PCB Layout for BGAs...

[gnuarm.del...@gmail.com]

A small board with a 100QFP is being redesigned for a new FPGA due to obsolescence. Gowin makes a 100QFP device that would be a good fit, but my customer has said \"no\" to the 100% Chinese brand... US government customers, ya know!

So now I\'m looking at a BGA. I don\'t want to get into fine PCB design rules, so 1.0 mm ball pitch is my preference. The only devices I can find that fit on the board have 196 or 256 pins. But the real problem is availability.

Digikey has a few of the XC7S15-1FTGB196I and more a scheduled for delivery in April. Add in the various speed and temperature flavors trickling in (mostly in April) and I should be ok for the initial delivery in August... if I can get my hands on those. I don\'t know if Digikey factors in the backlog orders in these counts.

Mouser shows great inventory of Efinix parts, particularly the T13 and T20 in a 0.8 mm 256 pin BGA, 10s of thousands in stock. But I\'d rather work with a 1.0 mm BGA. Oddly enough, LCSC shows part numbers, but zero inventory..

Anyone work with 0.8 mm BGAs? What PWB feature dimensions did you use? Did this impact the PWB cost?

--

Rick C.

- Get 1,000 miles of free Supercharging
- Tesla referral code - https://ts.la/richard11209

 

[David Brown]

On 07/01/2023 18:49, gnuarm.del...@gmail.com wrote:

A small board with a 100QFP is being redesigned for a new FPGA due to
obsolescence. Gowin makes a 100QFP device that would be a good fit,
but my customer has said \"no\" to the 100% Chinese brand... US
government customers, ya know!

So now I\'m looking at a BGA. I don\'t want to get into fine PCB
design rules, so 1.0 mm ball pitch is my preference. The only
devices I can find that fit on the board have 196 or 256 pins. But
the real problem is availability.

Digikey has a few of the XC7S15-1FTGB196I and more a scheduled for
delivery in April. Add in the various speed and temperature flavors
trickling in (mostly in April) and I should be ok for the initial
delivery in August... if I can get my hands on those. I don\'t know
if Digikey factors in the backlog orders in these counts.

Mouser shows great inventory of Efinix parts, particularly the T13
and T20 in a 0.8 mm 256 pin BGA, 10s of thousands in stock. But I\'d
rather work with a 1.0 mm BGA. Oddly enough, LCSC shows part
numbers, but zero inventory.

Anyone work with 0.8 mm BGAs? What PWB feature dimensions did you
use? Did this impact the PWB cost?

It\'s difficult to know the impact without knowing details of the board
you have at the moment. It is also somewhat dependent on the layout of
the balls on the part - some BGA\'s have missing balls, or their central
balls all connected to ground to make layout easier.

In general, 0.8 mm pitch should be doable in four layers, but you might
need finer tracks and clearances than you used before. In a recent
board we did, the 0.8 mm pitch BGA was fine with four layers and normal
vias. We did not need to switch to 6 layers with advanced vias until we
moved to the 0.65 mm version.

 

[gnuarm.del...@gmail.com]

On Sunday, January 8, 2023 at 7:52:02 AM UTC-5, David Brown wrote:

On 07/01/2023 18:49, gnuarm.del...@gmail.com wrote:
A small board with a 100QFP is being redesigned for a new FPGA due to
obsolescence. Gowin makes a 100QFP device that would be a good fit,
but my customer has said \"no\" to the 100% Chinese brand... US
government customers, ya know!

So now I\'m looking at a BGA. I don\'t want to get into fine PCB
design rules, so 1.0 mm ball pitch is my preference. The only
devices I can find that fit on the board have 196 or 256 pins. But
the real problem is availability.

Digikey has a few of the XC7S15-1FTGB196I and more a scheduled for
delivery in April. Add in the various speed and temperature flavors
trickling in (mostly in April) and I should be ok for the initial
delivery in August... if I can get my hands on those. I don\'t know
if Digikey factors in the backlog orders in these counts.

Mouser shows great inventory of Efinix parts, particularly the T13
and T20 in a 0.8 mm 256 pin BGA, 10s of thousands in stock. But I\'d
rather work with a 1.0 mm BGA. Oddly enough, LCSC shows part
numbers, but zero inventory.

Anyone work with 0.8 mm BGAs? What PWB feature dimensions did you
use? Did this impact the PWB cost?

It\'s difficult to know the impact without knowing details of the board
you have at the moment. It is also somewhat dependent on the layout of
the balls on the part - some BGA\'s have missing balls, or their central
balls all connected to ground to make layout easier.

In general, 0.8 mm pitch should be doable in four layers, but you might
need finer tracks and clearances than you used before. In a recent
board we did, the 0.8 mm pitch BGA was fine with four layers and normal
vias. We did not need to switch to 6 layers with advanced vias until we
moved to the 0.65 mm version.

What are \"normal\" vias and track/space widths? All of the BGAs I\'m looking at are solid arrays. I\'d give you identifiers of the packages, but I\'ve never found standard terms. Everyone has unique identifiers and more than one! Packaging is insane! I especially love how TI adds it to their data sheets with the reel packing and boxing data ahead of the package mechanical data. I\'m sure that stuff is important to someone, but they could have a separate data sheet with all that. Some companies do that with the package mechanical data. Gotta love diversity... or not.

My board is already 6 layers because of the high density and narrow width, 23 mm x 115 mm. One layer is pretty much nothing but lengthwise routing and it\'s still very congested. Surface layers are hard to route on. Between the parts and the vias, there\'s no room left. In the new layout, a couple of larger chips are going bye bye, so maybe that will improve things a bit.. But they are asking for some new features that will add some chips, so it may be an even swap.

--

Rick C.

+ Get 1,000 miles of free Supercharging
+ Tesla referral code - https://ts.la/richard11209

 

[David Brown]

On 08/01/2023 20:01, gnuarm.del...@gmail.com wrote:

On Sunday, January 8, 2023 at 7:52:02 AM UTC-5, David Brown wrote:
On 07/01/2023 18:49, gnuarm.del...@gmail.com wrote:
A small board with a 100QFP is being redesigned for a new FPGA
due to obsolescence. Gowin makes a 100QFP device that would be a
good fit, but my customer has said \"no\" to the 100% Chinese
brand... US government customers, ya know!

