Hello,

i wonder if the RMS voltage noise of the ADCLK905 family of chips can be calculated from the specifications. More specifically: If i apply a constant input signal, what is the output voltage RMS noise in the frequency range of 0-100MHz (assuming a prefect low pass)?

I hope this can be calculated from the specification of: -161 dbc/Hz phase noise, but i am not very familiar with this quantity. Is the following calculation sensible?

-161 dbc/Hz * 100MHz = 10e-16 * 10e8 = 10e-8 in units of signal to noise power.

Voltage noise is then sqrt(10e-8) = 10e-4

The output signal voltage is 800mV, so the RMS noise 800mV * 10e-4 = 0.4 µV

0.4 µV rms noise seems very low for a chip that is optimized for fast transitions, so i assume i made an error..

I hope this made sense. Any help is greatly appreciated.

Sincerely,

Tim

I am curious about what you are trying to accomplish. When you say the reference is assumed to be constant, are you referring to long term frequency, or are you suggesting we assume a jitter free reference source? That is a theoretical exercise that you won't be able to accomplish in the real world. Are you trying to isolate the performance impact of the buffer itself so that you can estimate the total noise in your system by summing the components?