Detecting DC offset (as in failure) on amplifier output

Sooo..... not sure if this is the right forum for this question, but here we go.

I looked around and could not find an answer specific to my question.

I would like to build a circuit that monitors the output of an amplifier (speaker output). When a positive DC offset is found (say.. more than .5 seconds), the arduino (or perhaps this is overkill?) would kill the relay.

This amplifier is an old Harman Kardon quad 75+. So.. the amplifiers (4 internally) can be operated in either 4 channel mode (each amp is a simple push/pull) or stereo mode (two amps are bridged for each channel via a switch on the rear).

My theory is as follows:

  1. Take the output of each individual amp and run it through 2 diodes to break out a + and - voltage (relative to ground).

  2. Pass each of the signals through a... 1K (?) resistor and then feed the signal to a 3-5 volt zener (to limit the voltage to arduino-safe levels).

  3. Monitor each of the eight signals with an arduino. I would not care about the level (so ADC is not needed).. only if the signal is persistent more than half a second or so.

I already have whipped up a simple power-on delay circuit that uses a transistor, zener and rc network to provide soft-start for the amplifier. I would prefer to move that to also be controlled by the arduino. Much easier this way - and centralized. I could even provide diagnostic LED's via the PCB this way in case anything went south.

Would the above actually work? Is my reasoning sound? Perhaps this is a much simpler way to do this that I am not thinking about?

I made something like that once for an amplifier I built, but I don't remember exactly how it worked... I didn't use a microcontroller. There might have been an NE555 timer involved, and a 555 can drive a relay, but I just don't remember.

  1. Take the output of each individual amp and run it through 2 diodes to break out a + and - voltage (relative to ground).

A full-wave bridge rectifier (4 diodes) will work if there is no common ground between your circuit and the amplifier. If there is a common ground you'll need a way to read/monitor the negative voltage.

  1. Pass each of the signals through a... 1K (?) resistor and then feed the signal to a 3-5 volt zener (to limit the voltage to arduino-safe levels).

You'll need two Zeners unless it's full-wave rectified. Or, it can be done with a pair of regular diodes connected "backwards" to the low-voltage power supply. [u]Over-voltage protection circuits[/u]

  1. Monitor each of the eight signals with an arduino. I would not care about the level (so ADC is not needed).. only if the signal is persistent more than half a second or so.

Half of those connections should be ground. :wink: (In stereo bridge-mode you're only using the 4 "hot" connections.)

...I'm pretty sure I built my amplifier protection circuit around [u]analog comparators[/u] running from a dual power supply so I could monitor the positive & negative halves of the signal. With an open-collector comparator you can "or" the outputs together which makes the logic easier. I don't remember how I handled the timing, but that's super easy if you're using a microcontroller.

What you have to do is RC filter the speaker output signal. Use a time constant of a second or two.
(gross faults will act in a tiny fraction of the time constant, note, but you don't want false triggering
on low bass notes).

The harder part is using the output voltage across the capacitor to control a relay. It can be either
polarity in a fault condition.

Passing it through a bridge-rectifier will give a signal of a single polarity, but floating around ground
(or one of the speaker terminals in the bridged case). Typically there won't be much current
or an opto isolator would be easy to use.

Standard circuits use the full +/- voltage rails of the amplifier and a differential sensing circuit
using discrete transistors. This makes the actual voltages irrelevant, the differential circuit can
track across the whole voltage range at audio frequencies if needed.

To avoid having to use such voltages I think you can use an opto coupler - use a high gain one, and
a sensitive receive side circuit. By using an AC opto coupler the bridge rectifier can be avoided.

For instance with an amp with +/- 30V rails, and a 10k / 100uF RC filter (use non-polarized cap), into
an ac-input optocoupler. This will have a threshold around 1.4V, but only a small current can
flow for a low voltage offset - a hard fault will put 3mA into the opto coupler, a high gain one might
give 5 to 10mA out for that input (but only 100uA or less for softer faults).

Hence a high sensitivity circuit on the output of the opto coupler...