So now I\'m looking at a BGA. I don\'t want to get into fine PCB
design rules, so 1.0 mm ball pitch is my preference. The only
devices I can find that fit on the board have 196 or 256 pins.
But the real problem is availability.

Digikey has a few of the XC7S15-1FTGB196I and more a scheduled
for delivery in April. Add in the various speed and temperature
flavors trickling in (mostly in April) and I should be ok for the
initial delivery in August... if I can get my hands on those. I
don\'t know if Digikey factors in the backlog orders in these
counts.

Mouser shows great inventory of Efinix parts, particularly the
T13 and T20 in a 0.8 mm 256 pin BGA, 10s of thousands in stock.
But I\'d rather work with a 1.0 mm BGA. Oddly enough, LCSC shows
part numbers, but zero inventory.

Anyone work with 0.8 mm BGAs? What PWB feature dimensions did
you use? Did this impact the PWB cost?

It\'s difficult to know the impact without knowing details of the
board you have at the moment. It is also somewhat dependent on the
layout of the balls on the part - some BGA\'s have missing balls, or
their central balls all connected to ground to make layout easier.

In general, 0.8 mm pitch should be doable in four layers, but you
might need finer tracks and clearances than you used before. In a
recent board we did, the 0.8 mm pitch BGA was fine with four layers
and normal vias. We did not need to switch to 6 layers with
advanced vias until we moved to the 0.65 mm version.

What are \"normal\" vias and track/space widths? All of the BGAs I\'m
looking at are solid arrays. I\'d give you identifiers of the
packages, but I\'ve never found standard terms. Everyone has unique
identifiers and more than one! Packaging is insane! I especially
love how TI adds it to their data sheets with the reel packing and
boxing data ahead of the package mechanical data. I\'m sure that
stuff is important to someone, but they could have a separate data
sheet with all that. Some companies do that with the package
mechanical data. Gotta love diversity... or not.

My board is already 6 layers because of the high density and narrow
width, 23 mm x 115 mm. One layer is pretty much nothing but
lengthwise routing and it\'s still very congested. Surface layers are
hard to route on. Between the parts and the vias, there\'s no room
left. In the new layout, a couple of larger chips are going bye bye,
so maybe that will improve things a bit. But they are asking for
some new features that will add some chips, so it may be an even
swap.

By \"normal\" vias I mean through-hole, without plugging or tenting - the
cheap ones. And by \"normal\" track widths and spacing, I mean the
perhaps 6 mil - cheap sizes that any pcb manufacturer can make without
extra charge. Basically, at 0.8 mm BGA we didn\'t need to do anything
special or consider it an especially high-density card. It was still an
effort to route, as there are a lot of connections in a small area. But
it didn\'t need extra cost for the pcb.

If you are already congested at 6 layers on the board, then you will
might have to go for beyond that. The first step is to talk to your pcb
and board manufacturers about via-in-pad, using tented or plugged vias.
If you can put the vias on the pads themselves, without causing voids
or blowouts during soldering, you can save a /lot/ of space. Next step
beyond that is microvias from outer layers to layers 2 and 5. We had to
do that for the 0.65 mm package BGA.

 

[gnuarm.del...@gmail.com]

On Monday, January 9, 2023 at 4:02:43 AM UTC-5, David Brown wrote:

On 08/01/2023 20:01, gnuarm.del...@gmail.com wrote:
On Sunday, January 8, 2023 at 7:52:02 AM UTC-5, David Brown wrote:
On 07/01/2023 18:49, gnuarm.del...@gmail.com wrote:
A small board with a 100QFP is being redesigned for a new FPGA
due to obsolescence. Gowin makes a 100QFP device that would be a
good fit, but my customer has said \"no\" to the 100% Chinese
brand... US government customers, ya know!

So now I\'m looking at a BGA. I don\'t want to get into fine PCB
design rules, so 1.0 mm ball pitch is my preference. The only
devices I can find that fit on the board have 196 or 256 pins.
But the real problem is availability.

Digikey has a few of the XC7S15-1FTGB196I and more a scheduled
for delivery in April. Add in the various speed and temperature
flavors trickling in (mostly in April) and I should be ok for the
initial delivery in August... if I can get my hands on those. I
don\'t know if Digikey factors in the backlog orders in these
counts.

Mouser shows great inventory of Efinix parts, particularly the
T13 and T20 in a 0.8 mm 256 pin BGA, 10s of thousands in stock.
But I\'d rather work with a 1.0 mm BGA. Oddly enough, LCSC shows
part numbers, but zero inventory.

Anyone work with 0.8 mm BGAs? What PWB feature dimensions did
you use? Did this impact the PWB cost?

It\'s difficult to know the impact without knowing details of the
board you have at the moment. It is also somewhat dependent on the
layout of the balls on the part - some BGA\'s have missing balls, or
their central balls all connected to ground to make layout easier.

In general, 0.8 mm pitch should be doable in four layers, but you
might need finer tracks and clearances than you used before. In a
recent board we did, the 0.8 mm pitch BGA was fine with four layers
and normal vias. We did not need to switch to 6 layers with
advanced vias until we moved to the 0.65 mm version.

What are \"normal\" vias and track/space widths? All of the BGAs I\'m
looking at are solid arrays. I\'d give you identifiers of the
packages, but I\'ve never found standard terms. Everyone has unique
identifiers and more than one! Packaging is insane! I especially
love how TI adds it to their data sheets with the reel packing and
boxing data ahead of the package mechanical data. I\'m sure that
stuff is important to someone, but they could have a separate data
sheet with all that. Some companies do that with the package
mechanical data. Gotta love diversity... or not.

My board is already 6 layers because of the high density and narrow
width, 23 mm x 115 mm. One layer is pretty much nothing but
lengthwise routing and it\'s still very congested. Surface layers are
hard to route on. Between the parts and the vias, there\'s no room
left. In the new layout, a couple of larger chips are going bye bye,
so maybe that will improve things a bit. But they are asking for
some new features that will add some chips, so it may be an even
swap.

