Does anyone know if the guys at batch PCB have stopped answering emails due to the Chinese new year? I have been emailing them for a week with no response.
Has anyone had success in submitting a design that fails their automated DRC bot? That bot seems to screw up a lot. In the last design I submitted the returned picture of the failure showed these big circles that are not in the Gerber file so the Bot screwed up. I was hoping they would acknowledge this problem and let me submit an order for this PCB. I have emailed them every day the last week and have not had any response
slammers:
Does anyone know if the guys at batch PCB have stopped answering emails due to the Chinese new year? I have been emailing them for a week with no response.
Has anyone had success in submitting a design that fails their automated DRC bot? That bot seems to screw up a lot. In the last design I submitted the returned picture of the failure showed these big circles that are not in the Gerber file so the Bot screwed up. I was hoping they would acknowledge this problem and let me submit an order for this PCB. I have emailed them every day the last week and have not had any response
Shawn
Shawn,
Batch PCB is a division of Spark Fun (Actually a spin off of Spark Fun). They do not celebrate the Chinese New Year. Their supplier of the boards does. This is based on how the service was working the last time I ordered. They may have restructured the process, for the board supplier to deal directly with the customers. I do not know.
David Carne was the contact for Batch PCB, but as I understand it, he is no longer employed with them. (It was not a quit or fire situation…it was more complicated than that.)
I’m not telling you to do this, but if it were me trying to get something answered…I would go to the beginning source.
I am pretty certain that BatchPCB is not going to process a board that failed DRC-- that is the point of the check. No question that the bot is not perfect and sometimes you need to tweak designs to make them pass (and I had one previous design which I could never get to pass). With that said, BatchPCB is a low-cost service that only works with high automation (else it doesn’t make financial sense). Thus, if the board does not pass DRC, they won’t make it.
Most likely they are ignoring your email. Again, as a barely break-even or even loss-leader business, they don’t have the time/manpower to analyze designs to determine why they failed DRC (even if it might be a DRC bug). The amount of time required to do the analysis of your board would exceed any possible “profit” on the order.
My suggestion is that you post details about your design, board and DRC bot output and see if folks here can help you diagnose the problem. There is likely a work-around which will let your design pass DRC and then you can submit you board.
I do disagree with James advice-- trying to figure out alternate ways to contact BatchPCB is a bad idea and does nothing but make them rethink the service.
Vraz:
I am pretty certain that BatchPCB is not going to process a board that failed DRC-- that is the point of the check. No question that the bot is not perfect and sometimes you need to tweak designs to make them pass (and I had one previous design which I could never get to pass). With that said, BatchPCB is a low-cost service that only works with high automation (else it doesn’t make financial sense). Thus, if the board does not pass DRC, they won’t make it.
Most likely they are ignoring your email. Again, as a barely break-even or even loss-leader business, they don’t have the time/manpower to analyze designs to determine why they failed DRC (even if it might be a DRC bug). The amount of time required to do the analysis of your board would exceed any possible “profit” on the order.
My suggestion is that you post details about your design, board and DRC bot output and see if folks here can help you diagnose the problem. There is likely a work-around which will let your design pass DRC and then you can submit you board.
I do disagree with James advice-- trying to figure out alternate ways to contact BatchPCB is a bad idea and does nothing but make them rethink the service.
Vraz,
The problem that I have, is the “Most likely they are ignoring your email.”
Any company should at least let you know, “you are wrong we are right, we will ignore you from now on.”
If this is going to be the norm, the service is already dead.
Even if the gerbers are wrong, someone needs to confirm right or wrong.
David C. was always good at letting you know your design was “all screwed up.”
But on the other hand…I didn’t ask if Slammer had ever submitted a design before, or if he had checked his gerbers with a gerber viewer.
I totally understand the need for them to keep costs down, it would just be nice to have a email out of courtesy saying there is nothing they can do if the bot fails. At this point I just do not know if they have received my emails or not.
Yes, I have submitted 6 or 7 designs to Batch PCB in the past. I use a quite high end CAD system from Mentor Graphics that has its own DRC and everything passes fine. I am not a novice at PCB design as this is what I do for a living. I just use Batch PCB for my own personal projects. Its a great service with awesome prices.
