via in pad

[John Larkin]

We're trying to decide on a new ST ARM chip. They come in 0.8 mm and
0.65 mm ball pitch. We are already comfortable with 0.8.

The 0.65 is preferred by some parties here, as it has more I/Os; 140
vs 82, but then 82 is still a lot.

Dog-bones and vias are tricky for the 0.65 version. The recommended
(ST app note) via drill is 0.2 mm (8 mils) and the finished/plated via
hole is 0.05 mm (2 mils.) Wow.

With a 2 mil finished plated via, would via-in-pad be practical? Not
much solder is going to slurp into a 2 mil hole.

==========

I know a guy who was working with a chip with 0.25 mm ball pitch. I
haven't heard how that worked out.

We are using the EPC GaN fets, BGA with 0.45 mm pitch. But only 4
balls and no dog-bones. We had some problems but it seems to be under
control now.

 

[Lasse Langwadt Christense]

onsdag den 16. oktober 2019 kl. 01.40.33 UTC+2 skrev John Larkin:

We're trying to decide on a new ST ARM chip. They come in 0.8 mm and
0.65 mm ball pitch. We are already comfortable with 0.8.

The 0.65 is preferred by some parties here, as it has more I/Os; 140
vs 82, but then 82 is still a lot.

Dog-bones and vias are tricky for the 0.65 version. The recommended
(ST app note) via drill is 0.2 mm (8 mils) and the finished/plated via
hole is 0.05 mm (2 mils.) Wow.

With a 2 mil finished plated via, would via-in-pad be practical? Not
much solder is going to slurp into a 2 mil hole.

==========

I know a guy who was working with a chip with 0.25 mm ball pitch. I
haven't heard how that worked out.

We are using the EPC GaN fets, BGA with 0.45 mm pitch. But only 4
balls and no dog-bones. We had some problems but it seems to be under
control now.

you can get the via filled and plated over

https://www.pcbuniverse.com/pcbu-tech-tips.php?a=5

 

On Tue, 15 Oct 2019 17:59:19 -0700 (PDT), Lasse Langwadt Christensen
<langwadt@fonz.dk> wrote:

onsdag den 16. oktober 2019 kl. 01.40.33 UTC+2 skrev John Larkin:
We're trying to decide on a new ST ARM chip. They come in 0.8 mm and
0.65 mm ball pitch. We are already comfortable with 0.8.

The 0.65 is preferred by some parties here, as it has more I/Os; 140
vs 82, but then 82 is still a lot.

Dog-bones and vias are tricky for the 0.65 version. The recommended
(ST app note) via drill is 0.2 mm (8 mils) and the finished/plated via
hole is 0.05 mm (2 mils.) Wow.

With a 2 mil finished plated via, would via-in-pad be practical? Not
much solder is going to slurp into a 2 mil hole.

==========

I know a guy who was working with a chip with 0.25 mm ball pitch. I
haven't heard how that worked out.

We are using the EPC GaN fets, BGA with 0.45 mm pitch. But only 4
balls and no dog-bones. We had some problems but it seems to be under
control now.

you can get the via filled and plated over

https://www.pcbuniverse.com/pcbu-tech-tips.php?a=5

Right, but that adds a lot of process steps to the board fab,
something like 15 by one account. Sounds expensive.



--

John Larkin Highland Technology, Inc

lunatic fringe electronics

 

[Robert Baer]

John Larkin wrote:

We're trying to decide on a new ST ARM chip. They come in 0.8 mm and
0.65 mm ball pitch. We are already comfortable with 0.8.

The 0.65 is preferred by some parties here, as it has more I/Os; 140
vs 82, but then 82 is still a lot.

Dog-bones and vias are tricky for the 0.65 version. The recommended
(ST app note) via drill is 0.2 mm (8 mils) and the finished/plated via
hole is 0.05 mm (2 mils.) Wow.

With a 2 mil finished plated via, would via-in-pad be practical? Not
much solder is going to slurp into a 2 mil hole.

* Seems even solder has problems filing a 4 mil hole. Then again it
matters not.

==========

I know a guy who was working with a chip with 0.25 mm ball pitch. I
haven't heard how that worked out.

