Number of spokes in thermal relief

Hand soldering is fussy about hole diameter and I have used exotic pad shapes on the solder layer to allow the iron better pad contact. Too often you end up heating the pin, not the pad

Hand soldering these connectors into these boards is going to be difficult enough with pads on 14 layers, at least 6 of which connect to planes.

These days component leads are tin plated and are intended to be soldered with Pb free solders. This is usually not an issue as soldering with Pb free solders reaches temperatures that melts the tin plating causing it to mix with the solder giving good coverage of the component lead with solder. Hand soldering even with Pb free solders does not always reach these temperatures at all never mind long enough but they certainly don’t when SnPb solder is used. Exposed tin is one source of whiskering although these whiskers tend not to grow that long (<1mm).

Contamination before and during the soldering process is the biggest cause of whiskering with these whiskers often reaching 10mm or more in length. The contamination causes stresses (compression stresses) within the joint and in order to relieve this stress a whisker is extruded. Component leads need to be clean before soldering and quality solders and fluxes (halogen free fluxes) used during soldering. It is also preferred to solder in a nitrogen environment rather than air. Boards assembled this way have shown no signs of whisker growth longer than a few microns.

You are not going to accomplish this with hand soldering!

Whiskers are even more of a problem in a vacuum since when a short occurs, and enough current is available, the vapourised tin forms a plasma which continues to conduct until either the current is interrupted or all of the available tin has been consumed. It apparently takes as little as 12V and around 2A to create this plasma but once the plasma is conducting it can conduct in excess of 200A.

Conformal coatings can help with whiskering not because they stop the whiskers from growing but because everything is coated with an insulating layer. It takes two whiskers to cause a short.


Man, you really know your way around town. Thanks for the info and sharing it.

PS: I’m amazed that KiCAD is being used for stuff like this.

Hopefully there is nothing in that post that the OP doesn’t already know, but the fact that they are hand soldering these boards makes me wonder. Imagine the stresses that build up due to the uneven heating and cooling cycles as you slowly work your way across the board soldering each pin. I’ve seen whiskers on motherboards of S100 card cages where the 100 pin sockets had been hand soldered.

Whiskering is a well known phenomenon, although not well understood there does seem to be some agreement that stresses are a primary cause. Stresses due to contamination in the joints and even mechanical stresses from mounting the PCB.

Several satellites have been lost after both the primary and secondary control computers have failed. Several satellites that are still in service are running on the secondary computer because the primary failed. There’s the famous communication outage across the USA back in 1998 when the Galaxy 4 satellite suddenly spun out of control and was lost. In all cases the suspected cause was whiskering. Even the famous Patriot missiles used in the Gulf war to shoot down Scuds suffered numerous misfires, again due to whiskering. Even pacemakers have had problems with whiskering.

Tin is not the only metal that can grow whiskers, zinc, cadmium and silver can too. The lead added to SnPb solder helps to mitigate whiskering which is why companies like IBM still use SnPb solder with a minimum of 3% Pb. Bismuth can also be used instead of lead.

Yes, I too am surprised that they are using KiCAD for this type work. Not because I have anything against KiCAD but aerospace companies are usually quite stringent about what software and associated libraries etc. can be used within the company. With software like KiCAD there is no one to hold accountable for problems or any guarantee you can get them fixed.

1 Like

not really. The current PCB is made with Mentor. I’m evaluating KiCAD to manufacture a demo PCB to solve only this wetting trouble. Mentor has the two spokes settings for thermal relief

Interesting… then - IMHO - get in contact with the Developers/CERN and sponsor the functionality you desire, if that’s the only point holding you back using it :wink:

Is there a reason for such a low temp? Customer requirement? Is the board temp measured near the connector pins?

Searching reveals pre-heat temps as high as 170 C, but you could start lower and work up.

Are you using a “made for the purpose” board pre-heater? You mention soldering over an oven.

early in this thread I asked how to do that, noone answered

this is an ESA requirement (written in an ECSS) we had to respect. This does not mean that limit temperature to 90° is always a good thing or does not cause troubles

How to contact kicad developers:
First option via the bugtracker.
(Mark your bug as wishlist.)
The highest chance of getting attention is by supplying a patch.

Or maybe contact the developers via mail:

Donating money:
(I don’t now if there is a way to tell them that this money should be used for implementing some feature you want.)

1 Like

We were probably in problem solving mode instead of sending you into the woods :wink:

As @Rene_Poschl already posted, those are the options to get to them and best chance would be to provide a patch.
Just keep in mind it’s open source and people are working on this mostly in their free time and because they want to do it (esp for themselves).
But maybe you’re lucky, who knows.

Can KiCad do hatched planes?

Set the plane connection to none in the pad properties, and connect it manually. That way you can chose number of spokes, track width, and so on.

OK, this can be feasible for few connections, but in our board there are more than 3600 thermal relief

I tried rotating pads by 45°, the thermal relief change to cross like.

I can’t understand why I should get two spokes increasing the clearance?
With small value I got 4 spokes connected to the plane, increasing the clearance over 0.4 mm I got plane disappear and remain only the 4 spokes

And how many other connections are there on the board routed by other means? How much additional work would it be to route the plane nets?

may be for your connector a solution can be a footprint with pre-assigned pad for thermal relief, and a keep-out zone with no-copper pour

thermal_relief.kicad_mod (2.7 KB)
or just a keep-out zone with no-copper pour for the connector and route manually as bpiphany suggested

1 Like

When I suggested you would be left with two spokes that was based on the assumption that not every pin in the area was part of the same net.

Is this really what your board looks like? On all six layers? No wonder you can’t solder it.

1 Like