Losing the 4th decimal place would not be a problem especially since they all are zero anyway.
Ah, good. I asked because I work for a PCB manufacturer, and we don't consider that kind of change to be 'modifying customer data'. So naturally I thought I should consider a PCB customer's viewpoint on the issue.
Actually, the drill machines we use aren’t accurate enough for that 4th digit anyway. The drill table steps in .0001" increments, but that’s only a theoretical accuracy. I don’t know what the actual accuracy is, but I suspect it’s up to about plus/minus .002"
I just checked a couple of PCB acceptability standards tables, and the tightest hole location tolerace is +/- .002", which is considered ‘reduced producability’ in the military standard (MIL-STD-275E). The most generous tolerance is in the IPC-D-300G standard (Class A) which is +/- .008" !!
We prefer 2.3 trailing because these are the default settings for the CAMtastic quick import that we use.
We have customers submit all sorts of formats. We are not savvy enough to be able to look at drill file in the raw and say “Oh, that’s 4.5 leading”. So instead, we get to import and guess. When you get the format wrong, the drill coordinates are way off. So we try again.
Integer : 1-6
Decimal : 1-6
Type : Absolute or Incremental
Zero Suppression : Leading, Trailing, or None
So you see we have 662*3 or 216 possible cominations. Believe me, we will give up after four tries and email you a rejection notice.
It’s just a whole lot easier on us if we don’t have to guess your format. 2.3 trailing is what we try first.
Some additional info for Steve and Martyn:
Usual problems:
Drill sizes are missing
If drill sizes are there, the coordinates are not all the way there
If coordinates are good, the trailing/leading format is screwy
If we finally get the drill file imported, odds are a layer is mirrored
If the layers look good and lined up, they will probably fail DRC
If the board passes DRC, the board is probably missing a border
If the board actually has a border, there are probably connectors hanging over the border - so we get to touch up the board
With a board finally ready to be panelized, the customer probably failed to tell us how many copies they wanted in the email order
If we know the number of copies, we can finally get the boards onto the final panel
So then take 40+ orders and have them all fail at different spots on the list. We then email them and get responses and new files that fail again at different spots on the list. It’s not much fun.
If you can knock one or two off this list at a time, that would be tremendous!
Ahhh! Thanks, Nathan, that makes much more sense. I had assumed that since the confirmation email had no comments, then everything went well; but in reality it may have taken several attempts to get the files imported properly.
I have to write a short script to combine the two files thst my program produces anyway, I will convert to 2.3 at the same time.
Thanks for the detailed explanation. It might be an idea, if it was not too much trouble, to include in your confirmation email if files imported on the first attempt or if it took several attempts so that folks could fine tune their files to import the first time.
Humm, this is something possible in PHP. i cannot do anything yet, but in the next few days i hope to ‘release’ the online DRC @ a decent level of accuracy.
Well, I just submitted another board. My drill files are now converted to 2.3 and they and the rest of the files passed the online test and the other DRC tests which I use as well, so we’ll see if it goes through on the first pass.
By the way, I use CAM350 from Downstream Technologies for DRC checking. It is very expensive for the full version but the demo version which I use works great for any of the boards I have designed. It does more tests than I need but works quite well.
i cannot do anything yet, but in the next few days i hope to ‘release’ the online DRC @ a decent level of accuracy.