By \"normal\" vias I mean through-hole, without plugging or tenting - the
cheap ones. And by \"normal\" track widths and spacing, I mean the
perhaps 6 mil - cheap sizes that any pcb manufacturer can make without
extra charge. Basically, at 0.8 mm BGA we didn\'t need to do anything
special or consider it an especially high-density card. It was still an
effort to route, as there are a lot of connections in a small area. But
it didn\'t need extra cost for the pcb.

So you don\'t know what size ball pad, via pad/drill and trace/space you actually used? I\'m finding it a bit harder than I expected to figure out dimensions that work. I\'d love to use the Xilinx 196 ball, 1.0 mm pitch part, but they are still pretty hard to get. Efinix has nothing but smaller pitch BGAs. 256 ball, 0.8 mm pitch, or 169 ball, 0.65 mm pitch. 0.65 mm pitch just won\'t work for me.

If you are already congested at 6 layers on the board, then you will
might have to go for beyond that. The first step is to talk to your pcb
and board manufacturers about via-in-pad, using tented or plugged vias.
If you can put the vias on the pads themselves, without causing voids
or blowouts during soldering, you can save a /lot/ of space. Next step
beyond that is microvias from outer layers to layers 2 and 5. We had to
do that for the 0.65 mm package BGA.

Absolutely not interested in 0.65 mm pitch, or any of the high density techniques like via in pad.

I can get away with nearly no vias, if I can route two traces between pins. That works at 1.0 mm, but at 0.8 mm, the trace/space has to be pretty small.

JLCPCB lists 0.2 mm drill (8 mil) and 0.45 mm via pads (18 mil). Assuming BGA lands of 0.4 mm means you can route 2 traces between pads only with 0.08 mm (3.2 mil) trace/space. Routing between the via holes is also tight, with trace/space of 0.08 mm (3.2 mil) and 0.13 mm (5.1 mil) (according to Xilinx). I\'ve read elsewhere that the space to the via drill should be 0.2 mm (8 mil) minimum for lower cost boards, so this doesn\'t sound good for routing two traces. With that restriction, I could only route two rows of pins without vias. Adding vias to escape the BGA makes the entire section of the board a difficult for routing.

Yeah, I\'d much prefer to go with a 1.0 mm part like the Xilinx 196 ball Spartan 7 parts. I just don\'t know if I\'ll be able to get them.

--

Rick C.

-- Get 1,000 miles of free Supercharging
-- Tesla referral code - https://ts.la/richard11209

 

[David Brown]

On 09/01/2023 11:42, gnuarm.del...@gmail.com wrote:

On Monday, January 9, 2023 at 4:02:43 AM UTC-5, David Brown wrote:
On 08/01/2023 20:01, gnuarm.del...@gmail.com wrote:
On Sunday, January 8, 2023 at 7:52:02 AM UTC-5, David Brown
wrote:
On 07/01/2023 18:49, gnuarm.del...@gmail.com wrote:
A small board with a 100QFP is being redesigned for a new
FPGA due to obsolescence. Gowin makes a 100QFP device that
would be a good fit, but my customer has said \"no\" to the
100% Chinese brand... US government customers, ya know!

So now I\'m looking at a BGA. I don\'t want to get into fine
PCB design rules, so 1.0 mm ball pitch is my preference. The
only devices I can find that fit on the board have 196 or 256
pins. But the real problem is availability.

Digikey has a few of the XC7S15-1FTGB196I and more a
scheduled for delivery in April. Add in the various speed and
temperature flavors trickling in (mostly in April) and I
should be ok for the initial delivery in August... if I can
get my hands on those. I don\'t know if Digikey factors in the
backlog orders in these counts.

Mouser shows great inventory of Efinix parts, particularly
the T13 and T20 in a 0.8 mm 256 pin BGA, 10s of thousands in
stock. But I\'d rather work with a 1.0 mm BGA. Oddly enough,
LCSC shows part numbers, but zero inventory.

Anyone work with 0.8 mm BGAs? What PWB feature dimensions
did you use? Did this impact the PWB cost?

It\'s difficult to know the impact without knowing details of
the board you have at the moment. It is also somewhat dependent
on the layout of the balls on the part - some BGA\'s have
missing balls, or their central balls all connected to ground
to make layout easier.

In general, 0.8 mm pitch should be doable in four layers, but
you might need finer tracks and clearances than you used
before. In a recent board we did, the 0.8 mm pitch BGA was fine
with four layers and normal vias. We did not need to switch to
6 layers with advanced vias until we moved to the 0.65 mm
version.

What are \"normal\" vias and track/space widths? All of the BGAs
I\'m looking at are solid arrays. I\'d give you identifiers of the
packages, but I\'ve never found standard terms. Everyone has
unique identifiers and more than one! Packaging is insane! I
especially love how TI adds it to their data sheets with the reel
packing and boxing data ahead of the package mechanical data. I\'m
sure that stuff is important to someone, but they could have a
separate data sheet with all that. Some companies do that with
the package mechanical data. Gotta love diversity... or not.

My board is already 6 layers because of the high density and
narrow width, 23 mm x 115 mm. One layer is pretty much nothing
but lengthwise routing and it\'s still very congested. Surface
layers are hard to route on. Between the parts and the vias,
there\'s no room left. In the new layout, a couple of larger chips
are going bye bye, so maybe that will improve things a bit. But
they are asking for some new features that will add some chips,
so it may be an even swap.

By \"normal\" vias I mean through-hole, without plugging or tenting -
the cheap ones. And by \"normal\" track widths and spacing, I mean
the perhaps 6 mil - cheap sizes that any pcb manufacturer can make
without extra charge. Basically, at 0.8 mm BGA we didn\'t need to do
anything special or consider it an especially high-density card. It
was still an effort to route, as there are a lot of connections in
a small area. But it didn\'t need extra cost for the pcb.

So you don\'t know what size ball pad, via pad/drill and trace/space
you actually used? I\'m finding it a bit harder than I expected to
figure out dimensions that work. I\'d love to use the Xilinx 196
ball, 1.0 mm pitch part, but they are still pretty hard to get.
Efinix has nothing but smaller pitch BGAs. 256 ball, 0.8 mm pitch,
or 169 ball, 0.65 mm pitch. 0.65 mm pitch just won\'t work for me.

