DGnd and AGnd?

[Ken Smith]

In article <AF57c.11433$rw6.217810@news.xtra.co.nz>,
Terry Given <the_domes@xtra.co.nz> wrote:

"Ken Smith" <kensmith@violet.rahul.net> wrote in message
At 10.16, the document implies that surface mount decoupling capacitors
are better than through hole parts. In practice correctly installed
through hole capacitors with a 0.1" lead spacing can be better than
surface mount. The lead provides a very low impedance path from the
actual capacitance directly to the power and ground planes.

I presume this is due to the via somehow? many leaded caps are actually smt
caps with leads attached to the metallised end caps. please elaborate

Its the trace that leads to the via, not the via its self. Standard
layout practice is to not put a via hole in a component pad but to instead
run a trace to the via. By time you include the length of the two traces
involved and plus the size of the body of the part, the total length is
longer than the 0.1" cap would have.

the biggest problem to overcome with leaded caps is their physical sixe.
trace the physical current loop....with an 0603 smt capacitor sitting 0.4mm
above the 0V plane (4-layer PCB) the physical loop can be no smaller than
(about) 1.5mm x 0.4mm.

I disagree with the 1.5mm number here. You left the length of the traces
out of the calculation.

= 2.5mm, the area WILL be bigger than that of an smt cap. Often leaded caps
have 0.1" leads bent out to 0.2" spacing,

Yes, don't use those they don't work well for bypassing. You want the
ones with the 0.1" spaced leads.

it is clear from the above discussion that traces etc. only make the problem
worse. keep them wide, and use multiple parallel vias.

If you are working with SMT you are limited as to trace size running to a
pad. This limit is imposed by the need to not wick all the heat away from
one part of the pad too quickly.

under its body. Some of the bypass capacitors should be on the back of
the PCB to make the impedance between the Vcc and DGND low.

if your production manager doesnt hit you with a stick.

I can run faster than him. Not that I've had to yet.

--
--
kensmith@rahul.net forging knowledge

 

[Terry Given]

"Ken Smith" <kensmith@violet.rahul.net> wrote in message
news:c3j2a8$ofh$2@blue.rahul.net...

In article <AF57c.11433$rw6.217810@news.xtra.co.nz>,
Terry Given <the_domes@xtra.co.nz> wrote:
"Ken Smith" <kensmith@violet.rahul.net> wrote in message
At 10.16, the document implies that surface mount decoupling capacitors
are better than through hole parts. In practice correctly installed
through hole capacitors with a 0.1" lead spacing can be better than
surface mount. The lead provides a very low impedance path from the
actual capacitance directly to the power and ground planes.

I presume this is due to the via somehow? many leaded caps are actually
smt
caps with leads attached to the metallised end caps. please elaborate

Its the trace that leads to the via, not the via its self. Standard
layout practice is to not put a via hole in a component pad but to instead
run a trace to the via. By time you include the length of the two traces
involved and plus the size of the body of the part, the total length is
longer than the 0.1" cap would have.

yep. With smt, normally need 25mils minimum track length to stop solder
paste wicking down holes (some would say use really small holes in the pads
but that is $$$)

the biggest problem to overcome with leaded caps is their physical sixe.
trace the physical current loop....with an 0603 smt capacitor sitting
0.4mm
above the 0V plane (4-layer PCB) the physical loop can be no smaller than
(about) 1.5mm x 0.4mm.

I disagree with the 1.5mm number here. You left the length of the traces
out of the calculation.

yep, deliberately - how far can we "push the envelope" well the device
itself cant physically get any smaller hence my use of it as a "lower bound"
for the sorts of stray inductance we see.

OK, taking into account the pad-hole constraint and using 0.1" leaded caps
the race gets a lot closer.

= 2.5mm, the area WILL be bigger than that of an smt cap. Often leaded
caps
have 0.1" leads bent out to 0.2" spacing,

Yes, don't use those they don't work well for bypassing. You want the
ones with the 0.1" spaced leads.

until production runs out, then they get the other ones (it always seems to
work out that way) and bend the legs to fit them into the smaller
lead-pitch, often leaving nice big long legs *sigh*

it is clear from the above discussion that traces etc. only make the
problem
worse. keep them wide, and use multiple parallel vias.

