Soft fade out circuit for turning off speakers

Question

I have some loud speakers connected to an amplifier. The problem is when I turn the preamp off, it sends a loud POP noise through the speakers. The turn off of the amp is delayed by 1 or 2 seconds, but that's not enough to kill the amp. So here's what I'm looking to achieve:

I'd like to slowly fade the speakers out as soon as as a remote wire goes LOW (it is 12V when preamp is on, but immediately becomes 0V when preamp turns off). And by slow I mean the whole thing happening is roughly half a second.

The speakers are pretty high powered. They draw 30 - 75 watts. So this circuit should withstand that.

If that's too difficult, I can try to implement this fade out function on the preamp outputs. It would be similar to a relay, except instead of a sudden off, I'd like to bring the voltage down to 0 gradually, so it gives the impression that audio is fading out.

Is this too difficult to achieve? Would a digital potentiometer connected in series to preamp signal, controlled by a Arduino, and then fed to amplifier be enough to achieve this? What are the repercussions? Can this be nicer implemented with capacitors? Is there a good way to switch the audio with a relay or something?

Answer 1

OK something is missing here... If you turn off the pre-amp there should be nothing to fade... at the amp. Further, the pop is caused by the transient of the pre-amp shutting down, and is short, so by the time you fade out the amp it's too late.

Since they are powered separately in different boxes I think you need to make the AMP mute the output when the pre-amp shows it is off. There may be a way in the pre-amp to fade out before it sets the shutdown line though.

I'd be interested to see if that line from the pre-amp really "immediately becomes 0V when preamp turns off" or if it actually fades out with the preamp power supply. It it's the latter you could use that to turn off the input more gracefully than a relay.

Perhaps with an LM13700 Dual Operational Transconductance Amplifier wired as a unity gain voltage controlled stereo volume control between the pre-amp and the amp. RC shown here would need to be adjusted for your output signal from the pre-amp

Answer 2

Why not use a digital pot, right before the PowerAmp (so you need 2), and immediately move the "wiper" to 100% attenuation?

We MUST keep the MCU logic0 and logic1 noise [digital LOW and HIGH levels] out of the DigiPot. This noise comes from internal clocking, memory fetches/writes, IO activity, etc which cause 0.2 to 0.5 volt spikes on MCU GND and on MCU VDD.

If MCU outputs directly connect to the DigiPot, that digital noise MUST have a return path. Even if there is LogicGND in the DigiPot, digital trash will cross into the resistive-network because the analog switches need to digitally controlled.

You will end up with MCU_clock_rate buzzing and hissing in your audio, and your PowerAmplifier will attempt to amplify that trash.

How to avoid this? SLOW DOWN the digital edges, with a dedicated RC LowPass Filter, with the Capacitor tied to the MCU gnd or gnd plane. Then run a high value resistor (this is a CMOS DigiPot, right? with no steady state currents other than 1nanoAmp leakages, so even 1MegOhm would suffice except the edge into 1pF would be 1uSec) over to the DigiPot region, and include a 2nd Capacitor to absorb most of what trash comes from the first Filter; this trash may be nothing more than a noise difference between the MCU gnd and the Audio gnd, but we don't care about the source; we just don't want any trash entering the DigiPot logic pins or its VDD.

The DigiPot has its own private capacitance from Logic Inputs to its GND, because ESD structures/diodes/etc have 3pf to 10pF capacitance. Hence I use 10,000 pF in the MCU trash filters; little trash actually gets into the DigiPot.

^{simulate this circuit – Schematic created using CircuitLab}