Charge pump and clock Feedthrough problem

[AdvaRes]

Hi members,

In my PLL's Charge pump (see fig) the UP signal of the charge pump causes a feedthrough. The current flows in one direction (wrong direction) for a while before changing it direction to charge the Filter. This result in a spike in the current signal and a ripple in the VCO Vtune signal.
Could someone explain what causes that and how efficently avoid that ?

 

[erikl]

In my PLL's Charge pump (see fig) the UP signal of the charge pump causes a feedthrough. The current flows in one direction (wrong direction) for a while before changing it direction to charge the Filter. This result in a spike in the current signal and a ripple in the VCO Vtune signal.
Could someone explain what causes that and how efficently avoid that ?

Hi AdvaRes,
it looks like injection feedthrough. If this is true, you could use the standard cancellation method: at the drain of the MOSFET connect another MOSFET with ½ of the W/L ratio (both source and drain short-cut and connected to the drain of the original MOSFET, using common bulk), its gate controlled by the inverse signal of the original switch. Thus you inject the same, but inverse amount of charge at the (nearly) same time, which cancels quite well the charge injection. (½*W/L, because drain and source together then deliver about the same amount of charge as the drain of the original switch).

Just give it a try!
Cheers, erikl

 

[AdvaRes]

Hi Erik !

I'm happy that there is a solution for such a phenomenon.

Thank you so much for the technique.


Cheers,
Advares.