32 bit USB uC in 48pins

Some shop won't do less than 16 mil holes. Production shop can handle it, but not prototype shop. I have to use both.

P packages:

than

=A0P =A0 =A0 =A0$ =A0 Wish

N ~LIN

MMU

half

tch

nds

k the

Ok, so I'm not crazy... well I guess that doesn't completely follow.

Only if you go below the minimum of the board house. I haven't seen any that won't support 15 mil without extra charge and most will go down to 12 mil or even 10. Live life dangerously and go with a 15 mil drill!

One advantage of larger via holes is that they hold the tip of an o'scope probe better. :)

Rick

QFP packages:

ss than

=A0 =A0P =A0 =A0 =A0$ =A0 Wish

CAN ~LIN

~MMU

sh *=3D 2

nly

ut half

pitch

ends

mm

ink the

ard

you

I guess it's my bad old habit of dealing with proto shops. Unless they are laser drilled, smaller drills cost them. Small bits break more often. They might not charge extra for the smaller holes, but would make it more difficult to negotiate other discounts.

TQFP packages:

less than

=A0 =A0P =A0 =A0 =A0$ =A0 Wish

~CAN ~LIN

=A0 ~MMU

lash *=3D 2

only

bout half

m pitch

ze ends

think the

board

e,

the

it

via

t you

After dropping AVR32 PIC24 and MCF51, it's exclusively Cortex-M0 and M1 now. But I am still open to suggestions for other chips. Current work in progress in NUC120 (#1) and LPC1343 (#2) in this layout

My spec: L(30)

TQFP packages:

less than

=A0 =A0P =A0 =A0 =A0$ =A0 Wish

~CAN ~LIN

=A0 ~MMU

lash *=3D 2

only

bout half

m pitch

ze ends

think the

board

e,

the

it

via

t you

I don't know which proto shops you are working with. The worst I have seen is for a house to take my 10 mil drill spec, add their "standard tolerance" of 0.003" to make it a 13 mil drill without asking/telling me. I've never had anyone take any issue with a 15 mil drill. I think I have seen a board house in Eastern Europe state that they don't use drills smaller than 20 mil, but that is because they are really, really cheap. I don't mind spending the extra $100 bucks on my proto cards. I focus on making my production boards cheap. Smaller vias can mean a smaller board or fewer layers on a tight design... or in this case, more choices in MCU chips in 44/64 pin packages.

BTW, are you considering QFN packages? I have only worked with them a little, but I've never seen any problems and they are typically 2-3 mm smaller than the equivalent QFP. I find them a bit harder to probe, but they are still "probable", unlike BGAs. Well, other than the weird, high density parts with three rows of pads (Actel 132 pin QFN).

Rick

That sort of requirement for a proto ship, was acceptable 15 years ago.

I have not worried about any proto shop doing smaller than that for years. Like I have not worried about number of holes for years either.

As an example (you can check online) I regularly from UK use PCB Pool and their min requirements for some time have been

Standard min requirements

=09Min track=090.15mm (6 mil) Min drill 0.3mm (12 mil)

More expensive min

=09Min track=090.125mm (5 mil) Min drill 0.2mm (8 mil)

This is for prototype quantities... I regularly pepper my double sided boards ground planes with extra 0.3mm vias to bind the ground plane on both sides.

Don't even get into high temperature or other special requirements for PCB materials. =20

As shown above.

--=20 Paul Carpenter | snipped-for-privacy@pcserviceselectronics.co.uk PC Services Timing Diagram Font GNU H8 - compiler & Renesas H8/H8S/H8 Tiny For those web sites you hate

in TQFP packages:

(less than

=A0L =A0 =A0P =A0 =A0 =A0$ =A0 Wish

=A0 ~CAN ~LIN

=A0 ~MMU

Flash *=3D 2

rd only

about half

mm pitch

size ends

by

I think the

e board

sue,

s the

on it

e via

n't you

mil

Yes, the initial version is a 64 pads QFN. You cannot put any vias in the center, because of the ground pad. Any one of the four TQFP48 will reduce the PCB space by 25%. If you can show me a 64 pins that is more cost effective than my list, I am all ears.

QFN64 and TQFP64 are same size (0.5mm pitch).

..

So, this is still false.

My supplier: sub-standard min: 7 mil track/space, 16 mil holes.

Then why not be on the safe side with 0.3mm (12mil) vias all the time?

I can do the first four easily with 7(lines)/16(holes) with LQFP48. I don't see much advantage in other sizes. NUC120 would be the best, if they have a real distributor coming near me. LPC1343 would be better with more flash. I wish there is a version of STM without CAN and LIN and SAM3 without MMU, but with lower cost.

My spec: L(30)

See:

I am curious why do you have the via at the left edge of the right 12M pad? The notch of the cyan (GND?) fill it's sitting in looks strange, and the via could potentially suck the solder off the pad.

The 12M crystal pads are on the top layer only. It is not connected to the ground fill on the bottom. The two holes (not vias) allow pinned crystals (HCS-49), rather than smd crystals.

nd

The fact remains that it is not a good idea to have holes or vias adjacent to pads if they are plated. Although as long as you are hand soldering the SMD part, you will eventually fill the via with solder and then the part will have enough to hold it.

Rick