Capacitor

[Paul]

Caps can do many different things. It depends on the type and cicruit
design. If you want a 'hint' about the use of electrolytic caps on a
modern mobo, look carefully at the rating....if the ratiing is 105
centigrade you can be almost sure it's in a switching
circuit...why...well mobo producers try to get the board cost down to
the lowest possible level. 105 C caps cost significantly (in
production x 1000's of brds) but are needed in switching
(regulating/dropping) applications.

As for "why it kept working with an electrolytic cap that
'vented'...well, if you take a close look on most mobos, you'll see
that in order to get the bulk capacitance they need and still have a
low profile, most boards use multiple caps (add the UF value) in
parallel. Having 3 to 7 caps so connected is common....so; when one
such cap failed the mobo might still appear to be "just fine"...but
without an understand and the ability to measure the voltage, amperage
and ripple current etc under heavy loads...well, assuming "all is
still ok" is a risky guess.

The number of capacitors used is not to achieve a large bulk
capacitance. There are several parameters to optimize at the
same time - capacitance, ESR, ESL. In fact, in terms of limits,
the total capacitance is seldom the issue, it is more to do
with the other parameters. (This is why, if you get a board
repaired, don't panic if the capacitance value of the replacement
caps is not exactly the same value.)

http://www.intersil.com/data/FN/FN4/FN4567/FN4567.pdf (pg. 9)

"In most cases, multiple electrolytic capacitors of small case
size perform better than a single large case capacitor."

HTH,
Paul

 

[Paul Busby]

Find someone who is handy with a soldering iron and get him (or her)
to change the capacitors. It isn't difficult (or expensive) if you
know what you are doing - but make sure the new capacitors are the
same value as the old ones and they are installed with the polarity
(+ and -) the same.

Or install the new capacitors with the correct polarity 'cos the original
ones may have been inserted the wrong way round which is why they leaked.
The value is not the most important factor, electrolytics have extremely
wide tolerances (-10 to 50/75%?). IMO, not fitting too high or too low a
voltage rating as a replacement is more important. Just as well that they
aren't tantalum caps - they usually go short circuit!

Regards

 

[Tom S]

You really don't understand it do you. On a 1A load from a 12v input, a
7805, will have to dissipate 7W. From a 24v supply, it'll have to
dissipate 19W. The heat most definitely _is_ different.

That guy doesn't understand Ohm's law, Roger, or how linear regulators work.

Tom S

 

[seeker]

The number of capacitors used is not to achieve a large bulk
capacitance. There are several parameters to optimize at the
same time - capacitance, ESR, ESL. In fact, in terms of limits,
the total capacitance is seldom the issue, it is more to do
with the other parameters. (This is why, if you get a board
repaired, don't panic if the capacitance value of the replacement
caps is not exactly the same value.)

http://www.intersil.com/data/FN/FN4/FN4567/FN4567.pdf (pg. 9)

"In most cases, multiple electrolytic capacitors of small case
size perform better than a single large case capacitor."

HTH,
Paul

======> Paul...

Perhaps I shouldn't have used the term "bulk". Still, I think we both
know the several reasons for caps in parallel...among which have to
due with size and other factors too. But, if you check the major
suppliers of Low ESR caps as used on mobos, you'll find that to get
Low ESR, hig temp, ability to stand ripple etc, will neraly only be
found in relatively small sized and value caps. I stand by the fact
that this is why multiples are used in parallel....but agree that this
point can become a "chicken vs the egg" discussion.. And, you are
right that if a cap has at least the same WVDC, Temp, type and ESR
characteristics, then small variationsin capacitance can be tolerated.
But mainly the problem to repair these mobos is that replcament caps
of equal VDC/LowESR/Uf value/Temp are just very hard to find in the
same physical pkg size as used by the mobo producer. In fact, about
2 years ago there was a huge problem with Abit and Intel MBs...(maybe
others) as these critical caps were failing/shorting/venting at an
alaraming rate. There was blame that these companies took the
cheapest crap they could find from China and that the caps were not
produced correctly...and in fact it was hard to do so with the same
values in the same case size.

 

[Roger Hamlett]

That guy doesn't understand Ohm's law, Roger, or how linear regulators

work.
Either that, or he has managed to invent perpetual motion by a
'roundabout' method (if the regulator still generates the same heat in the
lower voltage feed case, with the same load, he is getting power from
nowhere....

Best Wishes

 

[Leythos]

That guy doesn't understand Ohm's law, Roger, or how linear regulators work.

Sure I do, but I don't have to live in theory only, try it some time.
Take a regulator of your choice, the 7805 is a nice package, use an
input voltage of 24VDC and then 12VDC with a load of 500ma on the 5v
side - you won't notice the difference in temp.

 

[MM]

Sure I do, but I don't have to live in theory only, try it some time.

Take a regulator of your choice, the 7805 is a nice package, use an
input voltage of 24VDC and then 12VDC with a load of 500ma on the 5v
side - you won't notice the difference in temp.

