Hi,
I just got back my first 4-layer board from BatchPCB, and there seems to be a problem with my GND plane.
Pads that connect to the GND plane (one of the inner layers) by a via work fine - they’re all interconnected. But holes that connect to the GND plane with a thermal don’t seem to be connected.
Have a look at the attached image. It just shows the top and GND plane layers.
As I understand it, EaglePCB shows power planes in reverse. If that’s correct, then I think it ought to be working, but it’s not. Can anybody tell me what I did wrong?? I’m stumped.
–Dave
nerdfever:
Hi,
I just got back my first 4-layer board from BatchPCB, and there seems to be a problem with my GND plane.
Pads that connect to the GND plane (one of the inner layers) by a via work fine - they’re all interconnected. But holes that connect to the GND plane with a thermal don’t seem to be connected.
Have a look at the attached image. It just shows the top and GND plane layers.
As I understand it, EaglePCB shows power planes in reverse. If that’s correct, then I think it ought to be working, but it’s not. Can anybody tell me what I did wrong?? I’m stumped.
–Dave
I suggest you load your gerber files and drill file into gerbv and explore the layers until somebody who will recognize the cause of the problems comes by. I have not worked on 4-layer boards myself.
Things like the potentiometer will need a small trace in the schematic labeled “GND” (or whatever your ground plane is labeled…) from the pin to be connected.
Manually placed via’s also need to be named “GND”…
Perhaps sticking a screenie of the schematic may help identify the problem.
FartingMonkey92:
Perhaps sticking a screenie of the schematic may help identify the problem.
Thanks.
OK, here’s the relevant portion of the schematic. I circled the same pins and holes as on the layout I posted before.
According to Eagle all my GND lines are indeed connected together. And a previous version of this board (2-layer) worked fine.
I don’t mind posting the Gerbers if somebody wants; this is a hobby project so there’s nothing proprietary.
–Dave
Hmmm, if you move the “GND” symbols of the ones which don’t connect in the board editor, do the connections in the schematic editor follow them arround?
Also, use the “Info” tool and click on one of the failing green connections in the schematic editor, are they called “GND”?
Here are the Gerbers (attached), as I sent them to BatchPCB.
I’m still stumped - they looked OK to me.
Just had a look at those gerbers, i don’t see a ground plane…
Make sure you have the “Unrouted” layer turned on, you should see a few yellow air wires connecting the pins that don’t connect.
FartingMonkey92:
Hmmm, if you move the “GND” symbols of the ones which don’t connect in the board editor, do the connections in the schematic editor follow them arround?
Also, use the “Info” tool and click on one of the failing green connections in the schematic editor, are they called “GND”?
Yes, and Yes. I know what you’re getting at (you’re thinking that those pins & pads aren’t really connected). But they definitely are, in Eagle.
Re the Gerbers, the file “Rev4B.G2” is supposed to be the GND plane layer. Attached is the ViewPlot output from looking at that file.
To me it looks inverted - where there is white there ought to be copper, and no copper where it’s black.
But if that is the problem, then why/how are the 3 pads on the top layer interconnected? They are.
I’m still stumped.
FartingMonkey92:
Make sure you have the “Unrouted” layer turned on, you should see a few yellow air wires connecting the pins that don’t connect.
There are definitely no unrouted airwires. And it passed both the DRC and ERC fine. (and the DRCbot)
Yep, getting odd-er…
Have you remade the gerbers and had a look how they have come out?
I looked at the layers with Gerbv. It shows Rev4B.2G (corrected with edit) with color spots on the problem places, and it shows solid black on the places without the problem. I infer the black is referring to the places with copper, and the color is the places with copper removed. It looks like the negative of the image of the G2 layer on this thread.
I suspect the three pads at the upper left are getting connected via the GTL layer for the top 2 and the GBL layer for the lower of the three. Edit: The three pads I am referring to are just below the big top left mounting hole.
It does seem like an inverted gerber. EAGLE makes planes as traces, going back and forth to make a fill. so unless the CAM job was told to make an inverted output, it should be just like the traces on the rest of ya board.
Are the planes on completely seperate layers? Did you make your own CAM file?
What do the planes look like in the board editor?
Hmmm, why would it invert…
analogon:
I looked at the layers with Gerbv. It shows Rev4B.GTL with color spots on the problem places,
Thanks for looking, but I’m a bit confused.
