Very low-quiescent (<2mA), 5V, simple "speaker" driver

Question

How can I drive a 30ohm speaker from a 5V microprocessor in a simple, very low quiescent-current (<2mA) manner?

The "speaker" is actually a coil, not a speaker - and I'm trying to induce a magnetic field, like in an audio hearing loop.
It's driven from a PIC microprocessor, primarily with brief, digital (rail-to-rail) pulses, but occasional DAC/analogue signals.
It also needs to have separate gain control (that I can control from the PIC - there are plenty of spare legs) and (ideally) disable the whole thing entirely.
I have a 5V, 200mA single rail supply. The coil is 30ohms. Ideally I'd like to be able drop up to 150mA through the coil in short bursts.

Most important thing: low quiescent current, even when it's enabled. Preferably <2mA.
Important things: simplicity, compactness, and low-cost
Unimportant things: fidelity/distortion, temperature stability

I've looked at audio amps (which seem over-complicated, and Quiescent too high), standard op-amps (but they seem to have low current source/supply) and various BJT options (some kind of emitter follower?) but this analogue stuff is a little beyond me.

Answer 1

If you want to get 150mA with a 30 ohm coil (the resistance will increase with temperature) you'll need no more than a couple ohms switch resistance.

One way to get that would be to use two complimentary pairs of MOSFETs in an H-bridge configuration. For example, the DMHC3025LSD contains all four transistors and is less than 0.2 ohm for the total high+low switch resistance. To get a fairly clean 1kHz you'd need something like 25kHz-50kHz minimum PWM frequency, so you'd probably need a gate driver for the MOSFETs rather than being able to drive them directly from a microcontroller PWM.

If you drive them both high or both low, the current consumption will be minimal (microamperes). For an analog type output you would drive them with complimentary PWM waveforms to get bipolar output. It might be difficult to get much fidelity unless you're using close to full amplitude. If the inductance of the coil isn't enough to smooth the waveform at the PWM frequency you could add an LC filter, but that will affect the 'digital' waveforms too.

Answer 2

What frequency, turn on/off time? If it's slow (~1ms) you could use a SSR (solid state relay) For higher switching speeds you could look into analog switches. Here's an early hit on a digikey search. http://www.ti.com/lit/ds/symlink/ts5a4624.pdf
But there are thousands to choose from.

Gain control could be PWM, or if not take the dive into analog electronics, it's fun.

Answer 3

I would think that any of the smaller H-bridge drivers for small motors would be pretty much ideal for the application as you describe it. You can drive the coil with pulses of either polarity, and you can use "class-D" or delta-sigma techniques to create analog waveforms — the inductance of the coil itself will provide much of the required filtering.

Answer 4

After some further experimentation, I've decided to use a single BJT emitter follower, biased into forward-active to allow analogue input. However, I've added an extra line to a standard IO on the PIC. When set as an output, this will serve to drive digital signals, and when set as an input, it will revert to analogue. Gain will be controlled by a digital pot in place of (Re)

As far as I can tell this serves all my requirements - output fast sharp digital as well as clean analogue, low quiescent, and very simple/low-cost.

^{simulate this circuit – Schematic created using CircuitLab}