The size and shape of capacitors DOES make a difference. All microfarads are not "the same."
Example: GE Feedwater Control system DC power supply. Maximum ripple allowance: 50 mv.
We obtained new capacitors that were the same voltage and capacitance rating, brand new manufacture. The caps were installed, and the ripple was out of tolerance. The shop got out another brand new set, installed them, and had the same results. They even pulled out the capacitance tester, and verified the capacitors had the right rating, and were not shorted.
Engineering was called in to support the maintenance issue.
Going to first principles, we started by comparing the new capacitors to the old capacitors, and found the old ones were taller and wider. The new ones were shorter and skinnier.
We did a back of the envelope calculation of the capacitors (nobody wanted us to take apart the capacitors and how would we be able to tell what the electrolyte constant was anyway), but we showed that the smaller, skinnier caps would have to have a different internal structure. We postulated that the difference was causing higher ripple (AC leakage).
We went to Mainline Electronics (now out of business, formerly located in Plainfield, IL) and asked them if they stocked NOS capacitors that matched the voltage, capacitance, and approximately the size.
They did (they even had the original manufacturer!) We installed the NOS capacitors, and the ripple dropped to well within tolerance.
So, lets look at the "why" of this.
Again, capacitance is derived from the formula 0.225 x K x A / S, where:
0.225 is an eigenvalue
K is the Dielectric Constant
A is the Area of the Plates
S is the Space between the plates.
By using a different dielectric, I can reduce the area of the plates and the space between the plates.
Example:
K = 4
A = 11 sq inches
S = 0.1 inch
= 99 mF
Now tweak the dielectric constant:
K = 7
A = 6.3 (reduced area, less paper, cheaper to produce)
S = 0.1 inch
= 99.25 mF
So for filter caps, Size Does Matter. For your run of the mill caps, it may not. It's also why when you get a filter cap from The Real Bob Roberts, it looks like the old cap you removed, not a new svelte one.
By the way, the original New Capacitors were provided by GE Nuclear, which has a Nuclear Quality Assurance program governed by 10CFR50 Appendix B. The capacitors provided by GE had the right capacitance - it just didn't work in the filter circuit.
Believe it or not. Try it yourself on a power supply. Not all capacitors are equal, even if they have the same rating.