We are using the EPC GaN fets, BGA with 0.45 mm pitch. But only 4
balls and no dog-bones. We had some problems but it seems to be under
control now.

 

On Wednesday, 16 October 2019 05:16:18 UTC+1, Robert Baer wrote:

John Larkin wrote:


We're trying to decide on a new ST ARM chip. They come in 0.8 mm and
0.65 mm ball pitch. We are already comfortable with 0.8.

The 0.65 is preferred by some parties here, as it has more I/Os; 140
vs 82, but then 82 is still a lot.

Dog-bones and vias are tricky for the 0.65 version. The recommended
(ST app note) via drill is 0.2 mm (8 mils) and the finished/plated via
hole is 0.05 mm (2 mils.) Wow.

With a 2 mil finished plated via, would via-in-pad be practical? Not
much solder is going to slurp into a 2 mil hole.
* Seems even solder has problems filing a 4 mil hole. Then again it
matters not.


==========

I know a guy who was working with a chip with 0.25 mm ball pitch. I
haven't heard how that worked out.

We are using the EPC GaN fets, BGA with 0.45 mm pitch. But only 4
balls and no dog-bones. We had some problems but it seems to be under
control now.

I once accidentally put a via in the middle of a 0.5mm pitch BGA pad.
The via sucked the whole ball away, even though the volume of the
via was less than the ball volume. The board was rescued by cleaning
the remaining solder out of the via and forcing a stiff steel wire through
the via until it contacted the underside of the BGA. Fortunately this
was only a prototype. I may still have the X-ray images.

The bottom line: use plugged vias in pads under BGAs.

John

 

On Wed, 16 Oct 2019 13:07:30 +0000 (UTC),
DecadentLinuxUserNumeroUno@decadence.org wrote:

krw@notreal.com wrote in
news:c14eqetrkeeep6mq0ft6mvlhrpllsc0hgq@4ax.com:

On Tue, 15 Oct 2019 21:03:39 -0700,
jlarkin@highlandsniptechnology.com wrote:

On Tue, 15 Oct 2019 17:59:19 -0700 (PDT), Lasse Langwadt
Christensen <langwadt@fonz.dk> wrote:

onsdag den 16. oktober 2019 kl. 01.40.33 UTC+2 skrev John Larkin:
We're trying to decide on a new ST ARM chip. They come in 0.8
mm and 0.65 mm ball pitch. We are already comfortable with 0.8.

The 0.65 is preferred by some parties here, as it has more
I/Os; 140 vs 82, but then 82 is still a lot.

Dog-bones and vias are tricky for the 0.65 version. The
recommended (ST app note) via drill is 0.2 mm (8 mils) and the
finished/plated via hole is 0.05 mm (2 mils.) Wow.

With a 2 mil finished plated via, would via-in-pad be
practical? Not much solder is going to slurp into a 2 mil hole.

==========

I know a guy who was working with a chip with 0.25 mm ball
pitch. I haven't heard how that worked out.

We are using the EPC GaN fets, BGA with 0.45 mm pitch. But only
4 balls and no dog-bones. We had some problems but it seems to
be under control now.

you can get the via filled and plated over

https://www.pcbuniverse.com/pcbu-tech-tips.php?a=5

Right, but that adds a lot of process steps to the board fab,
something like 15 by one account. Sounds expensive.

Your production cost is a small part of your total cost so it
shouldn't be a bank buster. You're able to justify the cost of LT
parts, so filled vias should be on the menu where it gives any
advantage at all.


He can manage the via by placing mask over them. Being in a pad
negates that, so he can manage those that are 'in pad' areas by
ignoring them. A 4 mil via does NOT soak up enough solder to worry
about.

AlwaysWrong insists on being wrong.

Worried, add a mil to the stencil thickness. A bit cheaper than
having them filled by the fabs or your production floor.

Or, place a via in all four locations so the wicking effect is
equal on all four and thus the attachment integrity is the same
across all four.

Always.

 

On Wednesday, 16 October 2019 14:07:36 UTC+1, DecadentLinux...@decadence.org wrote:

A 4 mil via does NOT soak up enough solder to worry
about.

My experience is the opposite. I only have one data point because I
took care never to make that mistake again.