I\'m afraid I don\'t remember the sizes used - I was not directly involved
in the layout and routing. (I\'ve done fine-pitched BGA layout, but it\'s
probably 15 years since I did a pcb design myself.)

Is it the routing you see as a problem for 0.65 mm pitch, or the cost of
boards with high density features, or the production of them? We have
found that while the 0.65 mm pitch parts were harder for the layout and
a little more expensive for the boards, parts in these packages can be a
lot easier to get hold of. The choice of 0.65 mm or 0.8 mm was forced
by component availability, rather than as a preference by our layout
folk. (Our production people have no qualms about mounting small pitch
BGAs.)

If you are already congested at 6 layers on the board, then you
will might have to go for beyond that. The first step is to talk to
your pcb and board manufacturers about via-in-pad, using tented or
plugged vias. If you can put the vias on the pads themselves,
without causing voids or blowouts during soldering, you can save a
/lot/ of space. Next step beyond that is microvias from outer
layers to layers 2 and 5. We had to do that for the 0.65 mm package
BGA.

Absolutely not interested in 0.65 mm pitch, or any of the high
density techniques like via in pad.

I can get away with nearly no vias, if I can route two traces between
pins. That works at 1.0 mm, but at 0.8 mm, the trace/space has to be
pretty small.

JLCPCB lists 0.2 mm drill (8 mil) and 0.45 mm via pads (18 mil).
Assuming BGA lands of 0.4 mm means you can route 2 traces between
pads only with 0.08 mm (3.2 mil) trace/space. Routing between the
via holes is also tight, with trace/space of 0.08 mm (3.2 mil) and
0.13 mm (5.1 mil) (according to Xilinx). I\'ve read elsewhere that
the space to the via drill should be 0.2 mm (8 mil) minimum for lower
cost boards, so this doesn\'t sound good for routing two traces. With
that restriction, I could only route two rows of pins without vias.
Adding vias to escape the BGA makes the entire section of the board a
difficult for routing.

Mechanical drilling has a lot bigger tolerances than laser drilling, so
you do need to have extra space between the via hole and tracks on the
internal layers to account for the inaccuracies. Some manufacturers
will give you tighter specifications - in particular, some use lasers
for 0.2 mm holes. (And some, on the other hand, use mechanical drills
and charge extra for 0.2 mm holes due to extra breakage of the small
drill bits.)

Yeah, I\'d much prefer to go with a 1.0 mm part like the Xilinx 196
ball Spartan 7 parts. I just don\'t know if I\'ll be able to get
them.

That\'s always the big problem these days. I\'m afraid I can\'t give much
advice there (at least, nothing that you won\'t already have thought of
yourself) - we are all in the same boat.

 

[Theo]

gnuarm.del...@gmail.com <gnuarm.deletethisbit@gmail.com> wrote:

Anyone work with 0.8 mm BGAs? What PWB feature dimensions did you use?
Did this impact the PWB cost?

I\'ve not done this for real, but I did a bit of playing around with PCB
routing such BGA FPGAs. I was trying to see if it was feasible to use them
on cheap PCB processes with basic soldering (I never actually made any
boards to test it). One of the limitations of cheap processes is the
tolerances can be quite slack: eg 6mil track width/spacing, which makes
doing the BGA escapes hard.

An interesting (to me) observation was that you may be able to make PCB
routing easier by choosing your pinout wisely - eg don\'t route signals from
inner or adjacent balls where you don\'t need them. That means it wasn\'t
such a headache to have more pins because you can ignore many of them. For
some of them, it was fine to short adjacent pins to (safe) power rails if it
wasn\'t possible to separate them.

Another observation was that it might be fine to have a net route through
unused pins if they\'re all high impedance in the FPGA config. I\'d not do
this for fast signals, but maybe OK for slow/static ones.

Not ideal, but a couple of tricks where FPGAs offer a little bit more PCB
routing flexibility compared with off the shelf parts.

Theo

 

[gnuarm.del...@gmail.com]

On Monday, January 9, 2023 at 7:42:20 AM UTC-5, David Brown wrote:

On 09/01/2023 11:42, gnuarm.del...@gmail.com wrote:
On Monday, January 9, 2023 at 4:02:43 AM UTC-5, David Brown wrote:
On 08/01/2023 20:01, gnuarm.del...@gmail.com wrote:
On Sunday, January 8, 2023 at 7:52:02 AM UTC-5, David Brown
wrote:
On 07/01/2023 18:49, gnuarm.del...@gmail.com wrote:
A small board with a 100QFP is being redesigned for a new
FPGA due to obsolescence. Gowin makes a 100QFP device that
would be a good fit, but my customer has said \"no\" to the
100% Chinese brand... US government customers, ya know!

So now I\'m looking at a BGA. I don\'t want to get into fine
PCB design rules, so 1.0 mm ball pitch is my preference. The
only devices I can find that fit on the board have 196 or 256
pins. But the real problem is availability.

Digikey has a few of the XC7S15-1FTGB196I and more a
scheduled for delivery in April. Add in the various speed and
temperature flavors trickling in (mostly in April) and I
should be ok for the initial delivery in August... if I can
get my hands on those. I don\'t know if Digikey factors in the
backlog orders in these counts.

Mouser shows great inventory of Efinix parts, particularly
the T13 and T20 in a 0.8 mm 256 pin BGA, 10s of thousands in
stock. But I\'d rather work with a 1.0 mm BGA. Oddly enough,
LCSC shows part numbers, but zero inventory.

Anyone work with 0.8 mm BGAs? What PWB feature dimensions
did you use? Did this impact the PWB cost?

It\'s difficult to know the impact without knowing details of
the board you have at the moment. It is also somewhat dependent
on the layout of the balls on the part - some BGA\'s have
missing balls, or their central balls all connected to ground
to make layout easier.

In general, 0.8 mm pitch should be doable in four layers, but
you might need finer tracks and clearances than you used
before. In a recent board we did, the 0.8 mm pitch BGA was fine
with four layers and normal vias. We did not need to switch to
6 layers with advanced vias until we moved to the 0.65 mm
version.

