Searching \ for '[PIC]: Decoupling App Note' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/microchip/devices.htm?key=pic
Search entire site for: 'Decoupling App Note'.

Exact match. Not showing close matches.
PICList Thread
'[PIC]: Decoupling App Note'
2001\01\05@163605 by Kenneth Godee

flavicon
face
I read an application note about a month ago that was quite
good on Microchips web site concerning decoupling capacitors.
Now I'm trying to find it again. Their search engine is the pits!
Does anyone have or recall the number to this app note?


The Perfect Image Graphics Co.
2429 W. 12th St.
Tempe, AZ 85281

800-533-8732

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email spam_OUTlistservTakeThisOuTspammitvma.mit.edu with SET PICList DIGEST in the body


2001\01\05@170445 by David VanHorn

flavicon
face
At 02:24 PM 1/5/01 -0700, Kenneth Godee wrote:
>I read an application note about a month ago that was quite
>good on Microchips web site concerning decoupling capacitors.
>Now I'm trying to find it again. Their search engine is the pits!
>Does anyone have or recall the number to this app note?

Don't worry, be happy.

Put the cap at the ground pin. Run VCC from the chip to the cap on a wide
track, and from the cap to the system VCC on a narrow track.
Return your xtal caps to the chip's ground pin, by an isolated track.

The cap size is relative to the frequency, but you need to know where you
have the problem, before you know what size cap. Generally, 0.1 has least
impedance at 3 MHz, 0.01 at 30, and 0.001 at 300. Other values are in
between as you would expect. This holds true on most common caps, SMD and
thruhole. (I was surprised too)




--
Where's dave? http://www.findu.com/cgi-bin/find.cgi?kc6ete-9

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email .....listservKILLspamspam@spam@mitvma.mit.edu with SET PICList DIGEST in the body


2001\01\07@051019 by Russell McMahon

picon face
>At 02:24 PM 1/5/01 -0700, Kenneth Godee wrote:
>>I read an application note about a month ago that was quite
>>good on Microchips web site concerning decoupling capacitors.
>>Now I'm trying to find it again. Their search engine is the pits!
>>Does anyone have or recall the number to this app note?
>
>Don't worry, be happy.
>
>Put the cap at the ground pin. Run VCC from the chip to the cap on a wide
>track, and from the cap to the system VCC on a narrow track.
>Return your xtal caps to the chip's ground pin, by an isolated track.
>
>The cap size is relative to the frequency, but you need to know where you
>have the problem, before you know what size cap. Generally, 0.1 has least
>impedance at 3 MHz, 0.01 at 30, and 0.001 at 300. Other values are in
>between as you would expect. This holds true on most common caps, SMD and
>thruhole. (I was surprised too)


Others please critique these comments as required and add other useful
suggestions ...

All the above, plus -

Scatter caps of size as specced above around board - one per IC plus perhaps
a few others at meetings of the power supply.

Tantalum cap of around 10 uF per board helps but these are EXCELLENT fuses
which fail short circuit if you power supply has spikes above their rated
voltage so only use where you are SURE their rated voltage will NEVER be
exceeded.
If in doubt use SOLID aluminum in the same role.

If you are laying out a 2 layer PCB and are keen, run ground and earth
tracks parallel on opposite sides of the board.

Never have more than one path to the system ground point from any earth
point (ie no closed loops in the earth network). Same for Vcc but perhaps
less critical.

TRY to run earth network as a "tree" with the trunk base at the system earth
point and large branches feeding small branches feeding smaller etc.

Keep analog and digital grounds separate - ground run from each should start
at system ground point and proceed out to point of use with no commonality
BUT be aware that where an IC has mixed analog and digital inputs which are
related in some manner (eg micro + A2D or comparator or reference or some
combination) the ground circuit voltage drop may affect the device operation
and it MAY be better to common the digital and analog grounds for the chip
nearby. Each circuit with compromise ICs needs to be designed to optimise
the compromise aspect which is most important to you.

For super critical IC's consider using several caps of differing sizes in
parallel.

Keep cap leads as short as possible to minimise lead inductance.
Exception - leads can be cut to length required to generate a specific
series resonant frequency if main frequency of interest is known BUT this is
a rather expert and/or desperate approach.

Read regulator spec sheet carefully. Input and output capacitor requirements
can vary quite markedly and may affect stability (regulator can oscillate at
several MHz while providing apparently clean power! Some regulators set
specs on both maximum AND minimum equivalent series resistance (ESR). Too
"good" and ESR can cause problems.