The design looks fine in Viewmate as well which is what they recommend before submitting the design.
The large circles on the left are some strange artifact the bot is creating. It must have something to do with the thermal reliefs to my ground plane as the bottom of each circle is intersecting a pad that has a thermal relief. Unfortunately two of these circles are coming close to the purple pads which is what is causing the bot to choke. The easiest thing would be to move the headers slightly but for design reasons I can not do this.
The weird thing is that other circles are completely intersecting other pads and it is not complaining about that. But then again if its intersecting they may just think its supposed to be like that. Its just small clearances that it is objecting to.
I’ve noticed something about the circles. Every one of them intersect what appears to be a round pad (possibly a via). The bottom of the circle starts at the round pad.
This is the only common factor I see.
What I don’t understand, is why all of your via’s or round pads do not create the circle.
Is there anything different for these that have the strange circles associated with them, that the others do not?
Also notice the the circle inner diameter is almost centered on the pad. Also…the pad on the right doesn’t seem to have a thermal relief. So I don’t think it is totally related to the reliefs. More about the pads themselves (or possibly holes of them).
I would think this would be a problem with the drill file more than the regular gerbers.
I am sure its ground plane related. There is probably some gerber commands for the ground plane that is causing the bot to draw wrong.
The last design I submitted, the DRCBot just drew one ground plane over the entire surface of the board. No pads were visible at all. So of course it passed as there was nothing to check. But the PCBs came back as I designed them.
So it must be something to do with the way PADS draws its ground plan that is incompatible with the DRC.
Maybe I will try using different CAM settings and see if that has any effect.
I regenerated the CAM output on a newer version of Pads. Now the pads that had circles or completely filled in (by the bot, they still have thermal reliefs in the gerber viewer)
So now I do not get the clearence errors but there is one error and I have no idea what it is.
What is really strange is that the “offending” thermals show as solid, but there are other thermals on the board which show up correctly. Actually, looking more closely, all the pads which had the weird circles (including the one in the center of the board and the one on the right side) are solid now. So it seems there was something about the old output which caused the strange circles, but apparently it was actually the collision of those strange circles with the pads which caused the DRC clearance error. Very odd stuff.
I have seen other cases where applications handled Gerbers in different ways. Things like zero length line segments can cause strange behavior. Might be interesting the diff the old/new Gerber and see if you can spot the difference, though given the data volume, it might be pretty time consuming and thus impractical.
With respect to my earlier comment that they are most likely ignoring your email, I am not defending that as the appropriate response, but a likely one given the realities of their business model. They should have some “canned” responses ready to send back in such cases, but maybe they are busy or didn’t have anything appropriate. Unfortunately, they don’t know the experience level of the person submitting the message or what steps they have taken. Based on a few prior email samples they have posted, I can see them being a bit cautious at this point. The good part is that it looks like there is a workaround which is the real goal.
Whoops-- didn’t appreciate that while the other issues were gone, its still complaining about one. I don’t think the log data you posted is the problem as it essentially matches what I got from a recent passing design:
If I were to guess, the MergeCount is probably the number of polygons it merges together and tests is somehow the number of boundary tests. It then figures out the min/max x/y coordinates (the “rect”) and then subtracts them to get the size. No idea what “groups” or “lengthdb” mean, but our values seem similar.
Was there anything on the bottom later or the bottom log which looked suspicious?
Don’t think I have encountered a situation where the DRC bot returned an error but gave no indication of the cause (not even one color pixel). Will review some of my prior failures to see if anything rings a bell (and I have a few of those).
That was the end of the file so there was nothing after that.
There was another strange thing. It returned a log for both top and bottom but only a picture for the top side. I usually always get pictures of both sides. The bottom side log file had no errors either.
The bot doesn’t like copper pours that my PCB program (Pulsonix) generates. You could try removing the pour and running it past the bot again. I know the design won’t work, but will narrow down if the pour is the cause or not.
Also, the generated pictures for my boards come back looking wierd, but most of the time the PCB is made correctly.
