Searching \ for '[OT] Decoupling capacitors' 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/index.htm?key=decoupling+capacitors
Search entire site for: 'Decoupling capacitors'.

Exact match. Not showing close matches.
PICList Thread
'[OT] Decoupling capacitors'
1999\01\13@014105 by Lynx {Glenn Jones}

flavicon
face
Sorry this is off topic, but i think that this can help many people who
arent experts at electronics, and it is related to pics, in a way.

Anyway, My questions are, what are decoupling capacitors needed for? What
value works best? Which chips should i make sure to decouple? and What are
signs that decoupling capacitors are needed?

In case its not clear, by decoupling capacitor, i mean a capacitor placed
across the Vcc and GND pins of IC's. THanks for any and all help.

------------------------------------------------------------------------------
A member of the PI-100 Club:
3.1415926535897932384626433832795028841971693993751
058209749445923078164062862089986280348253421170679

1999\01\13@030934 by James Cameron

flavicon
face
Lynx {Glenn Jones} wrote:
> Sorry this is off topic, but i think that this can help many people
> who arent experts at electronics, and it is related to pics, in a way.

Yes!  Some of my old electronics around the house predate my
understanding of the need for decoupling capacitors.  And CMOS input
pullups!  Every now and then something stops working and I have to
figure out where I went wrong ten years ago.  The latest failure was the
connection from an optocoupler after a motion detector into a CMOS input
that was read by the "alarm" system.  I had no pull-up on the input, the
optocoupler was simply connected directly to ground.  Being woken up at
night is a very good way to refine my designs.

> Anyway, My questions are, what are decoupling capacitors needed for?

They decouple something.  Like a train.  Disconnecting the circuit from
the induced noise from outside.  ;-) ;-)  [i.e. I don't know]

> What value works best?

0.1uF

> Which chips should i make sure to decouple?

Anything with a counter, flip-flop, or oscillator is my rule.

> What are signs that decoupling capacitors are needed?

Spontaneous resets.  Resets or unexpected behaviour when you stick a
transmitting mobile phone near the circuit.  Unexpected input data when
an output is driven.

--
James Cameron                                      (spam_OUTcameronTakeThisOuTspamstl.dec.com)

OpenVMS, Linux, Firewalls, Software Engineering, CGI, HTTP, X, C, FORTH,
COBOL, BASIC, DCL, csh, bash, ksh, sh, Electronics, Microcontrollers,
Disability Engineering, Netrek, Bicycles, Pedant, Farming, Home Control,
Remote Area Power, Greek Scholar, Tenor Vocalist, Church Sound, Husband.

"Specialisation is for insects." -- Robert Heinlein.

1999\01\13@091933 by Harrison Cooper

flavicon
face
part 0 73 bytes


Attachment converted: wonderland:pcb bypass.pdf (PDF /CARO) (000261D0)

1999\01\13@112735 by Lynx {Glenn Jones}

flavicon
face
Ok, I now have a better idea of what these are needed for. However, I read
that electrolytic capacitors are not good for decoupling because of their
high inductance. Is the same true for the Gold foil capacitors which have
very high values? I'm thinking of using a .1F capacitor as a board filter
(accross the power lines where they enter the board). Do you guys think
this will work very well? I also plan to put .1uF capacitors on each of
the MCU's, A/DC's and other such chips. Thanks alot for all the help.

------------------------------------------------------------------------------
A member of the PI-100 Club:
3.1415926535897932384626433832795028841971693993751
058209749445923078164062862089986280348253421170679

1999\01\13@121450 by Peter L. Peres
picon face
Hello,

On Wed, 13 Jan 1999, James Cameron wrote:

> > What are signs that decoupling capacitors are needed?
>
> Spontaneous resets.  Resets or unexpected behaviour when you stick a
> transmitting mobile phone near the circuit.  Unexpected input data when
> an output is driven.

Actually 'scoping the Vdd line with a high frequency scope properly
attached (RF class grounding - not so easy sometimes) is the only way to
find out. If there are high frequency circuits involved, with
non-repetitive waveforms (such as HC gates, other logic, and
microcontrollers) you'll need a DSO to make sense of what is happening.