Electrolytic  Cap lifetimes halve for every 10 degrees C increase in
operating temperature - beware positioning near eg hot power resistors. Ecap
lifetimes drop even more drastically when stored hot with DC power removed
!!!
Ecaps die quicker when run markedly BELOW their rated voltage.

Tantalum capacitors are great fun - they can be persuaded to smell really
terrible, smoke, shriek, flame and explode (sometimes all of these one after
the other) when exposed to reverse polarity or voltage spikes over their
rated voltage. They usually fail short circuit protecting other components
but making trouble shooting "interesting". Use them in these circumstances
to amuse your customers.



regards


       Russell McMahon

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\01\07@123830 by Morgan Olsson

picon face
Russell McMahon wrote:
>Ecaps die quicker when run markedly BELOW their rated voltage.

I have heard differing opinions about that, but most say electrilytic caps live longer when operating below rated voltage, epecially if operating temperature is high.


Any links or explanations to why they would break at lower voltage?
What about zero voltage?  I have some 30 year old never electrified caps and I checked they are still OK...

I heard an guru once say that lagre electrolytics should be powered up slowly first time, and then be held at nominal voltage.  As i understood it, it would be because it could heal eventual cracks in the aluminium oxide.

If it is because of cracks in the oxide, I find it more reasonable to believe they wil heal more safely at lower voltage, but it is more risk of violent destructive reactions at higher voltage.

/Morgan

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\01\07@125258 by Dan Michaels

flavicon
face
Morgan Olsson wrote:
>
>Any links or explanations to why they would break at lower voltage?
>What about zero voltage?  I have some 30 year old never electrified caps
and I checked they are still OK...
>

Probably NOT a good idea to use 30 YO components, "especially" caps, in
anything today. Back in the olden days [before I was born, BTW], the
floors of TV repair shops were littered with bad caps - died almost
as often as did vacuum tubes.
============


>I heard an guru once say that lagre electrolytics should be powered up
slowly first time, and then be held at nominal voltage.  As i understood it,
it would be because it could heal eventual cracks in the aluminium oxide.
>
>If it is because of cracks in the oxide, I find it more reasonable to
believe they wil heal more safely at lower voltage, but it is more risk of
violent destructive reactions at higher voltage.
>

The oxide layer does have a certain ability to regenerate itself.
You can probably find info on this via web search. I would
imagine this assumes the electrolyte hasn't totally dried out,
however.

- danM

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\01\08@000651 by Russell McMahon

picon face
From: Morgan Olsson <morgans.rtspamKILLspamTELIA.COM>


>Russell McMahon wrote:
>>Ecaps die quicker when run markedly BELOW their rated voltage.
>
>I have heard differing opinions about that, but most say electrilytic caps
live longer when operating below rated voltage, epecially if operating
temperature is high.
>Any links or explanations to why they would break at lower voltage?
>What about zero voltage?  I have some 30 year old never electrified caps
and I checked they are still OK...


A "wet" aluminium (actually vaguely damp) electrolytic capacitor separates
its two aluminium plates by a VERY thin layer of oxide purposefully built on
one electrode. The thickness depends on (and establishes) the rated voltage.
AFAIU running below rated voltage tends to de-form the layer.

Running a cap at zero voltage is the same as having it off - if you do this
at elevated temperature it will indeed die sooner tyhan if it is on as the
applied voltage maintains the layer in the presence of temperature. I am not
acquainted with the fine detailo of the chemistry but a look at
manufacturers' spec sheets re storage at high temperatures will confirm the
above.

>I heard an guru once say that lagre electrolytics should be powered up
slowly first time, and then be held at nominal voltage.  As i understood it,
it would be because it could heal eventual cracks in the aluminium oxide.


Especially true of "good old daze" caps. The oxide layer tends to dissapear
with time and is 'reformed" by the application of lower voltage (actually
full voltage with a resistor to limit current to a very small value) for a
period. This does not seem to be a problem with modern caps of substantial
size but again, I don't know what chemical magic they have worked to make it
so.


Russell

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\01\08@020222 by Roman Black

flavicon
face
Morgan Olsson wrote:

> What about zero voltage?  I have some 30 year old never electrified caps and I checked they are still OK...

