| Electronics > Beginners |
| Why aren’t bypass capacitors built into IC’s? |
| << < (2/2) |
| T3sl4co1l:
FYI, the crossover point is around 100MHz and a few ohms. If the IC needs a supply impedance lower than that, or at a higher frequency than that, it has to be provided on package, or on die. If lower, then it can be provided on the PCB, to varying degrees of difficulty (worst case, say: multilayer board with stacks of alternating planes and tons of vias under a BGA?). Tim |
| amyk:
Look at a relatively recent CPU from Intel or AMD, the bypass caps are very obvious. |
| ejeffrey:
One reason is simply processing. You can't make a very large capacitor on a silicon chip compared to a multi-layer ceramic cap. You can add ceramic capacitors by bump bonding them to your chip or package, and that sort of thing is done when necessary but it adds extra fabrication steps. You can get more capacitance with higher ESL than on-chip capacitors. On the other hand, you typically get less capacitance with lower ESL than with capacitor mounted on the PCB. So on-package capacitors don't usually eliminate the need for PCB mounted capacitors, they typically just bridge the gap between what you can do on chip and on the PCB for applications where that is required. |
| helius:
Many of the classic ceramic PGA packages have gold pads on top for bypass capacitors. They were mainly for insurance for extra noise margin etc. since in most cases they weren't used. As packages changed to epoxy and metal IHS, etc, the caps got relocated and can be found on the interposer, under the spreader. I see a comment above showing an xray with capacitors embedded in the package, which would be possible if they were soldered to the lead frame before epoxy injection. Placing a capacitor in the IC itself would require a dielectric compatible with the wafer fab process which I would think is much more daunting. You would also have to calculate leakage from the plates to other structures in the solid state die which could be a problem. example: visible under the spreader: |
| danielbriggs:
Thanks everyone, Your replies are really interesting - thanks for taking the time to explain. All the best, Dan |
| Navigation |
| Message Index |
| Previous page |