Post Go back to editing

ADG794 video switching

Has anyone had success with this chip?

I'm experiencing really bad crosstalk between the two inputs. input 2 audio coming through to input 1 side.

Can anyone help on this?

The video is perfect... just the audio that's causing a problem!

  • You should verify how much of the audio is coming through, the switch is specified with a 70dB of isolation, so it's possible to hear some of that but it should be at a much much lover level. Can you measure and quantify the cross talk?

    If its more than what's on the datasheet then you have a coupling issue on the board, capacitive coupling from between the traces. This can be achieved by guarding your traces with ground shielding between them (signal trace, ground trace, signal trace, ground trace pattern).

    Additionally you should consider terminating the inputs with 10kohm or 1kohm resistors to ground, that also increases isolation.

    Also consider keeping the higher power audio signals (if you have an amplifier on the board) away from the lower level audio signals.

    Ultimately make sure your audio signal in within the power supply range of the switch, for exampel if you are AC coupled make sure the switch is powered by a dual supply.

    Understanding Crosstalk

    Performance degradation in switches

    Ask the Applications Engineer

    Minimize crosstalk

  • Thanks AnalogXPRT...

    I've been scouring the net for decent crosstalk information, but till your post... nothing decent found.

    The levels could be anything as it's used for different applications. The one application was the Audio from a computer's line out. The signal was very high and therefore must be more then what the switch can handle. I had designed the board with single trace and ground traces between. The audio tracks are also nowhere near each other. So therefore I feel it's definitely inside the chip that there's crosstalk.

    By taking the inputs to ground via a resistor it drops the volume greatly, it does minimize the crosstalk but not as much as I'd like it to. I've added a relay to the design, when it switches over to other input; the relay closes and takes the line-in of other input to ground via an 8ohm resistor. I found this worked very effectively and is easy and cheap enough to implement into my design.

    Could this not be implemented into the actual design of the chip? Two extra transistors that simply clamp the unused input to something close enough to ground.

    I don’t want it to go ALL the way to ground as it might be connected to a pre-amp and might blow the pre-amp... Or so I'm led believe... Am I correct with this thought of logic?

  • Grownding or shunting the unused signal is a technique often used for audio and wireless applications.

    Some switches implement this shunt internally (ADG918 for example for wireless)

    In audio applications, as you have mentioned, its hard to implement it on chip as there are many different impedances used and therefore the on chip shunt would have to be tailored to that line impedance.

    The ADG794 is a general purpose switch so it up to you to configure the shunt in a way that suits your system. Unfortunately I can only guess on how you route the signals.

    A way to implement what you are describing is to use multiple SPDT switches and use them as a mux with the second pole connected to GND with a 50ohm, 1Kohm or 8ohm resitor whatever works for your system. You may want to consider splitting the video switching and audio switching.

    Look also at Audio switches like the ADG884,

  • I'm glad to see my thought processes are not soooo far out 

    Unfortunately the boards are already built, So i'm simply adding in a relay and resitor "hack style"! but it's a cheap and viable solution. When we design the next version of the board for greater roll out, I'll be looking into seperate audio switching.

    Thanks once again for all your help!