AD9649: Switching clocks and input

We plan to use the AD9649 in combination with an (existing!) system that switches between about 20 inputs using an analog mux of sorts. Each input provides an analog signal and a clock. The sample clocks run at about 8MHz. The clocks are not synchronized.

We've evaluated the AD9649 in a setup with only a single device and it performs well. We did some experiments with switching inputs, basically by just manually removing connectors, and that appeared to work as well, but I couldn't find information on if we're doing this properly.

When switching inputs, the clock input will "glitch". Experiments show that the AD9649 apparently isn't picky about it, but I'm enquiring here to make sure that this will continue to work in the real world.

For example, I could put the AD9649 in "shutdown" or "standby" mode while switching inputs. Would that make any difference?

Currently I just use "standby" so the clock glitches won't propagate to the FPGA.

  • 0
    •  Analog Employees 
    on Dec 12, 2015 10:18 AM

    Hi Milosoftware,

    In switching between different signal sources for both the analog input and clock, the AD9649 is not picky or sensitive, but it might need some time to "come back" if the signals are idle while switching between sources.


    The input common mode voltage for the analog inputs is applied externally. If when you are switching between signal sources the input common mode voltage goes out-of-range (e.g. goes to 0V), then some time would be needed for this voltage to come back within range. The converter would still be working properly, but it would not be "seeing" a valid input signal until the input common mode voltage comes back within the proper range. If the input common mode voltage is maintained when you switch input signal sources, there is no problem.


    The CLOCK pins supply their own input common mode voltage. Even so, if the clock is held idle, there could still be a small wake-up time after the clock becomes active again.

    Can your system tolerate a small wakeup time after switching clocks and inputs?

    Thank you.


  • The analog input is actually a single-ended video signal. We connected VREF to Vin- to get a 0..2V range on the Vin+ input. I'd expect the common mode to not be a problem here (the large c.m. swing this may cause is probably bad for overall analogue performance, but the S/N of the source signal isn't that great to begin with). While switching, the input may drop to zero, but that wouldn't matter much. It won't "float" to a voltage outside its normal 0..1v5 range.

    We'll need to wait for a Vsync pulse after switching, and we'll probably ignore the first frame anyway, so we won't really start processing until after over a 100k samples have passed. I think that should give the ADC enough time to settle if I read your answer correctly.

  • 0
    •  Analog Employees 
    on Dec 15, 2015 12:51 AM


    I agree with your assessments. Based on this information, I believe the AD9649 will work fine.

    Thank you for using the AD9649. I wish the best for you and your project.