Why so many caps?

A

alpha_uma

Guest
I've noticed that the number of electrolytic capacitors on a typical "P4
socket 478" motherboard have mushroomed from, say, the previous "P3 socket
370" generations of motherboards. Why are they using so many electrolytic
capacitors these days? Not only are those caps an eyesore, they have such
short life-times--especially those used in the voltage regulator circuits.

They are not very noticeable on a socket 370 m/b, and you only see them
occasionally on, say, the first generation of Pentium motherboards (socket
7). I don't think I've ever noticed electrolytic capacitors on pre-Pentium
generations of motherboards. As a matter of fact, let me run to my storage
cabinet, and take a look at my 486 and 386 relics: no, I don't see them.

Can someone enlighten me? Thanks
Al-U
 
I've noticed that the number of electrolytic capacitors on a typical "P4
socket 478" motherboard have mushroomed from, say, the previous "P3 socket
370" generations of motherboards. Why are they using so many electrolytic
capacitors these days? Not only are those caps an eyesore, they have such
short life-times--especially those used in the voltage regulator circuits.

They are not very noticeable on a socket 370 m/b, and you only see them
occasionally on, say, the first generation of Pentium motherboards (socket
7). I don't think I've ever noticed electrolytic capacitors on pre-Pentium
generations of motherboards. As a matter of fact, let me run to my storage
cabinet, and take a look at my 486 and 386 relics: no, I don't see them.

Can someone enlighten me? Thanks
Click to expand...
The increase in processor speed has brought with it an increase in the
total power dissipated by CPUs in consumer-grade systems. The
switch to higher-density silicon processes ("deep submicron") has led,
for several reasons, to a need to decrease the operating voltages of
the chips. Lower voltage, and higher power implies much higher
current... if I recall correctly a lot of today's motherboards are
throwing around many tens of amperes, with a great deal of short-term
variation in the power consumption.

With such rapid changes in power consumption, and such high load
currents, you _cannot_ depend on off-board capacitance or regulation
to keep voltages stable. There's just too much speed-of-light delay
between the CPU and the power supply, and too much parasitic
inductance in the PC-board traces.

Modern motherboards place one or more high-speed, high-current voltage
step-down switching regulators _very_ close to the CPU, to step the 5
or 12-volt rail voltages down to the high-amperage sub-2-volt supplies
needed by many CPUs. These regulators need hefty amounts of
capacitance on their input sides to remain stable, and some on the
outputs as well.

You can't depend solely on small (.1 or .01 uF) chip capacitors for
bypassing. The capacitance can interact with the inductance of the PC
board traces to create a high-Q resonant circuit, which leaves you
with _lousy_ bypassing at the resonant frequency. It's necessary to
add low-Q electolytics to the traces, every few inch or two, to
"swamp" the resonances and restore proper bypass effectiveness.

Basically, today's CPUs are doing much more work, and are demanding a
lot more of their power supplies.

--
Dave Platt <dplatt@radagast.org> AE6EO
Hosting the Jade Warrior home page: http://www.radagast.org/jade-warrior
I do _not_ wish to receive unsolicited commercial email, and I will
boycott any company which has the gall to send me such ads!
 
"Dave Platt" <dplatt@radagast.org> wrote in message
news:10ekchhiubgqb44@corp.supernews.com...
I've noticed that the number of electrolytic capacitors on a typical
"P4
socket 478" motherboard have mushroomed from, say, the previous "P3
socket
370" generations of motherboards. Why are they using so many
electrolytic
capacitors these days? Not only are those caps an eyesore, they have
such
short life-times--especially those used in the voltage regulator
circuits.

They are not very noticeable on a socket 370 m/b, and you only see them
occasionally on, say, the first generation of Pentium motherboards
(socket
7). I don't think I've ever noticed electrolytic capacitors on
pre-Pentium
generations of motherboards. As a matter of fact, let me run to my
storage
cabinet, and take a look at my 486 and 386 relics: no, I don't see
them.

Can someone enlighten me? Thanks

The increase in processor speed has brought with it an increase in the
total power dissipated by CPUs in consumer-grade systems. The
switch to higher-density silicon processes ("deep submicron") has led,
for several reasons, to a need to decrease the operating voltages of
the chips. Lower voltage, and higher power implies much higher
current... if I recall correctly a lot of today's motherboards are
throwing around many tens of amperes, with a great deal of short-term
variation in the power consumption.