If you are working with SMT you are limited as to trace size running to a
pad. This limit is imposed by the need to not wick all the heat away from
one part of the pad too quickly.

Not much of an issue with decent soldering machines nowadays - look at
planar smps, they invariably have ludicrous amounts of Cu (> 8Oz, many
layers) and regularly drop smt parts straight onto solid planes.

Arguments about hand-soldering dont count either, because using a soldering
iron is the single worst thing you can do to smt components - differing CTE
of the metallisation on the endcaps and the substrate material mean gross
thermal shock causes cracking, which can lead to imminent failure or (worse)
gradual degradation, followed by failure in the field. This is whay all
manufacturers specify a reflow soldering temperature profile.

I am sure lots of people can provide examples of hand-soldering that
"worked," but a much more reliable approach is to use hot air rework if you
have to, and real machines to assemble the stuff. This is especially
important if say you build an expensive wodgy that might fail
catastrophically if something untoward happened. For the same reason I would
NEVER prototype anything important using a non-thru-hole plated pcb - far
too much time spent tracking stupid "not properly connected" faults.

under its body. Some of the bypass capacitors should be on the back of
the PCB to make the impedance between the Vcc and DGND low.

if your production manager doesnt hit you with a stick.

I can run faster than him. Not that I've had to yet.

always a handy skill

--
--
kensmith@rahul.net forging knowledge

Cheers
Terry

 

[Ken Smith]

In article <G1c7c.11756$rw6.222162@news.xtra.co.nz>,
Terry Given <the_domes@xtra.co.nz> wrote:

"Ken Smith" <kensmith@violet.rahul.net> wrote in message
[.. smt vs thru hole bypasses ...]
I disagree with the 1.5mm number here. You left the length of the traces
out of the calculation.

yep, deliberately - how far can we "push the envelope" well the device
itself cant physically get any smaller hence my use of it as a "lower bound"
for the sorts of stray inductance we see.

There is a technology where the bypass capacitors are actually imbedded
into holes in the PCB. If you want to have zero trace length on the SMT
devices, you have to compare to that version of thru hole to be fair.

[....]

until production runs out, then they get the other ones (it always seems to
work out that way) and bend the legs to fit them into the smaller
lead-pitch, often leaving nice big long legs *sigh*

I think I've met those guys.

Arguments about hand-soldering dont count either, because using a soldering
iron is the single worst thing you can do to smt components

You haven't been watching those guys mensioned above closely enough if you
think this is the worst thing you can do to smt part. I've seen
runs of boards with 100% failure rate due to failed SMT film caps.

"worked," but a much more reliable approach is to use hot air rework if you
have to,

I've assembled a smallish proto PCB this way. Zephertronics(sp) makes an
ok preheater.


--
--
kensmith@rahul.net forging knowledge

 

[Max Hauser]

"Terry Given" in news:AF57c.11433$rw6.217810@news.xtra.co.nz...

. . .
....a 1mm long via with an 0.016" hole has about 1.2nH of inductance.
By the time this becomes important, skin effect ensures the solid
component leg is no better than a tube of wall thickness
66mm/sqrt(F Hz) . . .

T. Given's detailed helpful response implicitly touches an ancient Usenet
issue, maybe worth incorporating into RFC1855 or other tutorials. Bear with
me for a moment. Classic newbie behavior is to pop onto unfamiliar
newsgroups and ask questions without looking to see past answers archived.
Some groups have FAQ lists to address these questions. (The current case
may not have been such a newbie question but it touched on Skin Depth, bear
with me.) New questioners could thus miss Good Stuff like in 1987 when a
feud broke out between warring camps, one insisting that skin-effect
conduction depth was the same as electromagnetic wavelength and the other
answering stubbornly that it was instead the _acoustical_ wavelength in air.
(Of course it's neither.) THAT is what an FAQ list can point to, so eager
newbies get the wisdom of the ages of Usenet.

(That exchange is at least fragmentarily archived on Major Public Archives,
including a boring posting news:1887@ucbcad.berkeley.edu, 23 September 1987,
on "skin depth," giving basics and references and therefore damping the
fun.)

