I'm building an adapter to connect an RJ9 headset output to a 3.5mm jack output(TRRS, mic included). I have designed it so far to use DG409 (4:1) analog multiplexers to choose what lines are connected to what (for example Pin 1 on the RJ9 connector will be connected to the Sleeve of the 3.5mm jack (TRRS). This is done by connecting one of the AUX outputs to the 1 side and choosing which of the 4 outputs to output to.
NOTE: the DG409's that I'm using have two MUX's in one chip so I'm basically grounding all the inputs/outputs of one of the muxes for each chip as I only have use for one PER chip (i'm using 4 chips to route 4 signals, Left (only mono audio needed), MIC and 2 GNDS)
The 4 outputs (on each chip) have a 1MOhm pull-down resistor on each of them while the inputs have different size pull-downs (don't know why I need this to output the audio properly). The Speaker channel has a 4.7kOhm pull-down, the MIC channel has a 220Ohm pull-down and the 2 GND lines have a 100kOhm pull-down.
I am currently powering/controlling everything off of the Hummingboard (which is outputting the audio) and trying to use the dual supply function (+/- 5V supply). I am supplying negative voltage via the use of a voltage converter (LTC660) which is also getting its power also from the Hummingboard.
Currently I'm testing it by outputting audio from a Hummingboard which is travelling through the analog MUX and being outputted to a RJ9 connector which is connected to an IP phone, and I'm listening to the output speaker on the IP phone.
Issues that I am seeing:
(1) Noise on the output - there is a distinct hum that I hear on the speaker when audio is playing through. There are peaks occurring at the 50Hz harmonics.
(2) the ground connected to the voltage converter (-ve voltage provider) is 2-7mV higher than the ground on the Hummingboard. NOTE: I also have connected 5 op-amps to be used as buffers (RC4558, so 10 buffers really) to the +ve and -ve voltage supplies.
NOTE: Because everything is powered/outputted from the Hummingboard and thus everything is sharing a ground. The audio GND and circuit ground are the same.
(1) Is the voltage converter's higher ground potential the cause of my AC hum?
(2) Could the use of capacitors in series with the audio signal lines help alleviate this problem? i.e. use 10uF or 22uF caps before inputting the analog MUX's? I've been looking at some audio switching schematics and they seem to use capacitors in series for the audio signal lines sometimes
High Level Schematic
As you can see I'm trying to choose where to output the SPK and MIC lines from the AUX to the RJ9 connector. For the SPK line, mux 2,3 and 4 are not outputting signals to the lines connected to the the mux 1 output line, thus they are an open switch.
Here's an image of the frequency analysis of the noise recorded on the MIC line.
As you can see the 50Hz harmonics are still prevalent.
NOTE: I am passing the audio GND lines through the analog multiplexer as well to choose what pin it is outputted to. Hence the GND line is seeing a small resistance in series (17Ohms) due to the switch being open. Is this possibly the cause of my AC hum/ground loop, as a difference in ground line resistance will be present? Or is this resistance small enough to be ignored....
Also could it be that if my buffers are outputting voltages slightly lower than what was inputted (due to the +/- 5V dual supply not being high enough) this could cause AC Hum?