With such rapid changes in power consumption, and such high load
currents, you _cannot_ depend on off-board capacitance or regulation
to keep voltages stable. There's just too much speed-of-light delay
between the CPU and the power supply, and too much parasitic
inductance in the PC-board traces.

Modern motherboards place one or more high-speed, high-current voltage
step-down switching regulators _very_ close to the CPU, to step the 5
or 12-volt rail voltages down to the high-amperage sub-2-volt supplies
needed by many CPUs. These regulators need hefty amounts of
capacitance on their input sides to remain stable, and some on the
outputs as well.

You can't depend solely on small (.1 or .01 uF) chip capacitors for
bypassing. The capacitance can interact with the inductance of the PC
board traces to create a high-Q resonant circuit, which leaves you
with _lousy_ bypassing at the resonant frequency. It's necessary to
add low-Q electolytics to the traces, every few inch or two, to
"swamp" the resonances and restore proper bypass effectiveness.

Basically, today's CPUs are doing much more work, and are demanding a
lot more of their power supplies.
Add to this, the fact that they are running at lower voltages, so the
Click to expand...
currents involved have risen even faster. The peak current at some
instants when particular combinations of gates switch, can be well over
100A, and the rate of change of current is enormous. The ESR of even good
quality capacitors is high enough that paralleling up several becomes
necessary.

Best Wishes
 
alpha_uma (none_such@home.com) said those last words:
I've noticed that the number of electrolytic capacitors on a typical "P4
socket 478" motherboard have mushroomed from, say, the previous "P3 socket
370" generations of motherboards. Why are they using so many electrolytic
capacitors these days? Not only are those caps an eyesore, they have such
short life-times--especially those used in the voltage regulator circuits.

They are not very noticeable on a socket 370 m/b, and you only see them
occasionally on, say, the first generation of Pentium motherboards (socket
7). I don't think I've ever noticed electrolytic capacitors on pre-Pentium
generations of motherboards. As a matter of fact, let me run to my storage
cabinet, and take a look at my 486 and 386 relics: no, I don't see them.

Can someone enlighten me? Thanks
Click to expand...
I guess it's because the requirement of power from processors increased.

If you search around on Google for "liquid nitrogen overclock" (without quotes)
you may be able to find an article from tomshardware.com, where he changed
capacitors on the motherboard, to increase power "ratings" for the motherboard,
to try overclock of Pentium 4 processor.

[]s
--
Š Chaos Master. | "I've tried so hard to tell
My Evanescence HP is at: | myself that you're gone
http://marreka.no-ip.com | And thought you were still
(most often offline... ) | with me..."
------------------------- -- Evanescence, "My Immortal"
 
"Dave Platt" <dplatt@radagast.org> wrote in message
news:10ekchhiubgqb44@corp.supernews.com...
snip
You can't depend solely on small (.1 or .01 uF) chip capacitors for
bypassing. The capacitance can interact with the inductance of the PC
board traces to create a high-Q resonant circuit, which leaves you
with _lousy_ bypassing at the resonant frequency. It's necessary to
add low-Q electolytics to the traces, every few inch or two, to
"swamp" the resonances and restore proper bypass effectiveness.
snip
Click to expand...
Hi, thanks for the explanations. You seem to be saying that the physics
behind the internals of the electrolytics have some "magic" over the chip
capacitors when it comes to counteracting the effects of LC resonances. I
know nothing about the internal construction of capacitors--come to think of
it, I know nothing about how resistors are constructed either. Can you
explain further some of this "magic" behind the electrolytics?
Al-U
 
"alpha_uma" <none_such@home.com> wrote in message news:<NYoGc.32385$P7.25753@pd7tw3no>...
I've noticed that the number of electrolytic capacitors on a typical "P4
socket 478" motherboard have mushroomed from, say, the previous "P3 socket
370" generations of motherboards. Why are they using so many electrolytic
capacitors these days? Not only are those caps an eyesore, they have such
short life-times--especially those used in the voltage regulator circuits.

They are not very noticeable on a socket 370 m/b, and you only see them
occasionally on, say, the first generation of Pentium motherboards (socket
7). I don't think I've ever noticed electrolytic capacitors on pre-Pentium
generations of motherboards. As a matter of fact, let me run to my storage
cabinet, and take a look at my 486 and 386 relics: no, I don't see them.