 

[John Woodgate]

I read in sci.electronics.design that Max Hauser <maxREMOVE@THIStdl.com>
wrote (in <105rriiae9n0hdc@corp.supernews.com&gt about 'Skin effect /skin
depth FAQ (Was: DGnd and AGnd?)', on Sun, 21 Mar 2004:

Classic newbie behavior is to pop onto
unfamiliar newsgroups and ask questions without looking to see past
answers archived. Some groups have FAQ lists to address these questions.

I think we probably have to accept that as a fact of life. It's
extremely difficult to search an FAQ for something that may not be there
anyway, when you don't have a totally unique key-word to search on.
Especially since many FAQs are totally unstructured; they 'just grow'
because no-one can spare the time to be an FAQ editor.

Furthermore, if you are a newbie (of even if not) you may well not know
whether there is an FAQ, and if there is, how you get to it.

Of course, it would be good if people asked, 'I want to find out about
skin effect. Where is the FAQ?'.
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

 

[Terry Given]

"Ken Smith" <kensmith@violet.rahul.net> wrote in message
news:c3ko8d$sf3$1@blue.rahul.net...

In article <G1c7c.11756$rw6.222162@news.xtra.co.nz>,
Terry Given <the_domes@xtra.co.nz> wrote:
"Ken Smith" <kensmith@violet.rahul.net> wrote in message
[.. smt vs thru hole bypasses ...]
I disagree with the 1.5mm number here. You left the length of the
traces
out of the calculation.

yep, deliberately - how far can we "push the envelope" well the device
itself cant physically get any smaller hence my use of it as a "lower
bound"
for the sorts of stray inductance we see.

There is a technology where the bypass capacitors are actually imbedded
into holes in the PCB. If you want to have zero trace length on the SMT
devices, you have to compare to that version of thru hole to be fair.

$$$$ looks pretty neat though

[....]
until production runs out, then they get the other ones (it always seems
to
work out that way) and bend the legs to fit them into the smaller
lead-pitch, often leaving nice big long legs *sigh*


I think I've met those guys.

Arguments about hand-soldering dont count either, because using a
soldering
iron is the single worst thing you can do to smt components

You haven't been watching those guys mensioned above closely enough if you
think this is the worst thing you can do to smt part. I've seen
runs of boards with 100% failure rate due to failed SMT film caps.

our production guys once built a pcb insertion jig as part of the assembly
process for a remote display for an AC motor drive. In a fit of inspired
brilliance, they used nylon pins to press down on the pcb at multiple points
to clip it into the housing. In a fit of inspired stupidity, they chose one
of those points to be a 1206 resistor. Funnily enough the "OK" LED didnt
work in every single one of the first 100 or so units made, even though they
passed the pcb test....LOL

then there was the production-led project to smt a gatedrive assembly (R&D
was too busy, and prod was making about 20,000 of these a year) thereby
saving abouyt $3 per pcb. The prod techs did a nice job of the layout, and
all worked well, BUT the prod mgr decided to V-groove the panels as it was
cheaper than routing (saving about $0.10 per pcb). Then the guillotine
device they built to cut the individual pcbs out of the assembled panel bent
the pcb's so far all the smt caps disappeared, leaving only their end-caps
behind. The really funny thing is, because of the destruction left by these
vandalised PCBs, production dropped the project like a hot potato, when all
they needed to do was re-layout the panel (1-2hrs) for routing, and throw
away their automatic smt cap remover - the cost increase of routing (vs
v-groove) was tiny compared to the savings gained by smt vs hand assembly
*sigh*

"worked," but a much more reliable approach is to use hot air rework if
you
have to,

I've assembled a smallish proto PCB this way. Zephertronics(sp) makes an
ok preheater.

my favourite technique involves Other People doing the shitty jobs......

Cheers from Aotearoa/New Zealand
Terry

 

[Terry Given]

"John Woodgate" <jmw@jmwa.demon.contraspam.yuk> wrote in message
news:3cJYh1GtcfXAFw1W@jmwa.demon.co.uk...

