Maker Pro
Maker Pro

Future: 0603 versus 0402 parts

J

Joerg

Jan 1, 1970
0
Ok, have to decide whether to default to 0603 or 0402 for resistors and
stuff on a new design. Wow, this time I am not size constrained.

Looking at Digikey it comes up with 143 pages (has risen) for 0402
resistors and 218 pages for 0603. Looking at prices 0603 still has a leg
up, around $0.003 for a 10K while its 0402 counterpart runs about twice
that. Is 0603 going to be a good choice for the next years? Or better
stick to 0402?

Of course from a debug point of view I hate to deal with 0402. With
chips the situation seems to be more clear cut. Things definitely seem
to be going TSSOP there.
 
J

John Larkin

Jan 1, 1970
0
Ok, have to decide whether to default to 0603 or 0402 for resistors and
stuff on a new design. Wow, this time I am not size constrained.

Looking at Digikey it comes up with 143 pages (has risen) for 0402
resistors and 218 pages for 0603. Looking at prices 0603 still has a leg
up, around $0.003 for a 10K while its 0402 counterpart runs about twice
that. Is 0603 going to be a good choice for the next years? Or better
stick to 0402?

Of course from a debug point of view I hate to deal with 0402. With
chips the situation seems to be more clear cut. Things definitely seem
to be going TSSOP there.

My production people hate 0402's... they tombstone a lot. We still use
0805's when there's plenty of room, 0603's for tight stuff, occasional
0402's for picosecond stuff where it matters.

You can still buy lots of 1206's and 2010's and such. They're not
going away.

Looks like some parts are appearing only in those drecky
leadless/chipscale packages. Yuk.

John
 
D

dalai lamah

Jan 1, 1970
0
Un bel giorno Joerg digitò:
Of course from a debug point of view I hate to deal with 0402.

I have to say that it's easier than I thought. They are still relatively
easy to hand solder; of course you need a microscope, but I wasn't able to
solder 0603 without it either. The problem with 0402 is the production: a
lot of assembliers still have problems with very close 0603 components,
with 0402 it will be even worse.
With
chips the situation seems to be more clear cut. Things definitely seem
to be going TSSOP there.

I think that very soon BGA (and hidden pins in general) will be the default
for all the ICs, not only the hundreds-pinned. You can already see this
trend, there are already a lot of logic ICs - even with few pins - with BGA
and/or flipchip package options. This worries me a lot more than discrete
components; 0402 can still be handled, but BGA means that you have to
change completely your procedures (design, PCB routing, prototyping,
testing...).
 
J

Joerg

Jan 1, 1970
0
John said:
My production people hate 0402's... they tombstone a lot. We still use
0805's when there's plenty of room, 0603's for tight stuff, occasional
0402's for picosecond stuff where it matters.

Hmm, I never had tombstoning with those. OTOH nearly all my clients
contract out circuit board stuffing so they don't have to bother with
that. Sometimes even the whole production. Sometimes I have a chat with
the contract producers and they'd tell me if they weren't happy with
parts of a design. Like when zeners began to miss data sheet limits I
had an instant email from China about it.

You can still buy lots of 1206's and 2010's and such. They're not
going away.

Good to know, thanks.

Looks like some parts are appearing only in those drecky
leadless/chipscale packages. Yuk.

I really hate it when there is a cooling pad that must be soldered, on
parts that consume less than 20mW. Arrgh. I have begun to shun parts
that don't come in TSSOP. Too much trouble. And when they stop the
migration path for a part at the SOIC level that is a red flag in terms
of remaining product life.
 
J

Joerg

Jan 1, 1970
0
dalai said:
Un bel giorno Joerg digitò:




I have to say that it's easier than I thought. They are still relatively
easy to hand solder; of course you need a microscope, but I wasn't able to
solder 0603 without it either. The problem with 0402 is the production: a
lot of assembliers still have problems with very close 0603 components,
with 0402 it will be even worse.

