wire Inductance vs capacitor

Question

Is it a way to find out what is the "admissible" length of a wire connection without "disturbing" a specific capacitor?

Context: I have several capacitor serial killer devices. I have to replace some specific capacitors inside of those devices in average 2 or 3 years after a continue and intensive use. Now I start to move some of those capacitors outside of the housing for 3 main reasons. Firstly, it's far easier to replace it. Secondly, the capacitors have cooler and well ventilated environment.Third, I have more space to install a bigger capacitor. For most of them, I have no problem to relocate them near their initial location. But for some, the "suitable" place is little bit far away.

So my question is: If I rewire the capacitor connections, what is the permissible wire length for a specific capacitor without being disturbed by the wire inductance. I know this depends on many factors. But is there a "rule of thumb" which I can count on to find out if I'm within the "safe zone"?

Edit with additional information: The capacitors are all electrolytic capacitors and almost all of them are related to VRM.

Thanks

Answer 1

A good rule of thumb is 1 nH per mm of wire. This is a surprisingly good ballpark number whether you are talking about a component wire, or a single wire in a reasonable-sized loop, for reasonable sized wires.

Note that a capacitor intended for low frequencies (electrolytics) may have a significant internal inductance, which renders any reasonable external inductance moot.

Answer 2

"Is there a way to find out what the "admissible" length of a wire connection is without "disturbing" a specific capacitor?"

No. You must know the design sensitivity to LdI/dt, resonance, crosstalk from ESR and ESL and how low ESR affects surge current on other parts.

Answer 3

The capacitors are all electrolytic capacitors and almost all of them are related to VRM.

With the extra wire inductance and resistance, it's not going to work. VRM operates at quite high frequency. So if you put long wires, adding a few hundreds of nH and tens of milliohms of inductance, what's going to happen is the high frequency ripple current will go through the low impedance path which is the caps that remain on the board. So the offboard caps will last longer, but the remaining caps on the board will die quicker.

If the caps die there can be several causes:

• Low quality caps: easily fixed by replacing with good quality caps.

• Wrong temperature caps: 85°C cap used in a hot zone, it will die. Again easily fixed by using 105°C cap instead.

• Wrong type of caps: high-ESR low ripple current "general purpose" cap instead of low-ESR high-ripple current cap: self heating is due to ripple current and internal resistance (ESR) so the more ESR the more self-heatiing. High-ESR caps in this position will overheat and die. Again easily fixed by using low-ESR or polymer cap.

• Bad thermal design where the device just gets way too hot: other stuff will die besides the cap, so you have to fix the airflow.

I'd simply remplace the dying caps with upgraded models like polymer or ultra low ESR caps and check the airflow.