What are \"normal\" vias and track/space widths? All of the BGAs
I\'m looking at are solid arrays. I\'d give you identifiers of the
packages, but I\'ve never found standard terms. Everyone has
unique identifiers and more than one! Packaging is insane! I
especially love how TI adds it to their data sheets with the reel
packing and boxing data ahead of the package mechanical data. I\'m
sure that stuff is important to someone, but they could have a
separate data sheet with all that. Some companies do that with
the package mechanical data. Gotta love diversity... or not.

My board is already 6 layers because of the high density and
narrow width, 23 mm x 115 mm. One layer is pretty much nothing
but lengthwise routing and it\'s still very congested. Surface
layers are hard to route on. Between the parts and the vias,
there\'s no room left. In the new layout, a couple of larger chips
are going bye bye, so maybe that will improve things a bit. But
they are asking for some new features that will add some chips,
so it may be an even swap.

By \"normal\" vias I mean through-hole, without plugging or tenting -
the cheap ones. And by \"normal\" track widths and spacing, I mean
the perhaps 6 mil - cheap sizes that any pcb manufacturer can make
without extra charge. Basically, at 0.8 mm BGA we didn\'t need to do
anything special or consider it an especially high-density card. It
was still an effort to route, as there are a lot of connections in
a small area. But it didn\'t need extra cost for the pcb.

So you don\'t know what size ball pad, via pad/drill and trace/space
you actually used? I\'m finding it a bit harder than I expected to
figure out dimensions that work. I\'d love to use the Xilinx 196
ball, 1.0 mm pitch part, but they are still pretty hard to get.
Efinix has nothing but smaller pitch BGAs. 256 ball, 0.8 mm pitch,
or 169 ball, 0.65 mm pitch. 0.65 mm pitch just won\'t work for me.

I\'m afraid I don\'t remember the sizes used - I was not directly involved
in the layout and routing. (I\'ve done fine-pitched BGA layout, but it\'s
probably 15 years since I did a pcb design myself.)

Is it the routing you see as a problem for 0.65 mm pitch, or the cost of
boards with high density features, or the production of them? We have
found that while the 0.65 mm pitch parts were harder for the layout and
a little more expensive for the boards, parts in these packages can be a
lot easier to get hold of. The choice of 0.65 mm or 0.8 mm was forced
by component availability, rather than as a preference by our layout
folk. (Our production people have no qualms about mounting small pitch
BGAs.)

I\'m trying to be as conservative as possible. I\'d rather not use BGAs at all, but the only QFPs I can find that fit on the board are Gowin, which is not on the approved vendor list with my customer. They are too Chinese.

I\'m concerned about adding cost for the boards, cost for the assembly and just an easy road forward. I spend the last two years building 8,000 units when the CODEC factory burnt down. The customer knows about this issue, but the previous CM turned flaky on me and all but stopped delivering product..

I have a new CM, but I don\'t want to go through production problems again.

If you are already congested at 6 layers on the board, then you
will might have to go for beyond that. The first step is to talk to
your pcb and board manufacturers about via-in-pad, using tented or
plugged vias. If you can put the vias on the pads themselves,
without causing voids or blowouts during soldering, you can save a
/lot/ of space. Next step beyond that is microvias from outer
layers to layers 2 and 5. We had to do that for the 0.65 mm package
BGA.

Absolutely not interested in 0.65 mm pitch, or any of the high
density techniques like via in pad.

I can get away with nearly no vias, if I can route two traces between
pins. That works at 1.0 mm, but at 0.8 mm, the trace/space has to be
pretty small.

JLCPCB lists 0.2 mm drill (8 mil) and 0.45 mm via pads (18 mil).
Assuming BGA lands of 0.4 mm means you can route 2 traces between
pads only with 0.08 mm (3.2 mil) trace/space. Routing between the
via holes is also tight, with trace/space of 0.08 mm (3.2 mil) and
0.13 mm (5.1 mil) (according to Xilinx). I\'ve read elsewhere that
the space to the via drill should be 0.2 mm (8 mil) minimum for lower
cost boards, so this doesn\'t sound good for routing two traces. With
that restriction, I could only route two rows of pins without vias.
Adding vias to escape the BGA makes the entire section of the board a
difficult for routing.

Mechanical drilling has a lot bigger tolerances than laser drilling, so
you do need to have extra space between the via hole and tracks on the
internal layers to account for the inaccuracies. Some manufacturers
will give you tighter specifications - in particular, some use lasers
for 0.2 mm holes. (And some, on the other hand, use mechanical drills
and charge extra for 0.2 mm holes due to extra breakage of the small
drill bits.)

JLCPCB does 0.2 mm holes without extra charge, along with 0.45 mm via pads. They charge a bit extra for 0.4 mm pads. I guess it makes for a smaller target. I look at various PCB maker\'s pages to see what they state they can do.

Yeah, I\'d much prefer to go with a 1.0 mm part like the Xilinx 196
ball Spartan 7 parts. I just don\'t know if I\'ll be able to get
them.

That\'s always the big problem these days. I\'m afraid I can\'t give much
advice there (at least, nothing that you won\'t already have thought of
yourself) - we are all in the same boat.

Yeah, I need to get in touch with a real distributor, rather than the web guys. Someone who has a sales person who will work with me. I remember in 1999, when I started working for myself, the sales people were dying to get their foot in the door. They quoted me great prices! Now, not so much.

--

Rick C.

-+ Get 1,000 miles of free Supercharging
-+ Tesla referral code - https://ts.la/richard11209

 

[Stef]

On 2023-01-09 gnuarm.del...@gmail.com wrote in comp.arch.fpga:

I\'m trying to be as conservative as possible. I\'d rather not use BGAs at all, but the only QFPs I can find that fit on the board are Gowin, which is not on the approved vendor list with my customer. They are too Chinese.

Digikey has a number of FPGAs in QFP100/144 in stock. Efinix,
Microchip, Lattice, Xilinx. Nothing that suits your needs?

--
Stef

Kinkler\'s First Law:
Responsibility always exceeds authority.

Kinkler\'s Second Law:
All the easy problems have been solved.

