Why do my thermal spokes show as not connected even though the pour looks fine?

Oct 21, 2026 Last reply: 4 hours ago 5 Replies

First reply

amy.morrison

Nine times out of ten when I've seen spokes that look connected but fail the check, there's a second zone sitting on top of the pour in that area, and the priority numbers are fighting each other. When two fills overlap, the one with the higher priority wins, and its clearance…

Read the full reply

I'm laying out a small board with a ground pour around a bunch of pads, and the thermal spokes on some of the pads just refuse to actualy join the pour. In the 3D view and in the printout the pad looks surrounded by copper, but the connectivity checker keeps flagging those pins as unconnected. The spokes are there, I can see them drawn, they just seem to stop short of the pour copper. I've tried bumping the spoke width up and shrinking the gap around the pad, and a couple of them fixed themselves but most didn't. It's almost like the pour isn't regenerating in that spot, or the clearance between the spoke and the void is eating the whole connection. Does the pour need to be refilled after changing those settings? I refilled once and nothing changed, but maybe I did it wrong. Or is there some minimum overlap the spoke needs with the pour before it counts as connected? I'd expect a spoke that visually touches the copper to count, but apparently the software disagrees with my eyeballs. If anyone has hit this before, what ended up being the actual fix? Is it a settings thing or just a rebuild thing?


Nine times out of ten when I've seen spokes that look connected but fail the check, there's a second zone sitting on top of the pour in that area, and the priority numbers are fighting each other. When two fills overlap, the one with the higher priority wins, and its clearance around your pads cuts a hole right through the lower zone's copper. That hole lands exactly where your spokes are supposed to reach, so the spokes get clipped at the ends and you end up with a hair-thin gap that's basically invisible at normal zoom but very visible to the connectivity checker.

What worked for me was clicking on every zone in that area, including any rule areas or keepouts that came in with a footprint, and looking at the priority number in the properties. Set your main ground pour to a lower priority than the smaller zones, so the small ones override it where they overlap, and the big pour fills everywhere else. Or if the second zone is a leftover duplicate you don't actually need, just delete it. Footprints sometimes ship with a built-in keepout right on the pads, which is sneaky because it doesn't look like copper until it eats your spokes. Zoom all the way in on one of the offending spoke ends and you'll probably see that sliver of gap. Once you fix the priorities, do a full refill, not just the one zone, because overlapping zones all get recomputed together. On your numbers, one thing jumps out: your spoke width is bigger than your thermal relief gap, so each spoke only barely pokes into the pour copper. That's legal, but it means any extra clearance from a competing zone wipes out the overlap entirely. Bumping the gap down or trimming the spokes so they clearly reach past the clearance ring makes the connection a lot less fragile.

Had almost the same thing on a board last year and it wasn't zones at all. The pads in question came from a footprint that had thermals turned off per-pad, so the pour still drew a ring around them but the spokes never actually tied in. Double-click the pad itself and check its own thermal settings, some library parts override your board rules and it's really easy to miss. Also worth checking which net each zone belongs to. I once had a pour that silently dropped to no-net after an edit, it filled fine and looked perfect but the checker rightly refused to call it ground.

I remember back in the days of thick copper pours and manual etching, you'd sometimes see these connections fail even if the design looked perfect. It wasn't the software, it was how the actual copper behaved during the process. If the board stackup was unbalanced, or if you were trying to push too much current thru a tiny spoke, the copper could lift or crack under the heat stress. That hair-thin connection you're worried about? It doesn't have much tolerance for manufacturing variance. I used to run those ground planes on the bottom layer mostly, and if the etching couldn't fully run right up to the pad edge due to slight mask misalignment, the spoke just couldn't make the connection even if the CAD said it would.

Dumb question maybe, but is your copper fill on the same layer as the pads? I chased something like this for a whole evening once and it turned out I had the pour on the bottom side while the pads were only on top. In the 3D view it looked all connected from certain angles and I never thought to flip it around. When I finally spun the board over there was just empty green underneath and the little connecting bits went nowhere. Is there an easy way in the software to show only one layer at a time so you can check that?

Following this because I had something similar on my first board and never fully figured it out. My fix turned out to be the net name on the pour not matching the pads, one had a typo from when I copied the schematic over, so the checker saw touching copper but thought they were different nets and flagged everything anyway. Maybe worth opening the properties on one bad pad and the pour and just eyeballing the net field. I only found mine because I clicked around out of frustration, so no idea if that's even a common cause, but it took me two evenings to spot it.

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required