So few capacitors on motherboard?

  • Thread starter Thread starter larrymoencurly
  • Start date Start date
L

larrymoencurly

I just bought an ECS K7VTA3 ver. 8c mobo and noticed that it has only
about a third as many electrolytic capacitors as most mobos. Has
anyone ever seen a fairly recent mobo with so few?

Before you say that ECS makes cheap mobos (this one was $40 at Fry's,
with Duron CPU), my other ECS mobos, including a K7S5A Pro ver. 5,
each have about 45 electrolytics, and even my old Soyo Intel 810i
mobo, which has only a single-phase CPU core voltage regulator, has
over 40.

The K7VTA3 also has about six places where capacitors were left out,
including one for the CPU core voltage regulator, while on my K7S5A
only about two were left out. Has anyone ever noticed an improvement
in stability by adding missing capacitors?
 
larrymoencurly said:
I just bought an ECS K7VTA3 ver. 8c mobo and noticed that it has only
about a third as many electrolytic capacitors as most mobos. Has
anyone ever seen a fairly recent mobo with so few?

Ummmm . . . although this should be obvious, not all capacitors are equal.
You can't make assumptions about the design of a motherboard based on
counting the number of physical capacitors. In order to compare two
motherboards, you would need to determine the function of *each* capacitor,
and then find the specifications for *each* capacitor that performs a
similar function. But for comparison purposes, I've got a fairly recent
Athlon XP motherboard by Epox (nforce 2 chipset) that has 13 capacitors
clustered around the voltage regulators. There are only 28 total on the
entire board, nowhere near 40 or 45. It has been running rock-solid stable
for several months, no problems at all. When I was planning the system, my
plan "B" was a board made by ECS. I don't think you'll have any problem
with your ECS board. The power supply has more to do with stability than
capacitors on the motherboard. Get a good power supply, and you should be
all set. -Dave
 
I just bought an ECS K7VTA3 ver. 8c mobo and noticed that it has only
about a third as many electrolytic capacitors as most mobos. Has
anyone ever seen a fairly recent mobo with so few?

Before you say that ECS makes cheap mobos (this one was $40 at Fry's,
with Duron CPU), my other ECS mobos, including a K7S5A Pro ver. 5,
each have about 45 electrolytics, and even my old Soyo Intel 810i
mobo, which has only a single-phase CPU core voltage regulator, has
over 40.

The K7VTA3 also has about six places where capacitors were left out,
including one for the CPU core voltage regulator, while on my K7S5A
only about two were left out. Has anyone ever noticed an improvement
in stability by adding missing capacitors?

Is it instable?
In what way if so? Memory bus or overclocking or ???
You may need test these parts in another system to get a feel for
their potential (if you haven't already) so the board is the only
variable.

Are the capacitors running warm->hot?

I'm sure it is a cost-cutting measure, since ECS doesn't use any caps
of high enough quality that fewer could substitute, like Rubycon MBZ,
Fujitsu hybrids, or OS-Cons. Then again, larger can-size means lower
ESR in same "family", so fewer would work. Strategically placed chip
caps could work too, but I think it's pretty much a given that the
board won't last as long, except that it's using a (now) relatively
low-powered CPU, which will help.

You could add some, but what's the point? A lot of time and $ on more
caps, when a different board would have more already.
i suppose if you have the spare caps lying around it's something to
try, though it might be better left unused, as a backup board, system,
in case of emergency.
 
Ummmm . . . although this should be obvious, not all capacitors
are equal. You can't make assumptions about the design of a
motherboard based on counting the number of physical capacitors.
In order to compare two motherboards, you would need to determine
the function of *each* capacitor, and then find the specifications
for *each* capacitor that performs a similar function. But for
comparison purposes, I've got a fairly recent Athlon XP motherboard
by Epox (nforce 2 chipset) that has 13 capacitors clustered around
the voltage regulators. There are only 28 total on the entire
board, nowhere near 40 or 45. It has been running rock-solid stable
for several months, no problems at all. When I was planning the
system, my plan "B" was a board made by ECS. I don't think you'll
have any problem with your ECS board. The power supply has more to
do with stability than capacitors on the motherboard.