 

[gnuarm.del...@gmail.com]

On Monday, January 9, 2023 at 7:30:34 PM UTC-5, Stef wrote:

On 2023-01-09 gnuarm.del...@gmail.com wrote in comp.arch.fpga:

I\'m trying to be as conservative as possible. I\'d rather not use BGAs at all, but the only QFPs I can find that fit on the board are Gowin, which is not on the approved vendor list with my customer. They are too Chinese.

Digikey has a number of FPGAs in QFP100/144 in stock. Efinix,
Microchip, Lattice, Xilinx. Nothing that suits your needs?

The QFP144 is far too large. Efinix has no QFP100 parts. Lattice has no QFP100 parts that I\'ve seen since the LFXP parts revision 2.0 of my board used. Xilinx has not had any in decades, unless you mean the very small, yet expensive CPLD thing they sell. Microchip might have some QFP100 parts in one of their older lines that I\'m not so familiar with. I believe their logic cells can be either logic, or FFs, but not both. So you need roughly double the count, if not more. They are very expensive too. I\'ve never been inclined to research such an old product, much like the Spartan 3. I get that in a QFP100, but how long will they continue to make Spartan 3 devices... which have also climbed significantly in price. It\'s what is called NRND. Gowin would have been perfect, but at one point they were put on a US list of CCMC (Communist Chinese Military Companies). Even though they were taken off, my customer sells a lot to the US government, so they don\'t like the \"optics\".

Did I miss any?

Trust me, I\'ve been traveling this road for the last eight years from when Lattice first announced the end of the line for the LFXP parts. Surprisingly, I noticed the other day that Arrow, who invested heavily in the line when Lattice announced EOL, still has almost 30,000 of them. In theory, I could continue to use those, but I do expect to sell more than 30,000 over the next 10 years. In fact, I am going to turn this over to a CM, who will in turn, pay me a royalty for every unit shipped. Other than consulting, I will be out of the business, but still receive payments for everything sold.. I won\'t be the guy worrying about where to find what part!

At this point, I\'d be happy with the Xilinx XC7S15-1FTGB196I. Digikey has over 800 in stock which is more than half of what I need. They claim they will have the rest by April. I just don\'t know how to get on the list to receive them. I guess the fact that they have inventory, means no one is in line to get the April lot. I really need to talk to someone who can give me straight answers about the future availability though. It would not help to design in a part that let\'s me ship the first order, then not receive any parts for the next year!

--

Rick C.

+- Get 1,000 miles of free Supercharging
+- Tesla referral code - https://ts.la/richard11209

 

[Stef]

On 2023-01-10 gnuarm.del...@gmail.com wrote in comp.arch.fpga:

On Monday, January 9, 2023 at 7:30:34 PM UTC-5, Stef wrote:
On 2023-01-09 gnuarm.del...@gmail.com wrote in comp.arch.fpga:

I\'m trying to be as conservative as possible. I\'d rather not use BGAs at all, but the only QFPs I can find that fit on the board are Gowin, which is not on the approved vendor list with my customer. They are too Chinese.

Digikey has a number of FPGAs in QFP100/144 in stock. Efinix,
Microchip, Lattice, Xilinx. Nothing that suits your needs?

The QFP144 is far too large. Efinix has no QFP100 parts. Lattice has no QFP100 parts that I\'ve seen since the LFXP parts revision 2.0 of my board used. Xilinx has not had any in decades, unless you mean the very small, yet expensive CPLD thing they sell. Microchip might have some QFP100 parts in one of their older lines that I\'m not so familiar with. I believe their logic cells can be either logic, or FFs, but not both. So you need roughly double the count, if not more. They are very expensive too. I\'ve never been inclined to research such an old product, much like the Spartan 3. I get that in a QFP100, but how long will they continue to make Spartan 3 devices... which have also climbed significantly in price. It\'s what is called NRND. Gowin would have been perfect, but at one point they were put on a US list of CCMC (Communist Chinese Military Companies). Even though they were taken off, my customer sells a lot to the US government, so they don\'t like the \"optics\".

Did I miss any?

Okay, QFP144 is too large, that severily limits the QFP options, but
Lattice does have a QFP100: ICE40HX1K-VQ100. But this one may not have
enough logic for you, its the smallest in the series.

--
Stef

Rube Walker: \"Hey, Yogi, what time is it?\"
Yogi Berra: \"You mean now?\"

 

[gnuarm.del...@gmail.com]

On Tuesday, January 10, 2023 at 3:52:54 AM UTC-5, Stef wrote:

On 2023-01-10 gnuarm.del...@gmail.com wrote in comp.arch.fpga:
On Monday, January 9, 2023 at 7:30:34 PM UTC-5, Stef wrote:
On 2023-01-09 gnuarm.del...@gmail.com wrote in comp.arch.fpga:

I\'m trying to be as conservative as possible. I\'d rather not use BGAs at all, but the only QFPs I can find that fit on the board are Gowin, which is not on the approved vendor list with my customer. They are too Chinese.

Digikey has a number of FPGAs in QFP100/144 in stock. Efinix,
Microchip, Lattice, Xilinx. Nothing that suits your needs?

The QFP144 is far too large. Efinix has no QFP100 parts. Lattice has no QFP100 parts that I\'ve seen since the LFXP parts revision 2.0 of my board used. Xilinx has not had any in decades, unless you mean the very small, yet expensive CPLD thing they sell. Microchip might have some QFP100 parts in one of their older lines that I\'m not so familiar with. I believe their logic cells can be either logic, or FFs, but not both. So you need roughly double the count, if not more. They are very expensive too. I\'ve never been inclined to research such an old product, much like the Spartan 3. I get that in a QFP100, but how long will they continue to make Spartan 3 devices... which have also climbed significantly in price. It\'s what is called NRND. Gowin would have been perfect, but at one point they were put on a US list of CCMC (Communist Chinese Military Companies). Even though they were taken off, my customer sells a lot to the US government, so they don\'t like the \"optics\".

Did I miss any?
Okay, QFP144 is too large, that severily limits the QFP options, but
Lattice does have a QFP100: ICE40HX1K-VQ100. But this one may not have
enough logic for you, its the smallest in the series.

No, that is too small. They also have a QFN84, which unfortunately is two rows, requiring finer than 0.1 mm trace/space. Again, only supporting the 1K size, and oddly enough, no support for the PLL.