Sure, you don't. 7805 will either blow up or will go into internal thermal
protection mode should you put it into such a circuit. It is a 100 mA
regulator and there is no way it can dissipate (24-5)*0.5=9.5W of power or
even 3.5W in case of 12V input.... Sorry.

/MM

 

[Leythos]

Sure, you don't. 7805 will either blow up or will go into internal thermal
protection mode should you put it into such a circuit. It is a 100 mA
regulator and there is no way it can dissipate (24-5)*0.5=9.5W of power or
even 3.5W in case of 12V input.... Sorry.

Try again, the LM7805 will handle 1A if properly heatsinked.
http://www.national.com/pf/LM/LM7805C.html

 

[Arnie Berger]

Hello,

perhaps this is dumb, but I do not know an answer.

I got 2 machines at work which work like a charm. No hangups, no bluescreen.
Perfect.

Now as I opened them for cleaning I noticed 2 capacitors with brown crusting
on the top. Seems as they had those for a little while. Nowhere else was the
crusting and it has dried.

Now my question is: The machines work, without a flaw, beleve it or not. Do
I have to change the mobo´s (boards) or can I stick with the present ones?
Like never change a running system?

Thanks for any advice-

If the caps are part of the switching power supply that converts the
the power supply voltages to the core voltages of the CPU, then I
doubt that your computer would still be working. However, the caps may
be filter caps that are spread around the board to provide some
additional filtering and noise reduction to the switching transients
on the ground and power planes. This is a fairly standard board design
practice. If that's the case, you'd probably never know the
difference.

arnie

 

[Frank]

If the caps are part of the switching power supply that converts the

the power supply voltages to the core voltages of the CPU, then I
doubt that your computer would still be working. However, the caps may
be filter caps that are spread around the board to provide some
additional filtering and noise reduction to the switching transients
on the ground and power planes. This is a fairly standard board design
practice. If that's the case, you'd probably never know the
difference.

Thanks for all the answers. As for now I´m no good at soldering, I have no
warranty on these 2 machines. I got 2 spare boards in the closet, so I´m set
for all things that can happen. As the still work like a charm I´m not doing
anything by now. Thanks for all the help, again.

Frank

 

[Frank]

.. Wenn nicht, arbeite

solange es geht mit dem Board weiter und tausch es gegen ein Neues aus,
sobald Fehler auftreten.

Hallo Carsten,

danke für die Hilfe, und genau das werde ich tun.

Gruß

Frank

 

[MM]

Try again, the LM7805 will handle 1A if properly heatsinked.

http://www.national.com/pf/LM/LM7805C.html

OK, I was looking at LM78L05, which is a lower power version of the LM7805.
With the original LM7805 you might be able to actually dissipate your 9.5W,
but take a look at the Maximum Power Dissipation graphs in the datasheet, it
will only work with a hefty heatsink or in a fridge. The bigger device
doesn't change the basics of the Ohm's law. It is actually bigger because of
these basics!


/MM

 

[Leythos]

OK, I was looking at LM78L05, which is a lower power version of the LM7805.
With the original LM7805 you might be able to actually dissipate your 9.5W,
but take a look at the Maximum Power Dissipation graphs in the datasheet, it
will only work with a hefty heatsink or in a fridge. The bigger device
doesn't change the basics of the Ohm's law. It is actually bigger because of
these basics!

In looking at the T0-3 unit, with heat-sink, sitting near my desk in a
power supply, and another one with the TO-220 style package, with heat-
sink, also sitting here (since I have about 30 of the 7805 devices and
another 10 LM338 and 317's and a bunch of power transistors and others.
I can assure you that the TO-3 and TO-220 units can deliver a sustained
500MA draw with little problem using a normal heat-sink (even one from
RadioShack) and a minimal amount of HSP.

They can also provide a sustained 1A of power, but, with a standard
heat-sink, you do notice a little heat from the fins. I would imagine if
you were to push beyond 1A a cooling fan might be required.

 

[MM]

I can assure you that the TO-3 and TO-220 units can deliver a sustained

500MA draw with little problem using a normal heat-sink (even one from
RadioShack) and a minimal amount of HSP.

I've no doubt it can deliver 500 mA, the question is how big a voltage drop
it can handle with this load. Somehow you don't want to admit the obvious.
The maximum junction temperature for both of these cases (TO3 and TO220) is
150 C. The junction-to-case thermal resistance is 4 C/W. Assuming your
heatsink has 10C/W resistance to ambient (which is pretty good) we get 14
C/W junction-to-ambient minimum resistance. At 9.5W of power dissipation
you'll have 9.5*14=133 C temperature differential between the junction and
ambient. Since the temperature in a computer case is usually higher than in
a room, let's say 35 C, you will have your junction at 168 C, which is well
above the spec. Do you still want to try 1 A at 24VDC input (19W to
dissipate)? Let us know how well it worked.