Lacking this, the decoupling rules outlined in design manuals and data
sheets should be obeyed, until proven wrong.

hope this helps,

       Peter

1999\01\13@133723 by mwalsh

flavicon
face
I've always thought that decoupling caps were used to decouple
the switching noise between the power and ground pins on a
digital device.

In CMOS circuits there is a P-channel switch tied to Vdd and
an N-channel switch tied to Vss and the output is taken between
the two drains that are tied together.  When the circuit is in a
stable state only one of the switches is turned on and the current
through the device is basically a function of the load attached to
the output.  When the device switches states, there is a brief
period where both the P and N channel switches are turned on.
A current spike with an amplitude limited only by the Rds(on)
of the two switches results.

A decoupling cap placed near the device supplies this energy
and keeps most of the current spike out of the power and
ground planes.  Since this is a high frequency, the decoupling
caps need to have low inductance and a higher self-resonant
frequency than the noise they are filtering.  Typically ceramic
caps meet these needs at a reasonable price.  I tend to use
0.1 uF caps, but I've never seen problems with boards that
use 0.01 uF decoupling caps

Reasonably sized power and ground traces, if you aren't doing
multilayer boards with power and ground planes, are almost
as important as the decoupling caps.  I consulted on a project
several years ago that had a lot of reliability problems.  They
had used 10 mil traces for power and ground and neglected
to decouple all of the chips.  I was seeing between 1 and 2
volt noise spikes between ground at one corner of the board
and ground at the other corner of a 4x5 inch PCB.  The
noise margins on the inputs were essentially zero when they
were referenced to this kind of ground.

Mark Walsh


Lynx {Glenn Jones} wrote:

{Quote hidden}

1999\01\13@161014 by paulb

flavicon
face
Lynx {Glenn Jones} wrote:

> I read that electrolytic capacitors are not good for decoupling
> because of their high inductance.  Is the same true for the Gold foil
> capacitors which have very high values?

 It most certainly *is* true.

> I'm thinking of using a .1F capacitor as a board filter (across the
> power lines where they enter the board).  Do you guys think this will
> work very well?

 *Absolutely useless!*

> I also plan to put .1uF capacitors on each of the MCU's, A/DC's and
> other such chips.

 That much is standard practice, and is what you must do.

 The "gold" capacitors appear to be intended solely for memory backup,
where a few tens or huundred ohms has no effect.  They are an improved
version of older high-capacity devices based as I have read on charcoal
(activated carbon) with high surface area.  This is of course, highly
resistive material.  In addition, they have some leakage so may or may
not be better or worse than using Ni-Cds (though *those* are decidedly
messy!).

 My "useless" comment therefore refers specifically to these miniature
100mF capacitors.  A standard electrolytic of the same value *would*
have some merit (though whether it would really be needed in addition to
a proper regulated supply?) but is the size of a can of food (if you
get a really modern one at 5V, perhaps only a can of baby food!).

 I have seen in the auto shops or similar, 1 Farad capacitors for extra
decoupling of the battery in automotive Hi-Fi systems.  These are the
size of an 850g tin of vegetables with decorative gold plating on
various parts and *include* a power indicator LED!

 Insofar as I assume, they simply incorporate an array of standard
electrolytics, which may be rated for low ESR/ high ripple, they may do
something close to what they propose, though I have great suspicions of
anything offered for this market.
--
 Cheers,
       Paul B.

1999\01\14@110049 by John Payson

flavicon
face
|  I have seen in the auto shops or similar, 1 Farad capacitors for extra
|decoupling of the battery in automotive Hi-Fi systems.  These are the
|size of an 850g tin of vegetables with decorative gold plating on
|various parts and *include* a power indicator LED!

|  Insofar as I assume, they simply incorporate an array of standard
|electrolytics, which may be rated for low ESR/ high ripple, they may do
|something close to what they propose, though I have great suspicions of
|anything offered for this market.

One of my coworkers has a 1F cap on his car stero; I don't know
why he needs a stereo that loud, but when he has it cranked and
the headlights on, they dim in time with the music.

The cap looked pretty much like an ordinary screw terminal 'lytic
to me except for its size: it was about 12" high or so by 4" dia-
meter if I remember right.

