0
\$\begingroup\$

I am designing an battery powered device (3.6-4.1 V regulated down to 3 V) which plays audio recordings through a speaker (8 Ω, 0.3 W). I want to drive the speaker by applying two differential inputs, but what I have is a DAC (20 mA output max) with single ended DC output for generating the signals. So I came up with the circuit below in order to:

  1. Increase the power to ~0.3 W (i.e. the drivers must be able to source/sink about 40 mA each @ 3 V).
  2. Convert the 0-3 V peak-peak DAC signal into a differential signal.

schematic

simulate this circuit – Schematic created using CircuitLab

As you can see, I chose to limit the current via resistors R9-R10 (simulation gives me 79 mA current through the speaker when the DAC is at +-3 V).

I have seen single ended designs where they use a ~220 uF cap in series with the speaker (with the second input grounded). Would it be better to have two ceramic caps instead of the two resistors? I am asking because I believe mismatches in the resistors are much easier to eliminate than with capacitors where tolerances are usually much higher. Also - ignore the name of the op amps (I will use other op-amps in the actual design).

\$\endgroup\$
10
  • 1
    \$\begingroup\$ Do you need to design this from discrete components? Why not just buy a speaker amplifier IC? \$\endgroup\$ Commented Apr 8, 2024 at 6:20
  • \$\begingroup\$ @Justme Its just a preference, most speaker amplifier ics I browsed through seemed to require about the same amount of external passive components, these ic's usually had added features which might be nice if I needed them, but I don't. But I am definitely open to suggestions. \$\endgroup\$ Commented Apr 8, 2024 at 6:30
  • 1
    \$\begingroup\$ @dvd280 The TDA8551 is a 1 W BTL amplifier that works from 5 V. It's cheap and ancient (hopefully not retired yet.) I have piles of them here. Easy to use. And yeah. It includes a "digital volume" and "mute" that you may consider unnecessary. But very few external parts needed. \$\endgroup\$ Commented Apr 8, 2024 at 6:37
  • \$\begingroup\$ @dvd280 I haven't looked into the details of your schematic, but putting series resistors on both sides of a speaker will not be better than on a single one, and even that is a terrible idea: you're limiting the power that goes into your speaker, which is the opposite of why you'd want an amplifier to begin with. Going with opamps of insufficient drive strength instead of an actual speaker amplifier is not going to take you where you want to be, a speaker amplifier. So, your R9 and R10 should have told you that the parts you're using aren't up to the task, and you need to rethink your approach \$\endgroup\$ Commented Apr 8, 2024 at 6:56
  • 3
    \$\begingroup\$ @dvd280 If you have a single 3.0V supply, those opamps are likely useless. \$\endgroup\$ Commented Apr 8, 2024 at 7:01

2 Answers 2

Reset to default
1
\$\begingroup\$

There is no functional difference between any of the following:

schematic

simulate this circuit – Schematic created using CircuitLab

In each case, the loudspeaker will see the same potential difference across it, regardless of the potentials at A and B, and regardless of the order of resistances and speaker. So you might as well choose the right-hand design.

Similarly with capacitors, the same principle applies. The speakers in each case below will all have the same potential difference:

schematic

simulate this circuit

There's absolutely no need to match the resistors, or the capacitors, since that will not alter the voltage across the speaker at all.

With a quiescent input (\$V_{IN} = V_{REF} = +1.5V\$) the two op-amp outputs at A and B might not be exactly equal, due to op-amp mismatches such as slightly different input offset voltages, or due to slight differences in feedback and input resistances.

This error will be small, but will cause DC current to flow between A and B. This will cause the op-amp output swings to be slightly asymmetric, and the loudspeaker diaphragm to be slightly offset from its rest position.

If that's a problem, then you should use a capacitor. As seen above, any of the following designs will behave identically:

schematic

simulate this circuit

It would be silly to implement anything but the design on the right, with fewer components.

Since the polarity of voltage across the capacitor will be both positive and negative, it's best to use an unpolarised model, but since that voltage will be so small, ±3V, even an electrolytic model will probably work (with some distortion).

You can reduce the resistor count for the amplifiers, too. As Bobflux showed, it's possible to use just two resistors (instead of 8) to obtain exactly the same behaviour.

\$\endgroup\$
1
\$\begingroup\$

All you need is one follower and one inverter:

enter image description here

There are several issues with this approach (and your schematic):

  • Opamp idle power usage, even when the speaker doesn't play

  • Requires a low voltage opamp with high current rail to rail output

  • Due to opamp/DAC offset, or resistor tolerances, output "zero" will not really be zero which causes constant DC current to be wasted in the speaker

The latter can be solved by AC coupling the speaker with a large electrolytic cap. Don't use ceramics unless you don't mind extremely high distortion.

A simple class D chip amp would use much less power. Also these usually have an enable pin, which will greatly reduce idle power usage. PAM8302 uses 1µA when it is disabled, which is nice.

As usual, the most important thing for efficiency of an amplifier is... the efficiency of the speaker (ie dB SPL at 1W/m) which can vary across many dB.

\$\endgroup\$
1
  • \$\begingroup\$ another serious problem is that a FET opamp is not going to provide enough output current anyway. You could probably get something audible with a 5532 which drives much more current, but all of these are quite unsuitable for a circuit running off 3V anyway, they typically need at least 8 or 10V of supply. They don't get anywhere near close enough to the rails. \$\endgroup\$ Commented Apr 8, 2024 at 14:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.