I'm probably worrying about nothing, but it's just strange that the
same manufacturer would cut way back on capacitors for a mobo that has
roughly the same features and number and types of chips as the the
K7S5A Pro (even though the former uses a VIA chipset and the SiS).
And while it's only ignorant speculation on my part, I can't believe
that the new mobo's chips are far more tolerant of spikes or dirty
signals than the other mobo's are. I also didn't see any surface
mount electrolytics, just ceramics, and there seemed to be just as
many ceramics on the older board. But an engineer friend suggested
that maybe the new mobo needs fewer bypass capcitors because it has a
second ground plane inside -- something about lowering the Q of the
board that way.
Get a good power supply, and you should be all set.

I'm sure won't use my $13 PSU with it; I'm gonna hook it up to my
deluxe $15 one! ;) Actually they're really good, Delta and
Fortron/Sparkle, bought as close-outs.
 
Is it instable?

It's been 100% stable since testing started last night, except when I
set a jumper wrong and overclocked the 1.6 GHz CPU to 2.0 GHz. I
don't overclock, except for my 300 MHz Celeron, which has been at 450
MHz for four years.
Are the capacitors running warm->hot?

They're all cool, as are all the transistors and chokes, except the
transistor of the linear regulator of the 2.5V memory supply. OTOH my
ECS P4S5A2 has some switching transistors and adjacent capacitors that
have run a bit warm since new, and that old Celeron mobo has a 3.3V
switching regulator transistor and capacitors that have been at 70C
for four years (was 90C without heatsink).
I'm sure it is a cost-cutting measure, since ECS doesn't use any caps
of high enough quality that fewer could substitute, like Rubycon MBZ,
Fujitsu hybrids, or OS-Cons.

All the capacitors are OST brand, except for a few black & white
Luxon-G
bypass caps. My other ECS boards use Luxons for bypass, another brand
for their
switching regulators. I didn't see any chip electrolytics, only chip
ceramics, and this K7VTA3 mobo doesn't seem to have any more of them
than my older ECS mobo.
You could add some, but what's the point? A lot of time and $ on more
caps, when a different board would have more already.
i suppose if you have the spare caps lying around it's something to
try, though it might be better left unused, as a backup board, system,
in case of emergency.

An engineer friend of mine bought three of these mobos and added the
missing capacitors, but he does that with every board he gets. He
suggested that the K7VTA3 could require fewer capacitors if it has a
second
ground plane inside that "cuts the Q" of the circuit board.
 
On 3 Jan 2004 12:41:19 -0800, (e-mail address removed)
(larrymoencurly) wrote:

But an engineer friend suggested
that maybe the new mobo needs fewer bypass capcitors because it has a
second ground plane inside -- something about lowering the Q of the
board that way.

It's a nice theory, but considering ECS, I doubt it.
I'm sure won't use my $13 PSU with it; I'm gonna hook it up to my
deluxe $15 one! ;) Actually they're really good, Delta and
Fortron/Sparkle, bought as close-outs.

The Delta should be pretty good for that... IIRC, it's output filter
caps are larger than most ~400W models... here on Noah's ark I have
two of everything, including those Deltas. I like the Compaq/LiteOns
more though.
 
I just bought an ECS K7VTA3 ver. 8c mobo and noticed that it has only
about a third as many electrolytic capacitors as most mobos. Has
anyone ever seen a fairly recent mobo with so few?

Before you say that ECS makes cheap mobos (this one was $40 at Fry's,
with Duron CPU), my other ECS mobos, including a K7S5A Pro ver. 5,
each have about 45 electrolytics, and even my old Soyo Intel 810i
mobo, which has only a single-phase CPU core voltage regulator, has
over 40.

The K7VTA3 also has about six places where capacitors were left out,
including one for the CPU core voltage regulator, while on my K7S5A
only about two were left out. Has anyone ever noticed an improvement
in stability by adding missing capacitors?
Pointless exercise. You won't gain any stability by adding them. The
reason they'll be missing is because the components for a particular
motherboard feature won't be there.
 
I'm probably worrying about nothing, but it's just strange that the
same manufacturer would cut way back on capacitors for a mobo that has
roughly the same features and number and types of chips as the the
K7S5A Pro

Err, WHY? Is it exactly the same design? No.
 
Err, WHY? Is it exactly the same design? No.

