EEVblog Electronics Community Forum

Electronics => Projects, Designs, and Technical Stuff => Topic started by: marcan on June 17, 2010, 03:58:37 am

Title: Audio AC coupling and DC bias
Post by: marcan on June 17, 2010, 03:58:37 am
I'm working on a project that includes a passive line-level audio mux (no buffering) that selects between passing through an audio input to an audio output, or instead selects an internal audio source to pass to the output. I've chosen the FSA2257 (http://www.fairchildsemi.com/ds/FS%2FFSA2257.pdf) mux for the job.

I understand that I need AC coupling for the external input and output, as the mux will only pass signals within its supply range (0-5V) and external signals are probably centered around 0V. I'll use 1µF caps on both the inputs and outputs (I believe they should provide adequate frequency response with line-level loads; this isn't meant for headphones/speakers). However, I haven't been able to figure out whether I also need explicit DC biasing on the input to set the DC offset to somewhere around Vcc/2 or not. Is letting the signal float after the input cap OK, or should I explicitly bias it? If so, how should I do it? A resistor to Vcc and another one to Gnd? Are there specific criteria for the value or would, say, 100K each to Vcc and Gnd be adequate? If I had to guess, it's a tradeoff between extra load impedance (obviously 1K resistors wouldn't cut it, as they'd drastically reduce low-frequency response vs. a 10K or higher line level load) and adequate biasing, but I'm not sure what that would mean in practice (would 1Meg resistors do the job?). Or maybe I just don't need the darned things at all.

You can probably tell analog is not my forte :)
Title: Re: Audio AC coupling and DC bias
Post by: jahonen on June 17, 2010, 06:50:14 am
If power supply pins of the MUX are connected to unipolar supply (+5V and 0 V for instance), then you definitely need the VCC/2 biasing on the signal. As a general rule, an input signal to the any IC should not exceed supply voltage (of course there are some exceptions).

Another way would be to choose such a MUX which can operate with split supplies and still accept logic level control signals, but that is probably not an option since you have already chosen the MUX chip. I personally prefer not to AC-couple whenever possible to reduce unnecessary roll-off and phase distortion of the low end of the signal. Of course that requires careful evaluation of DC offsets in each stage.

Regards,
Janne
Title: Re: Audio AC coupling and DC bias
Post by: marcan on June 17, 2010, 02:42:35 pm
This is going to be battery powered, so a bipolar supply would be too much of a pain. Audio quality isn't critical anyway (this is no hi-fi apparatus), so I'm not too concerned about having to AC couple.

So, given that I need Vcc/2 biasing, what would be the recommended way to do it?
Title: Re: Audio AC coupling and DC bias
Post by: jahonen on June 17, 2010, 03:03:11 pm
Simple resistor divider, like 100k from VCC to signal and 100k from signal to GND should work. This results effectively 50k input impedance.

Regards,
Janne
Title: Re: Audio AC coupling and DC bias
Post by: marcan on June 17, 2010, 03:42:27 pm
Sounds great, thank you for your help :)