (PWM) to generate the audio.

Wikipedia has a good explanation of PWM, and how it is used to make audio.

This image from the wikipedia page show pretty much what you can expect as output from your noise maker.

The red line is what your speaker "sees." The blue line is what your circuit generates.

enter image description here

Simply put, your program makes pulses of various widths on the two output pins. Wider pulses cause the speaker to move more than narrower pulses. The pulses have to occur faster than the speaker can respond to them. The inertia of the speaker smooths the pulses into something that resembles normal audio.

What is limiting the volume of the output is the current from the output pins. That is normally measured in milliamperes. Assume you can get 20mA out of your pins. At 7 volts and 20mA, you will get maybe 0.14 watts. Assuming you don't kill your processor doing this.

That's not much power. Compare it to what you expect from your stereo, and how many watts of power it can output.

To get more power, the simplest thing to to is to clean up the PWM signal and send it to a regular amplifier.

You only need one of the outputs pins. Your gadget uses two in attempt to get more power, but it really can't do much.

So, do this:

schematic

simulate this circuit – Schematic created using CircuitLab

Then connect the output to any standard audio amplifier. Buy and build a kit if you like, or connect it to the line in of a stereo.

That is a low pass filter. It filters out everything above 4kHz. I chose 4KHz because the project says to use a file with an 8kHz sampling rate. A file sampled at 8kHz can't contain any audio above 4kHz. Look up Shannon and Nyquist (in relation to sampling rates.)

The pulse rate of the PWM will have to be much higher than 8kHz, so using a 4 kHz filter doesn't lose any of your audio, but should smooth the square edges into a nice smooth wave.

There are small amplifiers that you could power from 3.3V. There are also many other amplifiers that run on higher voltages. There are cheap and common circuits (and kits) to build an amplifier from an LM386. The LM386 needs 5V or more, though. It is commonly operated on a 9V battery.

You could also look up H-bridges. If you drive an H-bridge with you speaker in the middle (the cross bar of the H) using your two processor outputs on the H-bridge inputs, then it should be quite loud. This could be tricky, though. It will be operating at the pulse rate, and amplifying the pulses rather than the smooth wave. That will cause a lot of electrical "noise" and probably make your speaker and the bridge get hot.

JRE
  • 67,678
  • 8
  • 104
  • 179
  • And it's a pretty poor output filter -- the sound will be discernibly muddied with that filter. **But** to make it sound good you'd need an active filter, and that may be beyond the OP's abilities. – TimWescott Mar 21 '19 at 18:22
  • 7
    Guys, there's no point in trying to make it sound "good." It's 8 bit, 8kHz audio generated by PWM. More effort is just trying to make a silk purse out of a sow's ear. – JRE Mar 21 '19 at 18:24