I use 3x glasses and do it sans microscope. Unless the client has a
camera-monitor setup.
I think that very soon BGA (and hidden pins in general) will be the default
for all the ICs, not only the hundreds-pinned. You can already see this
trend, there are already a lot of logic ICs - even with few pins - with BGA
and/or flipchip package options. This worries me a lot more than discrete
components; 0402 can still be handled, but BGA means that you have to
change completely your procedures (design, PCB routing, prototyping,
testing...).

Hmm, haven't seen that trend yet. Most of my stuff (analog and logic) is
migrating towards TSSOP.
 
J

Joerg

Jan 1, 1970
0
Eeyore said:
Joerg wrote:




Why exactly ?

Concerns about production life. Same back in the days when SO came out.
Stuff that wasn't migrated in due course had a tendency to become
unobtanium ;-)
 
B

Brian

Jan 1, 1970
0
Ok, have to decide whether to default to 0603 or 0402 for resistors and
stuff on a new design. Wow, this time I am not size constrained.

Looking at Digikey it comes up with 143 pages (has risen) for 0402
resistors and 218 pages for 0603. Looking at prices 0603 still has a leg
up, around $0.003 for a 10K while its 0402 counterpart runs about twice
that. Is 0603 going to be a good choice for the next years? Or better
stick to 0402?

Of course from a debug point of view I hate to deal with 0402. With
chips the situation seems to be more clear cut. Things definitely seem
to be going TSSOP there.

As a "board stuffer", I'll say a few things.

As you get smaller, assembly houses have to be better to do it. 0603
is a "knee" level for that right now. As the placement is needs to be
more accurate for smaller parts, more will be off one pad and
tombstone. However, with better equipment, the difference isn't as
bad.

Bigger parts are more robust in many ways. Thermal transfer,mechanical
strength, etc. Board contamination, water, etc., is less likely to
cause an issue with larger parts.

I say use the biggest parts you can fit in the required space, all the
way up to 1206 (the price is not that big a difference). There is no
real good reason to go smaller if you do not need the space. If your
cost of 0805 vs 0603 is killing your margins, you are better off
examining new markets or products. Yours are dead.

Yes, I know about high-speed and using smaller caps. But whenever I
here this, I see 0201's on a board running 20Mhz or less. No, you are
not improving anything then.
 
D

dalai lamah

Jan 1, 1970
0
Un bel giorno Joerg digitò:
I use 3x glasses and do it sans microscope.

And at the end of the day you can still find the way home? ;)
Hmm, haven't seen that trend yet. Most of my stuff (analog and logic) is
migrating towards TSSOP.

Same here, SOIC is becoming quite a rarity in my designs, almost everything
is TSSOP. But more and more frequently, when you open the datasheet of some
logic IC, you can find the microscopic BGA package options like VFBGA and
WCSP, or perhaps when you are lucky the (still annoying to hand solder)
QFN:

http://focus.ti.com/logic/docs/generalcontent.tsp?templateId=5985&navigationId=11373&contentId=4139
http://focus.ti.com/pdfs/logic/packcard7.pdf
 
J

Joerg

Jan 1, 1970
0
dalai said:
Un bel giorno Joerg digitò:




And at the end of the day you can still find the way home? ;)

Nope. Client brings me to the hotel. Then I only have to find the way to
a local pub :)))
Same here, SOIC is becoming quite a rarity in my designs, almost everything
is TSSOP. But more and more frequently, when you open the datasheet of some
logic IC, you can find the microscopic BGA package options like VFBGA and
WCSP, or perhaps when you are lucky the (still annoying to hand solder)
QFN:

http://focus.ti.com/logic/docs/generalcontent.tsp?templateId=5985&navigationId=11373&contentId=4139
http://focus.ti.com/pdfs/logic/packcard7.pdf

BGA is a pain. You can't see whether it's soldered right. Luckily all
the chips I need come in TSSOP. I was extremely glad when they even
ported the old CD4000 series to TSSOP.
 