Can someone enlighten me? Thanks
Al-U
Click to expand...
Many good explanations of the electrical aspects. What do you mean
'eyesore' ? These motherboards are modern works of art you could
previously only dream of. I bet these systems exceed what most of us
imagined 5,10 years back.
GG
 
"Glenn Gundlach" <stratus46@yahoo.com> wrote in message
news:acb22b57.0407101002.66cf8c44@posting.google.com...
snip
What do you mean 'eyesore' ? These motherboards are modern
works of art you could previously only dream of. I bet these
systems exceed what most of us imagined 5,10 years back.
Click to expand...
Yes, technology marches on for sure. But beauty is in the eye of the
beholder, I guess, :)

"Eyesore" is, perhaps, a bit harsh to describe the caps. But I think "pesky"
may be closer to what I meant. It's just that I think those electrolytic
caps are the weakest "links" on the entire motherboard (even if they were
made from the correct chemical formulas). I'd like my motherboards to last
at least five years, but it seems that these caps have much shorter
lifetimes. Personally, I just don't like the idea of having rows and rows of
cylinders containing "liquid gel" on my motherboard, :)

Assuming that a good chemical formula is being used to make the electrolyte,
what is the expected lifetime of an electrolytic cap, say, 1500uF, 105degC
and 6.3V? How long does it take for the electrolyte to dry up inside or leak
under "normal" usage and operating conditions?

Al-U
 
"alpha_uma" <none_such@home.com> wrote in message
news:kdbIc.1002526$Pk3.273388@pd7tw1no...
<...>
Assuming that a good chemical formula is being used to make the
electrolyte,
what is the expected lifetime of an electrolytic cap, say, 1500uF,
105degC
and 6.3V? How long does it take for the electrolyte to dry up inside or
leak
under "normal" usage and operating conditions?
Click to expand...
Thirty years or so?


--
Graham W http://www.gcw.org.uk/ PGM-FI page updated, Graphics Tutorial
WIMBORNE http://www.wessex-astro-society.freeserve.co.uk/ Wessex
Dorset UK Astro Society's Web pages, Info, Meeting Dates, Sites & Maps
Change 'news' to 'sewn' in my Reply address to avoid my spam filter.
 
On Sun, 11 Jul 2004 13:25:36 GMT, "alpha_uma" <none_such@home.com>
wrote:

Assuming that a good chemical formula is being used to make the electrolyte,
what is the expected lifetime of an electrolytic cap, say, 1500uF, 105degC
and 6.3V? How long does it take for the electrolyte to dry up inside or leak
under "normal" usage and operating conditions?
Click to expand...
The life of a capacitor can vary from microseconds to decades
depending on what it is subjected to. If the motherboard is properly
designed, all of the capacitors will still be good long after you get
tired of the computer. Caps in the power supply are much more likely
to fail since they are closer to the AC and its spikes and transients.
 
On Tue, 06 Jul 2004 03:25:33 GMT, the renowned "alpha_uma"
<none_such@home.com> wrote:

I've noticed that the number of electrolytic capacitors on a typical "P4
socket 478" motherboard have mushroomed from, say, the previous "P3 socket
370" generations of motherboards. Why are they using so many electrolytic
capacitors these days? Not only are those caps an eyesore, they have such
short life-times--especially those used in the voltage regulator circuits.

They are not very noticeable on a socket 370 m/b, and you only see them
occasionally on, say, the first generation of Pentium motherboards (socket
7). I don't think I've ever noticed electrolytic capacitors on pre-Pentium
generations of motherboards. As a matter of fact, let me run to my storage
cabinet, and take a look at my 486 and 386 relics: no, I don't see them.

Can someone enlighten me? Thanks
Al-U
Click to expand...
They want to use 0.50 worth of e-caps instead of $20 worth of ceramic
caps.

Best regards,
Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
speff@interlog.com Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com
 
"Spehro Pefhany" <speffSNIP@interlogDOTyou.knowwhat> wrote in message
news:gh05f0hh0pnj10mcbit3h2qiqbkacfkrjr@4ax.com...
They want to use 0.50 worth of e-caps instead of $20 worth of ceramic
caps.
Click to expand...
If money is of no concern (big IF, I know, but it's just hypothetical), and
if there are ceramic caps that would meet the specs, would you use ceramics
instead of electrolytics on a motherboard? In the current state of
consumer-oriented motherboard designs, are there (non-finanacially related)
reasons where electrolytics must be used in circuits?
Al-U
 
On Wed, 14 Jul 2004 09:14:54 GMT, the renowned "alpha_uma"
<none_such@home.com> wrote:

"Spehro Pefhany" <speffSNIP@interlogDOTyou.knowwhat> wrote in message
news:gh05f0hh0pnj10mcbit3h2qiqbkacfkrjr@4ax.com...