John

 

krw@notreal.com wrote in
news:c14eqetrkeeep6mq0ft6mvlhrpllsc0hgq@4ax.com:

On Tue, 15 Oct 2019 21:03:39 -0700,
jlarkin@highlandsniptechnology.com wrote:

On Tue, 15 Oct 2019 17:59:19 -0700 (PDT), Lasse Langwadt
Christensen <langwadt@fonz.dk> wrote:

onsdag den 16. oktober 2019 kl. 01.40.33 UTC+2 skrev John Larkin:
We're trying to decide on a new ST ARM chip. They come in 0.8
mm and 0.65 mm ball pitch. We are already comfortable with 0.8.

The 0.65 is preferred by some parties here, as it has more
I/Os; 140 vs 82, but then 82 is still a lot.

Dog-bones and vias are tricky for the 0.65 version. The
recommended (ST app note) via drill is 0.2 mm (8 mils) and the
finished/plated via hole is 0.05 mm (2 mils.) Wow.

With a 2 mil finished plated via, would via-in-pad be
practical? Not much solder is going to slurp into a 2 mil hole.

==========

I know a guy who was working with a chip with 0.25 mm ball
pitch. I haven't heard how that worked out.

We are using the EPC GaN fets, BGA with 0.45 mm pitch. But only
4 balls and no dog-bones. We had some problems but it seems to
be under control now.

you can get the via filled and plated over

https://www.pcbuniverse.com/pcbu-tech-tips.php?a=5

Right, but that adds a lot of process steps to the board fab,
something like 15 by one account. Sounds expensive.

Your production cost is a small part of your total cost so it
shouldn't be a bank buster. You're able to justify the cost of LT
parts, so filled vias should be on the menu where it gives any
advantage at all.

He can manage the via by placing mask over them. Being in a pad
negates that, so he can manage those that are 'in pad' areas by
ignoring them. A 4 mil via does NOT soak up enough solder to worry
about.

Worried, add a mil to the stencil thickness. A bit cheaper than
having them filled by the fabs or your production floor.

Or, place a via in all four locations so the wicking effect is
equal on all four and thus the attachment integrity is the same
across all four.

 

On Tue, 15 Oct 2019 21:03:39 -0700, jlarkin@highlandsniptechnology.com
wrote:

On Tue, 15 Oct 2019 17:59:19 -0700 (PDT), Lasse Langwadt Christensen
langwadt@fonz.dk> wrote:

onsdag den 16. oktober 2019 kl. 01.40.33 UTC+2 skrev John Larkin:
We're trying to decide on a new ST ARM chip. They come in 0.8 mm and
0.65 mm ball pitch. We are already comfortable with 0.8.

The 0.65 is preferred by some parties here, as it has more I/Os; 140
vs 82, but then 82 is still a lot.

Dog-bones and vias are tricky for the 0.65 version. The recommended
(ST app note) via drill is 0.2 mm (8 mils) and the finished/plated via
hole is 0.05 mm (2 mils.) Wow.

With a 2 mil finished plated via, would via-in-pad be practical? Not
much solder is going to slurp into a 2 mil hole.

==========

I know a guy who was working with a chip with 0.25 mm ball pitch. I
haven't heard how that worked out.

We are using the EPC GaN fets, BGA with 0.45 mm pitch. But only 4
balls and no dog-bones. We had some problems but it seems to be under
control now.

you can get the via filled and plated over

https://www.pcbuniverse.com/pcbu-tech-tips.php?a=5

Right, but that adds a lot of process steps to the board fab,
something like 15 by one account. Sounds expensive.

Your production cost is a small part of your total cost so it
shouldn't be a bank buster. You're able to justify the cost of LT
parts, so filled vias should be on the menu where it gives any
advantage at all.

 

On Wed, 16 Oct 2019 06:21:16 -0700 (PDT), jrwalliker@gmail.com wrote:

On Wednesday, 16 October 2019 14:07:36 UTC+1, DecadentLinux...@decadence.org wrote:

A 4 mil via does NOT soak up enough solder to worry
about.

My experience is the opposite. I only have one data point because I
took care never to make that mistake again.

John

Did you do via in pad on BGAs?

What happened?



