Dendritic growth on PCB

[helius]

The "dendrites" you are worried about are technically called CAF, Conductive Anodic Filaments. You can search for technical reports on the subject by using that term.

It has nothing to do with flux or tin, but is a metal migration process that happens inside the PCB laminate as copper ions are dissolved and redeposited in the electric field. To prevent it, you must visualize the electric field gradient and keep vias and tracks well away from it.

By making the spark gap sharply pointed, you can minimize the area of copper exposed to the field, which will restrict the amount of ions that can migrate. Thin copper is less subject to CAF than thick copper.

[tooki]

Dendrites only grow in liquids or in the presence of substantial moisture. This is something you can control. So in electronics, tin whiskers are the issue that's much more challenging (and still hasn't proven to be a huuuuge issue).

But they'll never, ever handle "tens of amperes" of current. We're talking about crystals of tin much thinner than hairs. The current to blow one open is, from what I've read, usually under 50mA.

This is why I asked above whether they're really an issue in a spark gap, where current would only be expected to flow when there's substantial voltage and current anyway.

Okay, I got what you mean but, if there is a tin whisker connecting both electrodes, will it do that the arc doesn't jump? Because in case that the arc will occur anyway, I think that the whisker will be melt, but then, the problem is that the tin will go onto the pcb surface and maybe can short something. Or atleast that's what I think

Just how much tin do you think it is? Remember, we're talking about crystals that are so thin they're practically invisible to the naked eye.

Or do as others have said and don't tin it at all, leave the spark gap as bare copper.

[marc1996]

The "dendrites" you are worried about are technically called CAF, Conductive Anodic Filaments. You can search for technical reports on the subject by using that term.

It has nothing to do with flux or tin, but is a metal migration process that happens inside the PCB laminate as copper ions are dissolved and redeposited in the electric field. To prevent it, you must visualize the electric field gradient and keep vias and tracks well away from it.

By making the spark gap sharply pointed, you can minimize the area of copper exposed to the field, which will restrict the amount of ions that can migrate. Thin copper is less subject to CAF than thick copper.

I will search info of CAF. Thank you for your info.
About the sharp points, thats true, the copper area that will be affected by the field will be less but then, the electrode will be damaged faster and the vertex will be destroyed faster. It's like a balance.
The info that I have recollected before do not recommend at all to do the shape very sharply because of that. So maybe I should search an equilibrium between them...

Dendrites only grow in liquids or in the presence of substantial moisture. This is something you can control. So in electronics, tin whiskers are the issue that's much more challenging (and still hasn't proven to be a huuuuge issue).

But they'll never, ever handle "tens of amperes" of current. We're talking about crystals of tin much thinner than hairs. The current to blow one open is, from what I've read, usually under 50mA.

This is why I asked above whether they're really an issue in a spark gap, where current would only be expected to flow when there's substantial voltage and current anyway.

Okay, I got what you mean but, if there is a tin whisker connecting both electrodes, will it do that the arc doesn't jump? Because in case that the arc will occur anyway, I think that the whisker will be melt, but then, the problem is that the tin will go onto the pcb surface and maybe can short something. Or atleast that's what I think

Just how much tin do you think it is? Remember, we're talking about crystals that are so thin they're practically invisible to the naked eye.

Or do as others have said and don't tin it at all, leave the spark gap as bare copper.

I thought that before about leaving the spark gap as copper but I do not know at all how this works and furthermore, the 80% of Printed spark gaps that I saw in the net are made of tin, that's why I thought that maybe it could be better..

[eb4fbz]

I would not tin the spark gap pads. Thickness will not be tightly controlled, not even using HASL process. There will be some risk of bridging, etc.

Flux residue will also affect breakdown voltage. Some residues are hygroscopic and would absorb moisture with time, with even worse consecuences.

Better use ENIG over them.

[marc1996]

I would not tin the spark gap pads. Thickness will not be tightly controlled, not even using HASL process. There will be some risk of bridging, etc.

Flux residue will also affect breakdown voltage. Some residues are hygroscopic and would absorb moisture with time, with even worse consecuences.

Better use ENIG over them.

.

[eb4fbz]

If ENIG is not available, check if any other chemical finishing is. Chemical finishing processes are much more controlled than mechanical ones.

It depends on the enviroment it will be used at, but i would not leave bare copper exposed.

[Wolfgang]

Just to play the devils advocate: Can PCB spark gaps be really worthwile investing such effort ? I mean, their primary deficiencies remain even after careful manufacturing:

- dependency on PCB material, moisture, atmopsheric pressure, dust, aging, ...
- dimensions must be very carefully controlled

If you need something reliable, who not resort to dedicated flashover components ?

[Kjelt]

Just to play the devils advocate: Can PCB spark gaps be really worthwile investing such effort ? I mean, their primary deficiencies remain even after careful manufacturing:
- dependency on PCB material, moisture, atmopsheric pressure, dust, aging, ...
- dimensions must be very carefully controlled
If you need something reliable, who not resort to dedicated flashover components ?

I am still not sure what the application is, if it is lighting protection than I would definitely recommend a dedicated sparkgap component. If like the first post states to prevent possible ESD problems, than it might be used on all I/O pins and than components would be quite a pricy solution. But I thought they normally used TVS diodes for that not sparkgaps ?

[marc1996]

.