They want to use 0.50 worth of e-caps instead of $20 worth of ceramic
caps.


If money is of no concern (big IF, I know, but it's just hypothetical), and
if there are ceramic caps that would meet the specs, would you use ceramics
instead of electrolytics on a motherboard? In the current state of
consumer-oriented motherboard designs, are there (non-finanacially related)
reasons where electrolytics must be used in circuits?
Al-U
Click to expand...
I really can't think of any, particularly in a motherboard that
doesn't have to operate below 0°C. Parts in the 100uF-330uF 6.3V range
are available. They are the weakest link in motherboard life (not
counting obsolescence and processor fans).

Best regards,
Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
speff@interlog.com Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com
 
If money is of no concern (big IF, I know, but it's just hypothetical), and
if there are ceramic caps that would meet the specs, would you use ceramics
instead of electrolytics on a motherboard? In the current state of
consumer-oriented motherboard designs, are there (non-finanacially related)
reasons where electrolytics must be used in circuits?
Click to expand...
It's my understanding that you do _not_ want a board's bypass
capacitance to consist solely of low-ESR, high-Q capacitors. If you
try this, you're likely to end up with some very severe ringing on the
power supply rails, at frequencies where the capacitance resonates
with the parasitic inductance of the PC-board traces. You end up with
a distributed set of high-Q tank circuits, and power/ground bounce
which will make your life really miserable.

The usual rule-of-thumb I've read, is to have no less than one
higher-capacitance, higher-ESR cap (e.g. 1 uF tantalum or 10 uF
aluminum electrolytic) for every few inches of power-supply trace... a
ratio of one such bulk-capacitance "Q-spoiler" cap for every 5 to 10
high-Q chip capacitors.

I suppose you could use a 1 uF ceramic cap in series with a 1-ohm chip
resistor for a similar purpose, but it seems unnecessarily clumsy.

[I once fought a prolonged battle with Boca, who had made an Ethernet
card based on an AMD PCNET Ethernet chip, and had blithely ignored
AMD's urgent recommendations about the required number of chip and
bulk capacitors which the chip needed for reliable operation. The
card had a severe ground-bounce problem on its receiver pins, and
ended up being bit-pattern-sensitive. Certain packets contained data
patterns which stimulated one of the high-Q resonances, bounced the
power and ground rails out of spec, and resulted in a misread and a
frame-check-sequence mismatch 100% of the time - most other packets
sailed through just fine. It was hell... FTP sessions would seize up
transferring certain files, while other files and TELNET sessions
worked just fine. I went through three cards, never got an
acceptable fix, and eventually demanded and got most of my money
back.]

For long life, I believe that either 105-degree-C-rated aluminum
electrolytics with a high surge-current rating, or some of the new
(and admittedly expensive) solid-polymer electrolyte caps can be used.
The latter seem to have a very long rated life, even at elevated
temperatures.

--
Dave Platt <dplatt@radagast.org> AE6EO
Hosting the Jade Warrior home page: http://www.radagast.org/jade-warrior
I do _not_ wish to receive unsolicited commercial email, and I will
boycott any company which has the gall to send me such ads!
 
"Dave Platt" <dplatt@radagast.org> wrote in message
news:10fauk7cl1a86bf@corp.supernews.com...
:
: It's my understanding that you do _not_ want a board's bypass
: capacitance to consist solely of low-ESR, high-Q capacitors. If
you
: try this, you're likely to end up with some very severe ringing
on the
: power supply rails, at frequencies where the capacitance
resonates
: with the parasitic inductance of the PC-board traces. You end
up with
: a distributed set of high-Q tank circuits, and power/ground
bounce
: which will make your life really miserable.
:
: The usual rule-of-thumb I've read, is to have no less than one
: higher-capacitance, higher-ESR cap (e.g. 1 uF tantalum or 10 uF
: aluminum electrolytic) for every few inches of power-supply
trace... a
: ratio of one such bulk-capacitance "Q-spoiler" cap for every 5
to 10
: high-Q chip capacitors.
:

The rule of thumb I have always observed is that I should have 100
times the energy in a bulk capacitor to maintain the rails
regardless of the inductance of the power wiring to the supply
source. So if I have 20 each .1mF I need a 200mF bulk capacitor
to hold the rail up..... But I never have seen this in a book or
article. Just my rule and I see a lot of application of it by
others.
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top