J

Joerg

Jan 1, 1970
0
Brian said:
As a "board stuffer", I'll say a few things.

It's always good to hear from the folks who actually have to produce
what we design. You guys really know what works and what doesn't.

As you get smaller, assembly houses have to be better to do it. 0603
is a "knee" level for that right now. As the placement is needs to be
more accurate for smaller parts, more will be off one pad and
tombstone. However, with better equipment, the difference isn't as
bad.


Well, looks like I should stay with 0603 then. 0805 would become a bit
tight on this board but I shall see when the schematics are done.
Bigger parts are more robust in many ways. Thermal transfer,mechanical
strength, etc. Board contamination, water, etc., is less likely to
cause an issue with larger parts.

I say use the biggest parts you can fit in the required space, all the
way up to 1206 (the price is not that big a difference). There is no
real good reason to go smaller if you do not need the space. If your
cost of 0805 vs 0603 is killing your margins, you are better off
examining new markets or products. Yours are dead.

Nope. I have designed on the cutting edge in terms of cost for a couple
of decades now. Most others cannot understand how a human being can
possibly enjoy doing that. But I do. Then there are the devices that
just have to be small. Medical disposables, for example. On many of
those a SOT-23 would look like a boulder. I remember not being able to
use a TLV431 because it didn't come any smaller than SOT-23 at that time.

Yes, I know about high-speed and using smaller caps. But whenever I
here this, I see 0201's on a board running 20Mhz or less. No, you are
not improving anything then.

I promise I won't do that unless necessary :)

BTW, do you mind sharing the name or web site of your company? My
clients are always on the lookout for good SMT prototype assembly.
 
B

Brian

Jan 1, 1970
0
Spehro said:
On Tue, 13 Mar 2007 22:51:16 GMT, the renowned Joerg
I saw some 0402 tombstoning on a fairly recent batch of boards from a
China assembly facility (SnPb process). Zero problems with the 0805s
(no 0603s on that design).

One of the problems with a mixed design (where there are also some large
devices on the PCB) seems to be the paste. Too much and you get
tombstoning, too little and the big parts won't solder reliably. Like
those pellet stoves where they haven't figured out how to automate
intake air. Throttle too high and it'll go out on "1". Throttle too low
and it'll smoke on "3" or higher.


You may not like the price, though.

Actually a 1ohm 1206 costs roughly the same as a 10K in 0402, around
$0.007. A 2010 size can pop to several cents. Still, I don't see a
reason for anything larger than 0603 unless it is in a current sense
path or RF bridge.

[...]

--
Regards, Joerg

http://www.analogconsultants.com- Hide quoted text -

- Show quoted text -

Use the largest you have room for, even if it is a 1206. Smaller is
not "better" unless you need the room (yes, I know, small caps at high
speed, but whenever I hear that, its on a sub 12Mhz design). A better
machine is needed to place 0402 vs 0603, and those can be "limits" for
certian places, meaning even those sizes may be marginal on placement.
If you see alot of green on your board, why not use larger components
the human eye can see? Why make it harder, potentially less reliable
and dificult to repair?

If a few tenths of a cent is going to break your profit margin, your
market is too old and ready to die. Reevaluate your products and
markets instead of shaving pennies.
 
B

Brian

Jan 1, 1970
0
One of the problems with a mixed design (where there are also some large
devices on the PCB) seems to be the paste. Too much and you get
tombstoning, too little and the big parts won't solder reliably. Like
those pellet stoves where they haven't figured out how to automate
intake air. Throttle too high and it'll go out on "1". Throttle too low
and it'll smoke on "3" or higher.
Actually a 1ohm 1206 costs roughly the same as a 10K in 0402, around
$0.007. A 2010 size can pop to several cents. Still, I don't see a
reason for anything larger than 0603 unless it is in a current sense
path or RF bridge.

http://www.analogconsultants.com-Hide quoted text -
- Show quoted text -

Use the largest you have room for, even if it is a 1206. Smaller is
not "better" unless you need the room (yes, I know, small caps at high
speed, but whenever I hear that, its on a sub 12Mhz design). A better
machine is needed to place 0402 vs 0603, and those can be "limits" for
certian places, meaning even those sizes may be marginal on placement.
If you see alot of green on your board, why not use larger components
the human eye can see? Why make it harder, potentially less reliable
and dificult to repair?