Actually it is to a large extent... supports similar processors,
chipsets, plug-in devices... might even have the exact same
regulators/controller as on their other boards with more capactors.
 
Pointless exercise. You won't gain any stability by adding them. The
reason they'll be missing is because the components for a particular
motherboard feature won't be there.

Not necessarily true. I've done it, in some cases it helps. This is
more when overclocking though, but it stands to reason that as the
capacitors age, if they once prevented overclocking by their bare
adequacy, they'll sooner prevent spec'd speed.

The bottom line is that it's done to cut costs. Clearly none of the
(other) manufacturers want to wastefully add expensive parts where
there's no benefit.
 
Actually it is to a large extent... supports similar processors,
chipsets, plug-in devices... might even have the exact same
regulators/controller as on their other boards with more capactors.
Having all of the above the same does not make the design the same. It
merely means there are shared components. There are 1000's of ways of
interconnecting.
 
Pointless exercise. You won't gain any stability by adding them.
The reason they'll be missing is because the components for a
particular motherboard feature won't be there.

On this mobo there aren't any missing chips, unlike on my Intel 801i,
but there seem to be missing bypass electrolytics, and for some of the
big chips the closest electrolytics are a lot farther away than they
are for the big chips on my other mobos.
 
Having all of the above the same does not make the design the same. It
merely means there are shared components. There are 1000's of ways of
interconnecting.

.... and thousands of way of cutting costs at the consumer's detriment.
What they did, quite simply, is cut the engineering margins down a few
notches, under the assumption that cheaper boards will sell better.

You would have us believe they came up with an advanced design that
suddenly does away with components, yet the other manufacturers
haven't figured out how? Or is it more likely that they've been
making cheap boards for years and just got a little cheaper, caring
only if it works till the warranty expires?
 
larrymoencurly said:
The K7VTA3 also has about six places where capacitors were left out,
including one for the CPU core voltage regulator, while on my K7S5A
only about two were left out. Has anyone ever noticed an improvement
in stability by adding missing capacitors?

Yikes, you are scaring me. Look under the hood of your car. If you notice
a "missing" carburetor, you should add one!
 
jeffc said:
Yikes, you are scaring me. Look under the hood of your car. If you notice
a "missing" carburetor, you should add one!

Careful inspections of cars reveals electrical connectors that "don't
go anywhere". This does not necessarily mean that you car is bad.

Jeff
 
Yikes, you are scaring me. Look under the hood of your car. If you notice
a "missing" carburetor, you should add one!

If they decided it'd be cheaper to leave out a spark plug, would you
add it?
 
jeff findley said:
Careful inspections of cars reveals electrical connectors that
"don't go anywhere". This does not necessarily mean that you car
is bad.

What about the Dodge Neon, built with something like 900 fasteners in
the chassis, or less than half the usual number? I heard that this
caused problems, and I'm wondering if the ECS K7VTA3 ver. 5 is the
Dodge Neon of mobos. Its CPU core voltage regulator is 2-phase with
seven 3300uF filter capacitors (8th one left out), compared to 3-phase
with eleven 3300uF capacitors for the K7VTA3 ver. 3. I thought that
more phases at a given base frequency meant less capacitance was
needed, and while I don't know the ver. 3 base frequency, ver. 8 runs
at about the same 200 KHz as another 3-phase ECS mobo, the K7S5A Pro
(nine 3300uF capacitors).
 
larrymoencurly said:
jeff findley <[email protected]> wrote in message

What about the Dodge Neon, built with something like 900 fasteners in
the chassis, or less than half the usual number?

That's not the point. That's a design problem. But just because there are
empty holes on a circuit board doesn't mean you should start filling them in
with stuff. Look at virtually any halfway complicated circuit board and
there will be empty holes.
I'm wondering if the ECS K7VTA3 ver. 5 is the
Dodge Neon of mobos.

That's a more meaningful question. The only answer I know of is to search
the web for reviews of that motherboard to see if anyone thinks it was
designed well. That or get an advanced degree in computer and electrical
engineering.
 
kony said:
If they decided it'd be cheaper to leave out a spark plug, would you
add it?

This is getting silly. If a 6 cylinder car only had 5 spark plugs, then
obviously you wouldn't buy that car. 6th grade analogies aside, this is a
moot point.
 
Back
Top