The current design is 3 kLUTs at 90% utilization. I was surprised it fit. I don\'t want to squeeze a new design so tightly, and I need to add some logic, so I\'m shooting for at least a 5 kLUT part and prefer larger. The current design has uLaw and I\'d like to be able to add Alaw, not that it\'s a lot of logic, but it\'s something more than a couple of LUTs. Then there\'s the multipliers... the current design uses one multiply, done using LUTs. I\'ve been asked to provide gain controls, requiring two more multipliers, 16 x 8 minimum. This could be done by shift and add, so not horribly large. But built in multipliers would help with the size and the iCE40 line has none.

Oddly enough, many of the Lattice lines (even relatively modern ones) have no multipliers. The entire XO3 line has no multipliers. The MachXO3D has a couple of parts in an QFN72 with enough I/Os (barely), but not much availability. Hmmm... this may have improved. The MachXO3D-9400 has enough inventory in various configurations to do the initial order. But the I/O count on the QFN72 probably won\'t be enough with the added features. Ignoring that, the current inventory might get me through the current order, but 53 weeks for any further inventory. Digikey tells you more are coming in April. When you enter a number, it\'s always available in April, even a million..

--

Rick C.

++ Get 1,000 miles of free Supercharging
++ Tesla referral code - https://ts.la/richard11209

 

[jan Coombs]

On Tue, 10 Jan 2023 06:19:58 -0800 (PST)
\"gnuarm.del...@gmail.com\" <gnuarm.deletethisbit@gmail.com> wrote:

> But built in multipliers would help with the size and the iCE40 line has none.

Just four or eight multipliers, and many more on-chip goodies:

https://www.latticesemi.com/en/Products/FPGAandCPLD/iCE40UltraPlus

dev board
https://tinyvision.ai/collections/kits/products/upduino-v3-1

Jan Coombs
--

 

[Michael Schwingen]

On 2023-01-09, gnuarm.del...@gmail.com <gnuarm.deletethisbit@gmail.com> wrote:

I\'m concerned about adding cost for the boards, cost for the assembly and
just an easy road forward. I spend the last two years building 8,000
units when the CODEC factory burnt down. The customer knows about this
issue, but the previous CM turned flaky on me and all but stopped
delivering product.

I have a new CM, but I don\'t want to go through production problems again.

0.8mm BGA should be no problem for any reputable CM - fine-pitch QFP is
usually more trouble.

JLCPCB does 0.2 mm holes without extra charge, along with 0.45 mm via
pads. They charge a bit extra for 0.4 mm pads. I guess it makes for a
smaller target. I look at various PCB maker\'s pages to see what they
state they can do.

That should be enough to fit a via between the 0.8mm-BGA pads - that\'s what
we do regulary. If you want blocking caps underneath the BGA, you will
probably require plugged/plated vias. You will have to look at the pinout
and do the fanout routing to see how many layers you need.

Talk to your PCB manufacturer about the details before doing the final
layout - there is some fine tuning (eg. drill size, annular ring, spacing)
where different PCB manufacturers have different preferences regarding which
rules will yield good results - when doing do, 0.8mm BGA should be possible
at modest PCB costs.

cu
Michael
--
Some people have no respect of age unless it is bottled.

 

[Michael Schwingen]

On 2023-01-10, gnuarm.del...@gmail.com <gnuarm.deletethisbit@gmail.com> wrote:

The QFP144 is far too large. Efinix has no QFP100 parts. Lattice has no
QFP100 parts that I\'ve seen since the LFXP parts revision 2.0 of my board
used.

We have used Lattice MachXO2 in TQFP100 in the past - not sure if these fit
your needs.

> At this point, I\'d be happy with the Xilinx XC7S15-1FTGB196I.

Okay, this sounds like the MXO2 might be two sizes to small for you.

cu
Michael
--
Some people have no respect of age unless it is bottled.

 

[gnuarm.del...@gmail.com]

On Tuesday, January 10, 2023 at 12:33:54 PM UTC-5, jan Coombs wrote:

On Tue, 10 Jan 2023 06:19:58 -0800 (PST)
\"gnuarm.del...@gmail.com\" <gnuarm.del...@gmail.com> wrote:

But built in multipliers would help with the size and the iCE40 line has none.
Just four or eight multipliers, and many more on-chip goodies:

https://www.latticesemi.com/en/Products/FPGAandCPLD/iCE40UltraPlus

dev board
https://tinyvision.ai/collections/kits/products/upduino-v3-1

Yes, sorry, they do make a few multiplier chips with FPGA tiles. I was referring to parts that I might be able to use. They have a couple of 8 kLUT parts, only one in a package that I could use. I can pick between a 0.8 mm ball pitch, or 0.65 mm. Not really excited about either, even though there\'s a bit of inventory of the 256 ball, 0.8 mm part. But no insight into future deliveries.

This looking for usable parts gets old fast, and I\'ve been doing it for over a year now. When I find the guy responsible for this shortage, I\'m going to give him a piece of my mind!

--

Rick C.

--- Get 1,000 miles of free Supercharging
--- Tesla referral code - https://ts.la/richard11209

 

[gnuarm.del...@gmail.com]

On Tuesday, January 10, 2023 at 1:35:24 PM UTC-5, Michael Schwingen wrote:

On 2023-01-09, gnuarm.del...@gmail.com <gnuarm.del...@gmail.com> wrote:

I\'m concerned about adding cost for the boards, cost for the assembly and
just an easy road forward. I spend the last two years building 8,000
units when the CODEC factory burnt down. The customer knows about this
issue, but the previous CM turned flaky on me and all but stopped
delivering product.

I have a new CM, but I don\'t want to go through production problems again.
0.8mm BGA should be no problem for any reputable CM - fine-pitch QFP is
usually more trouble.

Part of my problem is a lack of having designed with BGAs before. I can find footprint recommendations, but they are different for every manufacturer.. It didn\'t occur to me that this might be because even though they have the same pitch and ball count, they may not have the same ball size.

The two primary choices right now are a 196 ball, 1.0 mm pitch and 256 ball, 0.8 mm pitch. Can you share the design rules you used for these parts?

