AnsweredAssumed Answered

charge injection and clock feedthrough in analog switches

Question asked by Analog_sup on Jul 29, 2015
Latest reply on Jul 31, 2015 by mayp


I have few questions regarding charge injection and crosstalk (clock feedthrough) in analog switches.

1) I was looking for analog switches which can reliably transfer a minimum signal level of 50uV. This would correspond to better than 0.05pc of charge injection with an output capacitor of 1nF. Since charge injection contribution mostly comes from the NMOS and PMOS mismatch, I was wondering if by increasing the output capacitor 10 times, I can improve on the change in the output voltage 10 times. Most of the spec sheets report use of 1nF of capacitor for the testing and my doubt comes from the fact that if it was straightforward, why a better charge injection performance is not reported for a slightly bigger capacitor.

2) If my understanding is correct, clock-feedthrough (which is one of the crosstalk component in spec sheet) would cause similar error in output voltage. I found that most of the data sheets report this parameter in terms of attenuation factor (dB). If I have to translate this factor into change in output level, what value of capacitor (Cg) should I assume between gate and signal channel for hand calculation. I believe that for certain output capacitor (Cl), there will be a change of (Cg/(Cg+Cl)) * Vclock_high, at the output of switch (across Cl). With the attenuation factor of say 100dB, this should be lowered by 10^5 times. Please correct if I am wrong.

Would be glad if someone can clear my doubts.