If a few tenths of a cent is going to break your profit margin, your
market is too old and ready to die. Reevaluate your products and
markets instead of shaving pennies.- Hide quoted text -

- Show quoted text -

Sorry for the semi "double" post. Google groups sucks, but my ISP has
no newsserver :( Seemed to not go thru the first try, then shows up
later. Go figure.
 
D

DJ Delorie

Jan 1, 1970
0
dalai lamah said:
Un bel giorno Joerg digitò:
And at the end of the day you can still find the way home? ;)

I use 2x and 3.5x mag visors, and after a while with the 3.5x on, I
can't focus on regular things any more. It takes a few minutes to
re-adjust. So, I try to use the *weakest* lens that works for the
part sizes I'm working with.
Same here, SOIC is becoming quite a rarity in my designs, almost
everything is TSSOP.

My last design (still debugging it ;) is primarily 0603 and tssop,
with sot-323 transistors. The occasional ssop (0.65mm pitch) seems
big to me now. I had SOICs on my last board, but only because it was
a RAM array and I wanted to run traces between the pins without
horrendously expensive fab costs, and it was a 5v design which limited
chip selection.

But, I've done boards by hand with tvssop (0.4mm pitch) and 01005
parts. It can be done, and it puts the larger (hah) parts in
perspective.
 
S

Spehro Pefhany

Jan 1, 1970
0
Hmm, I never had tombstoning with those. OTOH nearly all my clients
contract out circuit board stuffing so they don't have to bother with
that. Sometimes even the whole production. Sometimes I have a chat with
the contract producers and they'd tell me if they weren't happy with
parts of a design. Like when zeners began to miss data sheet limits I
had an instant email from China about it.

I saw some 0402 tombstoning on a fairly recent batch of boards from a
China assembly facility (SnPb process). Zero problems with the 0805s
(no 0603s on that design).
Good to know, thanks.

You may not like the price, though.
I really hate it when there is a cooling pad that must be soldered, on
parts that consume less than 20mW. Arrgh. I have begun to shun parts
that don't come in TSSOP. Too much trouble. And when they stop the
migration path for a part at the SOIC level that is a red flag in terms
of remaining product life.




Best regards,
Spehro Pefhany
 
J

Joerg

Jan 1, 1970
0
Spehro said:
I saw some 0402 tombstoning on a fairly recent batch of boards from a
China assembly facility (SnPb process). Zero problems with the 0805s
(no 0603s on that design).

One of the problems with a mixed design (where there are also some large
devices on the PCB) seems to be the paste. Too much and you get
tombstoning, too little and the big parts won't solder reliably. Like
those pellet stoves where they haven't figured out how to automate
intake air. Throttle too high and it'll go out on "1". Throttle too low
and it'll smoke on "3" or higher.
You may not like the price, though.

Actually a 1ohm 1206 costs roughly the same as a 10K in 0402, around
$0.007. A 2010 size can pop to several cents. Still, I don't see a
reason for anything larger than 0603 unless it is in a current sense
path or RF bridge.

[...]
 
J

Joerg

Jan 1, 1970
0
Spehro said:
Well, if you're buying them one at a time, it's mostly overhead.

Suggest you get quotes in production quantities where appropriate.

That was for larger quantities.
 
J

Joerg

Jan 1, 1970
0
Hawker said:
I tend to use 0603 and only go 0402 if I have to.
There are a few reasons for this.
From an assembly standpoint, as another poster said 0402 is a knee in
the manufactures abilities. Also 0402 uses a different feeder for most
P&P machines than 1210-0603 so often CMs don't have as many feeders for
0402.

