charger injection effect

[shanmei]

I read the book from Maloberti, which says that:

"A charge injection into a low-impedance node will only cause a glitch whose duration depends on the value of the node impedance. However, a charge injection into a high impedance or capacitive node causes an offset that can be problematic, especially when the injected charge is signal dependent in a non-linear manner."


I don't know why the low impedance node will have a glitch due to the charge injection, and the high impedance will have offset error? Anyone can help me? Thanks.

 

[erikl]

Think of the time constant of such a charge injection: Injected charge δq = C*δV , the form of the voltage change δV (the glitch) depends 1. on the generation impedance of the charge injector, and 2. on the load impedance, which is the capacitance C in parallel to any load resistance.
So both the rise time (which usually is small) as well as the fall time of the glitch depend on the impedances of the generator (the injector) and of the load.

If the load is a low impedance node, the fall time of the glitch will also be short, so the voltage change δV due to the glitch will be (nearly) decayed until the next measurement.

However, if the load has a high impedance, the fall time of the glitch will be rather long, hence there may be still a considerable "offset" when the next measurement starts. This offset will be added to the proper measurement value, which is transferred from the sample & hold capacitor.

 

[D.A.(Tony)Stewart]

Another way to say same thing...


by Conservation of charge Q=C1V1=C2V2 thus offset V2/V1 (%) = C1/C2 (%)

Charge injector into high impedance will offset according to capacitance ratio of initial charge voltage before (1) and after(2).


LOw impedance will use same C ratio offset but decay quickly so appears as spike.