As for ESR, there really wouldn't be much point to such a cap if
it had too high an ESR, though I would not be surprised if a cap
that size really tends to behave like:

+ ----+--R--+--R--+--R--+--R--+--R--.
     |     |     |     |     |     |
     C     C     C     C     C     C
     |     |     |     |     |     |
- ----+-----+-----+-----+-----+-----'

where the R's are non-trivial but most of the C's are to the left.
Probably easier than having a cap which behaves as a single RC.

1999\01\16@114432 by dave vanhorn

flavicon
face
At 08:26 AM 1/13/99 -0800, Lynx {Glenn Jones} wrote:
>Ok, I now have a better idea of what these are needed for. However, I read
>that electrolytic capacitors are not good for decoupling because of their
>high inductance. Is the same true for the Gold foil capacitors which have
>very high values? I'm thinking of using a .1F capacitor as a board filter
>(accross the power lines where they enter the board). Do you guys think
>this will work very well? I also plan to put .1uF capacitors on each of
>the MCU's, A/DC's and other such chips. Thanks alot for all the help.



You are about to discover parasitic inductance.
The gold caps (not gold foil, it's just a name) have very high resistance,
they will not supply high pulse currents. a 1000uF cap would do a better job.

Bypasses need to be sized for the operating frequency. Square waves have
hard edges, which means that the chip needs to have a lot of current
available on the third and fifth harmonics. Even surface mount caps have
significant inductance.

Figure the third harmonic of your clock. 0.1uF works well at 3 MHz, .01 at
30, but above this you really need to check the specific capacitors.

1999\01\16@122558 by dave vanhorn

flavicon
face
>  I have seen in the auto shops or similar, 1 Farad capacitors for extra
>decoupling of the battery in automotive Hi-Fi systems.  These are the
>size of an 850g tin of vegetables with decorative gold plating on
>various parts and *include* a power indicator LED!
>
>  Insofar as I assume, they simply incorporate an array of standard
>electrolytics, which may be rated for low ESR/ high ripple, they may do
>something close to what they propose, though I have great suspicions of
>anything offered for this market.
>--
>  Cheers,
>        Paul B.


One wonders why they don't put the C on the amplifier instead.
The input is 12V, which means that a 1F cap has 72j stored, but with a huge
inductance and resistance.
Inside the amps, they switch the DC input up to +/- 35V or better.
Adding the caps at that point, where 0.5*C*V^2 works out so much better,
would seem an obvious and large improvement. Much smaller C, much more "punch"
As a final injury, putting those huge Cs on the input means that their
current to the load is limited by the amp's switcher as well as their own
internal L and R.

Dumb, Dumb, Dumb....

1999\01\16@123651 by dave vanhorn

flavicon
face
>As for ESR, there really wouldn't be much point to such a cap if
>it had too high an ESR, though I would not be surprised if a cap
>that size really tends to behave like:
>
>+ ----+--R--+--R--+--R--+--R--+--R--.
>      |     |     |     |     |     |
>      C     C     C     C     C     C
>      |     |     |     |     |     |
>- ----+-----+-----+-----+-----+-----'

Put some small Ls in there too, and you've got the picture!

1999\01\16@215512 by paulb

flavicon
face
Dave VanHorn wrote:

> One wonders why they don't put the C on the amplifier instead.
> Dumb, Dumb, Dumb....

 Hey, look, the *important* part of my posting was the final line:
"though I have great suspicions of anything offered for this market".

 I don't think it really matters as long as, for Fair Trading Practice
purposes, they can be shown to have a 1 Farad capacity.  They are sold
to a particular market, the one that buys gold plated connectors (but
with what underneath?), Oxygen Free Copper, cables with specific "in"
and "out" ends etc. etc.  Need I say more?

 Well, yes, don't forget the Power LED!
--
 Cheers,
       Paul B.

1999\01\16@215723 by dave vanhorn

flavicon
face
At 01:52 PM 1/17/99 +1000, Paul B. Webster VK2BZC wrote:
>Dave VanHorn wrote:
>
>> One wonders why they don't put the C on the amplifier instead.
>> Dumb, Dumb, Dumb....
>
>  Hey, look, the *important* part of my posting was the final line:
>"though I have great suspicions of anything offered for this market".