I read in sci.electronics.design that Max Hauser <maxREMOVE@THIStdl.com
wrote (in <105rriiae9n0hdc@corp.supernews.com&gt about 'Skin effect /skin
depth FAQ (Was: DGnd and AGnd?)', on Sun, 21 Mar 2004:

Classic newbie behavior is to pop onto
unfamiliar newsgroups and ask questions without looking to see past
answers archived. Some groups have FAQ lists to address these questions.

I think we probably have to accept that as a fact of life. It's
extremely difficult to search an FAQ for something that may not be there
anyway, when you don't have a totally unique key-word to search on.
Especially since many FAQs are totally unstructured; they 'just grow'
because no-one can spare the time to be an FAQ editor.

FAQ that for a job - too much like hard work (sorry, couldnt resist the
pun - must be the lure of the Dark side...)

Furthermore, if you are a newbie (of even if not) you may well not know
whether there is an FAQ, and if there is, how you get to it.

I fall into this category, mostly because I dont care enough about my PC to
learn that sort of thing the hard way (PC is just a tool, and a particularly
unreliable one at that)

Of course, it would be good if people asked, 'I want to find out about
skin effect. Where is the FAQ?'.

NSL. Worse, they tend to use inexplicable (and unexplained) TLA's

My favourite has to be those who mis-use engineering jargon - I used to work
with a self-taught guy like this, and it was infuriating - if you dont know
what hysterisis is/means, dont use the word

 

[Max Hauser]

"John Woodgate" in news:3cJYh1GtcfXAFw1W@jmwa.demon.co.uk...

I read in sci.electronics.design that Max Hauser
wrote (in <105rriiae9n0hdc@corp.supernews.com&gt :

Classic newbie behavior is to pop onto
unfamiliar newsgroups and ask questions without looking to see past
answers archived. Some groups have FAQ lists to address these questions.

I think we probably have to accept that as a fact of life.

Yes I think that's pretty well established, steadily since the dawn of
newsgroups (at UNC-CH) 25 years ago.

It's extremely difficult to search an FAQ for something that
may not be there anyway, ...

Yes indeed indeed but that was not so much my point (actually the sh+t you
may find even if you DO ask was my point) but still, the issue with classic
newbie behavior is not about difficulty or ease of succeeding but about lack
of trying. Good information may be archived or it may not but you'd hardly
know one way or the other if you won't lift a finger to find out, eh?

 

[Mike]

On Sun, 21 Mar 2004 11:34:43 -0800, Max Hauser wrote:

"Terry Given" in news:AF57c.11433$rw6.217810@news.xtra.co.nz...
. . .
....a 1mm long via with an 0.016" hole has about 1.2nH of inductance.
By the time this becomes important, skin effect ensures the solid
component leg is no better than a tube of wall thickness
66mm/sqrt(F Hz) . . .

T. Given's detailed helpful response implicitly touches an ancient Usenet
issue, maybe worth incorporating into RFC1855 or other tutorials. Bear with
me for a moment. Classic newbie behavior is to pop onto unfamiliar
newsgroups and ask questions without looking to see past answers archived.
Some groups have FAQ lists to address these questions. (The current case
may not have been such a newbie question but it touched on Skin Depth, bear
with me.) New questioners could thus miss Good Stuff like in 1987 when a
feud broke out between warring camps, one insisting that skin-effect
conduction depth was the same as electromagnetic wavelength and the other
answering stubbornly that it was instead the _acoustical_ wavelength in air.
(Of course it's neither.) THAT is what an FAQ list can point to, so eager
newbies get the wisdom of the ages of Usenet.

(That exchange is at least fragmentarily archived on Major Public Archives,
including a boring posting news:1887@ucbcad.berkeley.edu, 23 September 1987,
on "skin depth," giving basics and references and therefore damping the
fun.)

I suppose you could argue that a FAQ would at least be centrally located
and subject to review by the newsgroup participants, but these days it
would be difficult to replace google as the FAQ of choice. Maybe there
should be a SED-FAQ toolbar for IE? No other reference provides so much
information (with such varying levels of quality) so easily.

Except, of course, posting a query in the newsgroups...

-- Mike --

 

[Roy McCammon]

Is there a question about skin effect?

--
local optimization seldom leads to global optimization

my e-mail address is: <my first name> <my last name> AT mmm DOT com