I have some too, they test ok for capacitance
but the ESR gets shot due to drying out of
the electrolyte. The larger sized electros
from that era were very well made and sealed
though. :o)
-Roman

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\01\08@051700 by Tim Forcer

flavicon
face
Some thoughts on this often-ignored topic.

Without decoupling, it is common for a PIC'84 to reset itself when several
outputs change state simultaneously - the current spike involved in the
latter trips the internal "brownout" detector.  I suspect that this
behaviour is common to many (most? all?) PICs.  Some years back we added
specific requirements for decoupling to our lab notes for PIC experiments -
without decoupling it was almost unknown for breadboard systems to work
reliably.

It is possible to have too BIG a decoupling capacitor.  Check out things
like self-inductance and frequency of minimum impedance (the object is, in
some respects, low impedance rather than just capacitance).

Always route connections from the decoupler direct to the power pins of the
IC being decoupled (if using a multi-layer board with power/ground planes,
the decoupler should still have direct connections independent of the
power/ground planes).  The object is to decouple the *IC*, not the power
tracks (indeed, the old system of having "railway line" power tracks
running along under the rows of ICs on a board, with decouplers between
ICs, tends to CREATE problems rather than reduce them (power current spike
causing voltage glitch on one IC is coupled to two other ICs, rather than
decoupled).  Power tracks should be decoupled separately.

Surface mount decouplers are great.  They are mechanically small, and
providing short direct tracks is relatively simple.  Also they tend to have
less self-inductance than leaded types.

Decoupled IC sockets are great, and often have little or no cost penalty
over providing the two separately (also make life easier when laying out
the board).  The type with surface-mount capacitors also provide a mini
ground plane under the IC, which makes a small contribution to improving EMC.

If you want to experiment with undecoupled behaviour which is relatively
easy to observe and play around with, get hold of a TTL "555" timer chip.
The output totem pole circuit is particularly poor for "feed through"
current during the changes between LO and HI output, creating enormous
current glitches.

At one time, AMD produced an excellent application note about decoupling,
with specific reference to their range of PLDs.  This is no longer on the
Lattice site (AMD's programmable logic business was restyled "Vantis",
then, after a couple of years, sold to Lattice) - anybody got a current URL?

[Digression - the failure of tantalums due to over-voltage stress should
not be used for over-voltage protection, IMO.  In my experience, over-volt
damage can occur VERY fast, and requires close-tolerance components to
clamp any spikes.  The voltage at which a tantalum will fail is NOT
close-tolerance.  As in-system-programmability becomes ever common,
designers need to be aware that IC absolute maximum voltage rating states a
voltage above which a device will self-destruct, not simply fail to
operate.  ICs programmed using an external supervoltage will generally be
usable after suffering an over-voltage (although sometimes will need
re-programming).  isp types generally turn into multi-pin zero-Ohm devices.]

[Further digression: the ground connection from a PIC's oscillator
component(s), whether resonator or crystal, should be routed direct to the
PIC's ground pin rather than to the generic ground net.  This ensures that
oscillator currents have a good local return path.]

Tim Forcer               .....tmfKILLspamspam.....ecs.soton.ac.uk
Department of Electronics & Computer Science
The University of Southampton, UK

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\01\08@103207 by David VanHorn

flavicon
face
>
>Decoupled IC sockets are great, and often have little or no cost penalty
>over providing the two separately (also make life easier when laying out
>the board).  The type with surface-mount capacitors also provide a mini
>ground plane under the IC, which makes a small contribution to improving EMC.

They're better than nothing.  Notice the long lead between the ground pin
and the cap? Extra inductance you don't need.  Also, since the leads form a
"T" rather than a "V" with the power tracks, the decoupling is much less
effective.


>[Digression - the failure of tantalums due to over-voltage stress should
>not be used for over-voltage protection, IMO.

They also tend to blow if inrush current isn't controlled.  20 years ago,
they were great bypass elements. Today, I think electrolytics have eaten
their lunch.


>[Further digression: the ground connection from a PIC's oscillator
>component(s), whether resonator or crystal, should be routed direct to the
>PIC's ground pin rather than to the generic ground net.  This ensures that
>oscillator currents have a good local return path.]

A very often missed point!

--
Where's dave? http://www.findu.com/cgi-bin/find.cgi?kc6ete-9

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


More... (looser matching)
- Last day of these posts
- In 2001 , 2002 only
- Today
- New search...