From an electrical standpoint for caps there has been some recent
research showing that in the smaller geometries (0402 and smaller) the
caps do not hold there rated capacitance as well when they have a DC
bias on them (as in used for bypass caps). I am still learning about
this but apparently the actual capacitance value can go way down under
DC bias for smaller geometries and this is not as prevalent for 0603 and
larger.

Interesting. 0402 caps in my case would only be 0.01uF and smaller
though, the rest would be mostly 0805. But as Brian said, it may be best
to keep it as large as practical. 0603 for resistors and 0805 for most
caps is certainly in the cards on this design, maybe even 0805 for
everything.
 
J

Joerg

Jan 1, 1970
0
Brian said:
Spehro Pefhany wrote:

@removethispacbell.net> wrote:
John Larkin wrote:
@removethispacbell.net> wrote:
Ok, have to decide whether to default to 0603 or 0402 for resistors and
stuff on a new design. Wow, this time I am not size constrained.
Looking at Digikey it comes up with 143 pages (has risen) for 0402
resistors and 218 pages for 0603. Looking at prices 0603 still has a leg
up, around $0.003 for a 10K while its 0402 counterpart runs about twice
that. Is 0603 going to be a good choice for the next years? Or better
stick to 0402?
Of course from a debug point of view I hate to deal with 0402. With
chips the situation seems to be more clear cut. Things definitely seem
to be going TSSOP there.
My production people hate 0402's... they tombstone a lot. We still use
0805's when there's plenty of room, 0603's for tight stuff, occasional
0402's for picosecond stuff where it matters.
Hmm, I never had tombstoning with those. OTOH nearly all my clients
contract out circuit board stuffing so they don't have to bother with
that. Sometimes even the whole production. Sometimes I have a chat with
the contract producers and they'd tell me if they weren't happy with
parts of a design. Like when zeners began to miss data sheet limits I
had an instant email from China about it.
I saw some 0402 tombstoning on a fairly recent batch of boards from a
China assembly facility (SnPb process). Zero problems with the 0805s
(no 0603s on that design).
One of the problems with a mixed design (where there are also some large
devices on the PCB) seems to be the paste. Too much and you get
tombstoning, too little and the big parts won't solder reliably. Like
those pellet stoves where they haven't figured out how to automate
intake air. Throttle too high and it'll go out on "1". Throttle too low
and it'll smoke on "3" or higher.
You can still buy lots of 1206's and 2010's and such. They're not
going away.
Good to know, thanks.
You may not like the price, though.
Actually a 1ohm 1206 costs roughly the same as a 10K in 0402, around
$0.007. A 2010 size can pop to several cents. Still, I don't see a
reason for anything larger than 0603 unless it is in a current sense
path or RF bridge.

http://www.analogconsultants.com-Hide quoted text -
- Show quoted text -

Use the largest you have room for, even if it is a 1206. Smaller is
not "better" unless you need the room (yes, I know, small caps at high
speed, but whenever I hear that, its on a sub 12Mhz design). A better
machine is needed to place 0402 vs 0603, and those can be "limits" for
certian places, meaning even those sizes may be marginal on placement.
If you see alot of green on your board, why not use larger components
the human eye can see? Why make it harder, potentially less reliable
and dificult to repair?

If a few tenths of a cent is going to break your profit margin, your
market is too old and ready to die. Reevaluate your products and
markets instead of shaving pennies.- Hide quoted text -

- Show quoted text -


Sorry for the semi "double" post. Google groups sucks, but my ISP has
no newsserver :( Seemed to not go thru the first try, then shows up
later. Go figure.

Yeah, lots of people have that issue. I feel blessed. My ISP changed
hands twice, PacificBell -> SBC -> AT&T, but they always kept a nice NG
access. Even binaries are allowed which is really helpful for schematics
and photos.
 
Top