JLCPCB does 0.2 mm holes without extra charge, along with 0.45 mm via
pads. They charge a bit extra for 0.4 mm pads. I guess it makes for a
smaller target. I look at various PCB maker\'s pages to see what they
state they can do.
That should be enough to fit a via between the 0.8mm-BGA pads - that\'s what
we do regulary. If you want blocking caps underneath the BGA, you will
probably require plugged/plated vias. You will have to look at the pinout
and do the fanout routing to see how many layers you need.

Talk to your PCB manufacturer about the details before doing the final
layout - there is some fine tuning (eg. drill size, annular ring, spacing)
where different PCB manufacturers have different preferences regarding which
rules will yield good results - when doing do, 0.8mm BGA should be possible
at modest PCB costs.

You mean my CM who orders the PWBs? Yeah, I\'ve tried asking before and they say they would need a design so they could get a quote. I know, that sounds lame, but I used four different CMs over the last decade and they have all said the same thing. They don\'t have design rules, that\'s for me to know.

--

Rick C.

--+ Get 1,000 miles of free Supercharging
--+ Tesla referral code - https://ts.la/richard11209

 

[gnuarm.del...@gmail.com]

On Tuesday, January 10, 2023 at 1:39:43 PM UTC-5, Michael Schwingen wrote:

On 2023-01-10, gnuarm.del...@gmail.com <gnuarm.del...@gmail.com> wrote:

The QFP144 is far too large. Efinix has no QFP100 parts. Lattice has no
QFP100 parts that I\'ve seen since the LFXP parts revision 2.0 of my board
used.
We have used Lattice MachXO2 in TQFP100 in the past - not sure if these fit
your needs.
At this point, I\'d be happy with the Xilinx XC7S15-1FTGB196I.
Okay, this sounds like the MXO2 might be two sizes to small for you.

I have the Lattice selection guide, the Lattice package guide, Lattice data sheets and have marked all the packages I might use on all the parts I might use.

No, the MXO2 parts are pretty limited for this. The existing design is shoehorned into the 3 kLUT XP part. I was surprised it routed. I have to add some logic and I don\'t have the same confidence in routing, so I want plenty of room in this part. 5 kLUTs should do the job, but I\'d like to have a bit more. The Xilinx part has 6000 LCs, but there is no such thing as a LC in a Xilinx part, that\'s a marketing term. The XC7S6 has 3,752 6LUTs and twice that many FFs, so it should be about like an 8 kLUT device, but who knows if the routing is up to it. The only one in stock is the XC7S15, so that\'s the one I would use, at least to start. But I need a better price than the $25, and some expectation of getting more parts.

Anyone see signs of the shortages easing?

--

Rick C.

-+- Get 1,000 miles of free Supercharging
-+- Tesla referral code - https://ts.la/richard11209

 

[David Brown]

On 10/01/2023 23:17, gnuarm.del...@gmail.com wrote:

Yes, sorry, they do make a few multiplier chips with FPGA tiles. I
was referring to parts that I might be able to use. They have a
couple of 8 kLUT parts, only one in a package that I could use. I
can pick between a 0.8 mm ball pitch, or 0.65 mm. Not really excited
about either, even though there\'s a bit of inventory of the 256 ball,
0.8 mm part. But no insight into future deliveries.

This looking for usable parts gets old fast, and I\'ve been doing it
for over a year now. When I find the guy responsible for this
shortage, I\'m going to give him a piece of my mind!

The reason you can parts in high-density packages, but not low-density
packages, is that there are lots of people such as yourself who are so
reluctant to use the small pitch devices. (This is not criticism - you
have solid reasons for preferring larger pitch devices, as do many
others.) Big manufacturers often prefer smaller pitch and higher
density, as it can lead to lower overall costs for their products, even
if design is more costly and the pcbs are more expensive.

There have been component supply issues for several years now, with only
gradual improvement in many areas. But there is a general pattern of
somewhat higher availability in smaller pitch parts.

 

[David Brown]

On 10/01/2023 23:44, gnuarm.del...@gmail.com wrote:

On Tuesday, January 10, 2023 at 1:35:24 PM UTC-5, Michael Schwingen
wrote:
On 2023-01-09, gnuarm.del...@gmail.com <gnuarm.del...@gmail.com
wrote:

I\'m concerned about adding cost for the boards, cost for the
assembly and just an easy road forward. I spend the last two
years building 8,000 units when the CODEC factory burnt down. The
customer knows about this issue, but the previous CM turned flaky
on me and all but stopped delivering product.

I have a new CM, but I don\'t want to go through production
problems again.
0.8mm BGA should be no problem for any reputable CM - fine-pitch
QFP is usually more trouble.

Part of my problem is a lack of having designed with BGAs before. I
can find footprint recommendations, but they are different for every
manufacturer. It didn\'t occur to me that this might be because even
though they have the same pitch and ball count, they may not have the
same ball size.

The two primary choices right now are a 196 ball, 1.0 mm pitch and
256 ball, 0.8 mm pitch. Can you share the design rules you used for
these parts?

The board stackup, routing and bypassing recommendations from FPGA
manufacturers are basically bollocks. I believe it is primarily a
matter of being able to fob off complaints and support requests by
saying \"Did you follow our layout application notes, impossible though
they may be? If not, it\'s not /our/ fault that you have problems.\"

OK, that\'s a bit of an exaggeration, but you can ignore the suggestions
of 16 layers with 8 different power planes and a dozen different
capacitor sizes mounted directly below the device.


Yes, there are complications for BGA layouts. And I\'m afraid you are
going to have to do some research, some learning, and some discussions
with both PCB manufacturers (or their proxies) and board builders.

For the same pitch of BGA, there can be different sized balls, and
different sized pads on the underside of the BGA device which will
affect the shape of the ball after soldering. Pad size on the pcb has
different options. You have a key decision between solder mask defined
and non-solder mask defined pads, which affects mechanical strength,
thermal stability, solder paste masks, routeability, and manufacturing
requirements. And BGA soldering has different requirements in
production than non-BGA devices.


I have no doubt that this is something you can master quite quickly -
it\'s not /that/ hard. But it\'s not something you can learn just by a
thread on a newsgroup.