I know, I was just expounding on the silliness of the approach.
Now back to greenlining my CDs :)

1999\01\18@140049 by John Payson

flavicon
face
|One wonders why they don't put the C on the amplifier instead.
|The input is 12V, which means that a 1F cap has 72j stored, but with a huge
|inductance and resistance.
|Inside the amps, they switch the DC input up to +/- 35V or better.

Although I don't know whether such designs are used in automotive
audio equipment, I read an article which was plugging an amplifier
design where the switching supply was modulated to stay a bit above
the desired audio output level (so that the output transistor would
only have to drop that few volts).  If a particular amplifier uses
such a design, placing a monster-sized cap on the (varying) high
voltage 'supply' could be a very bad thing.

Given that the headlights on my co-worker's car will flash in time
with the music when the stereo is cranked, it would seem that in
addition to the audio-frequency variations in the amp's current req-
uirement there are also sub-audio-frequency variations.  While any
ESR in the 'bypass' cap will translate into wasted energy, the low-
frequency variations in current demand should be within the range of
what the cap can help filter; the most significant concern would be
the cap's ability to withstand the relatively high ripple currents
involved.

Of course, the most logical solution to all these problems would be
to rotate the "volume" knob about 1/2 turn counterclockwise, but who
am I to suggest such things?

1999\01\18@140711 by dave vanhorn

flavicon
face
>Although I don't know whether such designs are used in automotive
>audio equipment, I read an article which was plugging an amplifier
>design where the switching supply was modulated to stay a bit above
>the desired audio output level (so that the output transistor would
>only have to drop that few volts).  If a particular amplifier uses
>such a design, placing a monster-sized cap on the (varying) high
>voltage 'supply' could be a very bad thing.

That's a carver thing, they've got some cute name for it. They reduce
dissipation that way.

>Given that the headlights on my co-worker's car will flash in time
>with the music when the stereo is cranked, it would seem that in
>addition to the audio-frequency variations in the amp's current req-
>uirement there are also sub-audio-frequency variations.  While any
>ESR in the 'bypass' cap will translate into wasted energy, the low-
>frequency variations in current demand should be within the range of
>what the cap can help filter; the most significant concern would be
>the cap's ability to withstand the relatively high ripple currents
>involved.

A small battery would do as well, probably better.

>Of course, the most logical solution to all these problems would be
>to rotate the "volume" knob about 1/2 turn counterclockwise, but who
>am I to suggest such things?

Exactly. After a certain point, it ceases being music, and becomes special
effects.

1999\01\18@203828 by Alan King

picon face
dave vanhorn wrote:

> >Of course, the most logical solution to all these problems would be
> >to rotate the "volume" knob about 1/2 turn counterclockwise, but who
> >am I to suggest such things?
>
> Exactly. After a certain point, it ceases being music, and becomes special
> effects.

 You're still thinking from your or my sonic viewpoint.
These people NEED 500+ watts to hear the music through
several years worth of hearing damage!  (Some of them could
use 50000 volts to the temples too..)
Alan

1999\01\19@151255 by John Payson

flavicon
face
|  I have seen in the auto shops or similar, 1 Farad capacitors for extra
|decoupling of the battery in automotive Hi-Fi systems.  These are the
|size of an 850g tin of vegetables with decorative gold plating on
|various parts and *include* a power indicator LED!

After some contemplation, I suspect that the power indicator LED
may be not a mere gimmick, but an important safety feature.  I'm
not sure what the ESR on those monster caps is, but it's probably
well under an ohm.  If someone were to remove such a cap from a
car without discharging it, it could probably do some pretty severe
damage if shorted.  If initially charged to 12 volts, it could but
over 100 joules of energy into a 1ohm resistor in less than a second.
While installed bleeder resistors often work well for high-voltage
capacitors, a bleeder that could quickly reduce the cap voltage to a
safe level would impose a significant load on the cap while it's in
use.

A power-on LED, though, with a resistor so it receives 20mA at 12
volts, would serve to alert anyone who might be working with the cap
that it's "live".  Although the LED would of course add to the loading
of the cap, the 20mA load is really pretty trivial (if the cap were
removed from the circuit, a 20mA load would cause it to drop by one
vold every 50 seconds.  As the voltage/current decline, the rate of
drop would be even slower.

More... (looser matching)
- Last day of these posts
- In 1999 , 2000 only
- Today
- New search...