The 76 micron number is the standard for drill positioning accuracy because
that is what standard industrial solid carbide industrial bits can hit. It
is kind of interesting that the diameter of the bit doesn't matter: Small
bits tend to "whip" at the >100krpm used for drilling, while large bits
"walk" on the little flat line that is at the bottom of standard bits (not
split-point). Positioning error on a good CNC drill is below 25 microns.
Alignment of drill to internal features can be another 76 microns. Throw in
laminating press stackup errors, then final etch artwork alignment to
drilled holes. A lot of board routers align optically to external features,
so maybe that portion is very small for a high end fabricator.

A high throughput design will also stack a pile of boards in the drill at
the same time, so you can have larger drill alignment errors due to the
"wandering" caused by a longer bit and uneven chip compaction during
drilling.

There are ways of dealing with all of the inaccuracies noted above, but
they are usually time consuming, so they cost more. Lasers are relatively
slow, and have severely limited depth capability, but no "whip" or "walk",
and can be programmed for optical pattern recognition for alignment.
Likewise Laser Direct Imaging can be used to target chemical etch and
automatically adjust artwork (either locally, on a single "up" or globally
across the panel.

Wayne Thayer

On Thu, Aug 13, 2020 at 4:56 AM Bush, Jeffrey D. (US) <
[log in to unmask]> wrote:

> Generally the DTP is specified for all characteristics in the PCB - Drill,
> Image, Lamination, etc and .010" DTP is a standard value with many shops
> improving that capability for certain designs.  With LDI and real-time
> Drill optimization systems a .007" DTP is viable.
>
> Jeffrey Bush
> Program Quality Engineering Manager
> BAE SYSTEMS
> Electronic Systems Division
> C4ISR Solutions – Space Systems Group
> P.O. Box 868
> Nashua, NH 03061-0868
> T: (603) 885-8137  |  M: (603) 318.8056   |  E:
> [log in to unmask]
> 144 DW Highway, Merrimack, NH 03054  Mail Stop: MER24-116A-A4
>
> Important:
> This email may contain BAE SYSTEMS INCORPORATED Proprietary Information
> for the sole use of the intended recipient. Any review or distribution by
> others is strictly prohibited.  If you are not the intended recipient,
> please contact the sender and delete all copies.
>
> -----Original Message-----
> From: TechNet <[log in to unmask]> On Behalf Of Dwight Mattix
> Sent: Wednesday, August 12, 2020 5:35 PM
> To: [log in to unmask]
> Subject: Re: [TN] Drilling positional tolerance (hole-to-hole)
>
> *** WARNING ***
> EXTERNAL EMAIL -- This message originates from outside our organization.
>
>
> Drill position to etched pattern std capability is +/- 3 mils (76um).
>
>
>
>
> -----Original Message-----
> From: TechNet <[log in to unmask]> On Behalf Of Jack Olson
> Sent: Wednesday, August 12, 2020 1:40 PM
> To: [log in to unmask]
> Subject: [EXT] [TN] Drilling positional tolerance (hole-to-hole)
>
> If a spec says that PCB drilling positional tolerance is .0762 mm, doesn't
> that mean that the drill hole PATTERN can be off that much in relation to
> other features?
> (like to board outline or copper pattern?) What is a safe number for
> hole-to-hole positional tolerance?
>