/MM

 

[Leythos]

I've no doubt it can deliver 500 mA, the question is how big a voltage drop
it can handle with this load. Somehow you don't want to admit the obvious.
The maximum junction temperature for both of these cases (TO3 and TO220) is
150 C. The junction-to-case thermal resistance is 4 C/W. Assuming your
heatsink has 10C/W resistance to ambient (which is pretty good) we get 14
C/W junction-to-ambient minimum resistance. At 9.5W of power dissipation
you'll have 9.5*14=133 C temperature differential between the junction and
ambient. Since the temperature in a computer case is usually higher than in
a room, let's say 35 C, you will have your junction at 168 C, which is well
above the spec. Do you still want to try 1 A at 24VDC input (19W to
dissipate)? Let us know how well it worked.

While you won't like my answer, and I'm sure you say it's not true, I've
been doing it for years. Almost always use +24 to the supply side for
the +15, +10, +5 supply circuits, and -24 for the -15, -10 circuits. The
have handled 500MA loads for years with the standard heat-sink and HSP.

You and I can both read the spec's, and I have never disagreed with
them, not once, but what works in the real world is different that the
specs many times. Heck, I even have one +/- 15vdc follower that I built
in the early 80's that still works (changed the caps a few years back),
and it still follows within about 3mv all day long.

 

[MM]

While you won't like my answer, and I'm sure you say it's not true, I've
been doing it for years. Almost always use +24 to the supply side for
the +15, +10, +5 supply circuits, and -24 for the -15, -10 circuits. The
have handled 500MA loads for years with the standard heat-sink and HSP.

You are a strange person I should say. There is a huge difference between
15V and 5V output in this case! With 5V at 500mA you are on the edge of what
can be done with a 7805. At 15V you do have some margin.

You and I can both read the spec's, and I have never disagreed with
them, not once, but what works in the real world is different that the
specs many times.

If this happens, it usually means either error in measurements or in how
these measurements are being interpreted. Besides, this time you are
challenging much more than a spec, you are challenging one of the basic laws
of physics.

/MM

 

[Leythos]

You are a strange person I should say. There is a huge difference between
15V and 5V output in this case! With 5V at 500mA you are on the edge of what
can be done with a 7805. At 15V you do have some margin.

Maybe I should have been more clear - the +15 and +10 are not done with
the LM78XX series. The 5V is and works fine, even at loads of 1A over
hours of continuous operation.

If this happens, it usually means either error in measurements or in how
these measurements are being interpreted. Besides, this time you are
challenging much more than a spec, you are challenging one of the basic laws
of physics.

Ah, the old, specs could be wrong, physics could be wrong, something
must be wrong, cause you can't explain how it can work on paper when it
works in real life.

Here's and Idea, instead of telling me something that I'm doing can't be
done, that the spec's don't allow for it, try it yourself, the LM7805C's
are cheap, pick your package type, and try it - just try it yourself
before you tell me what I'm doing isn't possible.

 

[MM]

works in real life.

Here's and Idea, instead of telling me something that I'm doing can't be
done, that the spec's don't allow for it, try it yourself, the LM7805C's
are cheap, pick your package type, and try it - just try it yourself
before you tell me what I'm doing isn't possible.

I am doing this kind of thing pretty much every day. That's what I am being
paid to do. So far, I haven't been able to break the Ohm's law. Why don't
YOU try to load your perpetum mobile with a 5A load and feed it with 100VDC.
It shouldn't make any difference in heat, right? Go and measure your
voltages, currents and temperatures properly and then we can see whether
theory is coherent with practice.

/MM

 

[Leythos]

works in real life.

I am doing this kind of thing pretty much every day. That's what I am being
paid to do. So far, I haven't been able to break the Ohm's law. Why don't
YOU try to load your perpetum mobile with a 5A load and feed it with 100VDC.
It shouldn't make any difference in heat, right? Go and measure your
voltages, currents and temperatures properly and then we can see whether
theory is coherent with practice.

I hear that same rhetoric from people all the time. I don't care if
you're paid to do it, like that means your qualified. I don't care if
you just want to whine about it not being possible. I don't care if you
don't seem to understand that it's already being done and working. I've
been building power supply units since the 70's, and unlike you, I've
not been blinded by the "I can only do what the spec's say I can do"
mentality.

Like I said, try it yourself, put aside your big headed idea that it's
not possible, just try it, you might surprise yourself. What I've seen
from your replies is symptomatic of the typical "the spec's say it can't
work, so it can't work type of people". Open your eyes once and quit
being such a closed minded person and just try it, it's a simple
circuit, you should be able to build it in under an hour (even if you
have to solder it yourself). You might even have the parts, since you're
pretty much paid to do this this every day - if you don't have the TO-3
LM7805 you can order one for about $5 from suppliers.

One of these days you might want to actually try something before you
speak out that it's impossible, when you do try it you're going to look
funny posting that it did work (not that I expect that from someone like
you).

 

[MM]

Like I said, try it yourself, put aside your big headed idea that it's
not possible, just try it, you might surprise yourself.

You are not hearing me. I am trying this kind of thing every day and I am
not seeing the results you see. If you want me to try something be more
specific. What it is precisely you want me to try and what it is I will see?
What are the design goals/parameters:

Input voltage, load, max and min ambient temperature, heat sink thermal
resistance, etc?

/MM