Rev4B.GTL is the Top Layer. It’s supposed to have copper rest rings around all the holes. All BatchPCB board holes are “plated-thru”, which I think means they have copper inside the hole, all the thru the board and thru all layers. Whether or not a given hole connects to an inner layer depends on whether or not copper on that layer touches the hole.
So, I think the problem places (places where a hole is supposed to connect to GND but doesn’t) are supposed to have copper around them on the top layer. Unless I misunderstand (which is possible).
analogon:
and it shows solid black on the places without the problem.
I don’t understand how you’re seeing that. Gerbv seems to be a Linux app (I’m running Win7), so I can’t try it, but in Viewplot (another Gerber viewer) the top layer shows copper pads where they’re supposed to be for the GND contacts. I don’t see “black” (missing copper) there.
analogon:
I infer the black is referring to the places with copper, and the color is the places with copper removed. It looks like the negative of the image of the G2 layer on this thread.
I suspect the three pads at the upper left are getting connected via the GTL layer for the top 2 and the GBL layer for the lower of the three.
I’m just confused. Which 3 pads do you mean?
FartingMonkey92:
It does seem like an inverted gerber. EAGLE makes planes as traces, going back and forth to make a fill. so unless the CAM job was told to make an inverted output, it should be just like the traces on the rest of ya board.
Are the planes on completely seperate layers? Did you make your own CAM file?
What do the planes look like in the board editor?
Hmmm, why would it invert…
According to the EaglePCB manual, “A Supply layer is displayed and output inverted.” (page 281 of the v5.6 manual).
And on page 59 of the Tutorial, it says
“Supply layers defined with $… are plotted inversely, i.e., objects with the color of the supply layer define copperfree areas. The Thermal symbols connect the ground plane with the throughhole using four conducting paths.”
It doesn’t say WHY it does this - my guess is that it’s to make the layer easier to look at (since a supply layer is mostly copper with a few holes, while a normal layer is mostly bare with a few copper traces).
So I wasn’t totally shocked when the GND layer seemed to look inverted in both the Eagle layout and in the Gerber. But I don’t understand if the board was really made inverted or not. If it was inverted, then how are the GND pads that ARE connected, in fact connected? If it’s NOT inverted, then why don’t the thermal holes work?
nerdfever:
analogon:
I looked at the layers with Gerbv. It shows Rev4B.GTL with color spots on the problem places,
Thanks for looking, but I’m a bit confused.
Rev4B.GTL is the Top Layer. It’s supposed to have copper rest rings around all the holes. All BatchPCB board holes are “plated-thru”, which I think means they have copper inside the hole, all the thru the board and thru all layers. Whether or not a given hole connects to an inner layer depends on whether or not copper on that layer touches the hole.
So, I think the problem places (places where a hole is supposed to connect to GND but doesn’t) are supposed to have copper around them on the top layer. Unless I misunderstand (which is possible).
analogon:
and it shows solid black on the places without the problem.
I don’t understand how you’re seeing that. Gerbv seems to be a Linux app (I’m running Win7), so I can’t try it, but in Viewplot (another Gerber viewer) the top layer shows copper pads where they’re supposed to be for the GND contacts. I don’t see “black” (missing copper) there.
analogon:
I infer the black is referring to the places with copper, and the color is the places with copper removed. It looks like the negative of the image of the G2 layer on this thread.
I suspect the three pads at the upper left are getting connected via the GTL layer for the top 2 and the GBL layer for the lower of the three.
I’m just confused. Which 3 pads do you mean?
1. I made a mistake... I meant G2 edited the post to correct my posting.
-
Added that I was referring to the three pads just below the upper left mounting hole. I now see they are not relevant to your problem.
-
I am running Gerbv version 2.4.0 on Windows 7. http://sourceforge.net/projects/gerbv/f … rbv-2.4.0/ I am not saying that it is better than whatever else you might be running, but it’s what I had.
Note that my comments and observations on your board may not be useful or accurate. I am not experienced in this.
The only way i can see how to get this working how we want it to set layer 2 or 15 as a normal signal layer (untick supply layer in layer properties), set the layer options in the DRC as (12+316) or (12+1516) and make ground pours with the polygon tool on the layer you want.
This will make them display as normal without all this annoying “inverted-ness”… Don’t know what a fab house would say about them.