--

John Larkin Highland Technology, Inc

lunatic fringe electronics

 

[DemonicTubes]

On Tuesday, October 15, 2019 at 5:40:33 PM UTC-6, John Larkin wrote:

With a 2 mil finished plated via, would via-in-pad be practical? Not
much solder is going to slurp into a 2 mil hole.

I think you will be surprised by how much trouble you will have with this.

I would highly recommend having them filled and plated by the board house. The modern ones do this all the time and are well setup for it, so the cost shouldn't be significant enough to gamble against the trouble you will bring yourself.

 

[LM]

ST has some info somewhere about how how to wire the PCB.
On Tue, 15 Oct 2019 16:40:23 -0700, John Larkin
<jlarkin@highland_atwork_technology.com> wrote:

We're trying to decide on a new ST ARM chip. They come in 0.8 mm and
0.65 mm ball pitch. We are already comfortable with 0.8.

The 0.65 is preferred by some parties here, as it has more I/Os; 140
vs 82, but then 82 is still a lot.

Dog-bones and vias are tricky for the 0.65 version. The recommended
(ST app note) via drill is 0.2 mm (8 mils) and the finished/plated via
hole is 0.05 mm (2 mils.) Wow.

With a 2 mil finished plated via, would via-in-pad be practical? Not
much solder is going to slurp into a 2 mil hole.

==========

I know a guy who was working with a chip with 0.25 mm ball pitch. I
haven't heard how that worked out.

We are using the EPC GaN fets, BGA with 0.45 mm pitch. But only 4
balls and no dog-bones. We had some problems but it seems to be under
control now.

 

On Wednesday, 16 October 2019 16:02:45 UTC+1, jla...@highlandsniptechnology.com wrote:

On Wed, 16 Oct 2019 06:21:16 -0700 (PDT), jrwalliker@gmail.com wrote:

On Wednesday, 16 October 2019 14:07:36 UTC+1, DecadentLinux...@decadence.org wrote:

A 4 mil via does NOT soak up enough solder to worry
about.

My experience is the opposite. I only have one data point because I
took care never to make that mistake again.

John

Did you do via in pad on BGAs?

What happened?

I accidentally put a via in the middle of a pad. It was 0.5mm pitch
BGA package. The ball completely disappeared. Production of the
prototype was done by a very competent UK contract manufacturer
(the one who now makes Raspberry Pis).
I may be able to find an X-ray image.

John

 

[Dimiter_Popoff]

On 10/16/2019 2:40, John Larkin wrote:

We're trying to decide on a new ST ARM chip. They come in 0.8 mm and
0.65 mm ball pitch. We are already comfortable with 0.8.

The 0.65 is preferred by some parties here, as it has more I/Os; 140
vs 82, but then 82 is still a lot.

Dog-bones and vias are tricky for the 0.65 version. The recommended
(ST app note) via drill is 0.2 mm (8 mils) and the finished/plated via
hole is 0.05 mm (2 mils.) Wow.

With a 2 mil finished plated via, would via-in-pad be practical? Not
much solder is going to slurp into a 2 mil hole.

==========

I know a guy who was working with a chip with 0.25 mm ball pitch. I
haven't heard how that worked out.

We are using the EPC GaN fets, BGA with 0.45 mm pitch. But only 4
balls and no dog-bones. We had some problems but it seems to be under
control now.

My experience is with vias in pad for 1.27 pitched BGAs but it should be
relevant.

You should not worry about how much solder is soaked up into the via,
it is negligible (even with 0.2mm holes as we have been doing it for
decades now).
However you do have to worry about solder balls flowing through the
hole entirely, leaving nothing to connect the BGA pad to the PCB pad.
Happens when a solder ball comes coldly attached to the BGA case; about
1 ball in 100 in my experience are like that.
The remedy is simple - just bake the BGA balls up with enough decent
flux prior to putting it on the board so there are no cold balls.
For our low quantity products this is not an issue, don't know if it
would be one for you.
But trying to bake the BGA right away onto a board with holes in
the pads (we do all BGA pads with holes) is almost 100% sure to
get you into trouble.

Dimiter

======================================================
Dimiter Popoff, TGI http://www.tgi-sci.com
======================================================
http://www.flickr.com/photos/didi_tgi/