What happens when you seriously increase capacitance?

[Fizzycapola]

Say you replace 6.3v 1000uF capacitors at RAM slots with premium brand 25v 4700uF or greater, what could happen? has anyone tried increasing capacitance all the way to failure? if so what is the failure that happens?

[willawake]

Re: What happens when you seriously increase capacitance?

there wont be a failure

though eventually there will be too much bulk capacitance your psu might not be able to start the computer

[Fizzycapola]

Re: What happens when you seriously increase capacitance?

So say your PSU has infinite power on all rails, it would be ok to use gigantic capacitors?

[kc8adu]

Re: What happens when you seriously increase capacitance?

in most cases low esr is more important than bulk.
a few boards like abit vp6 repond favorably to an increase like from stock 1500 to 2200 uf..

[Topcat]

Re: What happens when you seriously increase capacitance?

in most cases low esr is more important than bulk.
a few boards like abit vp6 repond favorably to an increase like from stock 1500 to 2200 uf..

You are correct. We both tested the hell out of that and came to the same conclusion on the vp6. However, on that note, don't overdo it. Don't start replacing all your 1500uF caps with 3300uF caps, you may end up with a dead board.

[PeteS in CA]

Re: What happens when you seriously increase capacitance?

One or more components in the compensation network in the feedback loop compensates for the RC network formed by the ESR and the capacitance of the output capacitor. A large increase in the capacitance is likely to cause changes in the regulator's response to load transients or even cause the regulator to oscillate.

[starfury1]

Re: What happens when you seriously increase capacitance?

Humm well for my money....and not directly related to your question on ram...others know this better then me

here's my looooong winded answer

The world of electronics is a matter of trade offs.

A components is usually selected to do a particular job in the circuit and its value is based on the design of the whole circuit.

that said alone come the bean counters and use the bare minimum part from a cost perspective they can get away with.

A capacitor can mostly in the DC world be considered a bucket...so what your saying is ok lets use the biggest bucket.

So what happens....its going to take more time to fill it...you have just change a parameter of the circuit.

So circuit operation will not be the same....and now other parameters will have to change to compensate
this will fall into two cat...
1 because you changed something and
2 if you want the circuit to work cause you did.

So for a quick example an analog PSU transformer bridge rectifier and capacitor.

All 3 components would have been selected to work together as one unit but suppose I put the biggest cap there I can find without regard to the transformer or rectifier.

What will happen is the in rush current demand of this whooping great capacitor may well kill the bridge rectifier....so we are back to trade offs....if I want it to work I am probably going to have to use a more current capable bridge and probably a different transformer, this of course depends on how big a cap I use.

Now I might get away with it if I only use the next size up in value.

So the above advice as outlined by the others applies.

Depending on the circuit design and how drastically the components have been down graded..a next size or two would be an advantage in some cases
but you must remember that it will affect other parts and circuit operation to some extent.

Then there is a point when there no advantage in going further, and you may actually be causing harm to the circuit by over stressing other components in the chain.

Generally speaking

If you don't know play safe and get as close to original spec part or go maybe 1 or 2 higher in value.

As to the ram...well my question is
1 how would it effect operation, as you have asked...and
2 would there be any advantage

If you only had a cap that size could you use it get away with it

OK HTH and yeah I know its long winded reply

Cheers all


your second reply...short answer
no it wouldn't, because there are most probably other components between them.....

[Fizzycapola]

Re: What happens when you seriously increase capacitance?

Thanks all for your prompt replies.

I will continue using exactly same capacitance replacements, just better quality as I've been doing, seeing some motherboards still run ok with faulty capacitors that only register ~20% of their specification, it makes values like 1000-2200uF seem quite adequate if they're quality enough to maintain this.

[starfury1]

Re: What happens when you seriously increase capacitance?

Well you can use higher as pointed out and higher voltage is not really a worry.
just that you don't want to push too far.


if you search through the forums you will get a good idea of what you can and can't get away with depending on their use

Size can be a problem too the larger you go

But as Kc8adu has pointed out ESR of the cap would probably be the more important factor with MB circuits then the actual capacitance

like a GP cap will work, probably...but die quick cause of its higher ESR

Also the marked value of a cap is plus and minus in the tens of percent
(20% according to the wiki )

So a nominal 1000uF could be in reality 800uF or 1200 uF worst case

Generally speaking here

If in any doubt just use the same and I guess you can't go wrong
(quality Low ESR)

Suppose the other is post what you got and what MB your working on
Someone will know if its ok or not.

here's a wiki on electrolytic capacitors

Cheers

[999999999]

Re: What happens when you seriously increase capacitance?

So say your PSU has infinite power on all rails, it would be ok to use gigantic capacitors?

No, because even with infinite power, there are still wiring, trace impedance and inductance inherant on supply circuits so even with that infinitely powerful PSU, it takes more time to reach the working voltage. Staying with the lowest ESR reasonable, you also increase the surge current which could eventually be a problem, failure point, though this is probably not enough of an increase to cause that, but on the other hand I don't know of any data on the long term effects of doing it.

[PCBONEZ]

Re: What happens when you seriously increase capacitance?

Back to the front:

""Say you replace 6.3v 1000uF capacitors at RAM slots with premium brand 25v 4700uF or greater, what could happen? has anyone tried increasing capacitance all the way to failure? if so what is the failure that happens?""

~~~~~~

Here is what happens because the voltage is too high:

[erosion 1]
The aluminum oxide layer slowly 'dissolves' into the electrolyte over time until an equilibrium is reached.
[ BTW: This is where the shelf life rating of caps comes from. ]

[erosion 2]
Ripple currents strip the aluminum oxide layer into the electrolyte due to the rapidly changing current direction. (Basically by electrolysis.) Some of the ions (charged particles) drift into solution and go too far from the dielectric to be replaced when the current direction reverses. The layer becomes thinner and thinner over time.

[rebuilding]
DC voltage replaces the layer. (Basically by more 'normal' electrolysis process.) The constant non reversing voltage in the correct direction builds up the layer (until the electrolyte runs out of ions.) - This repairs the damage done by both erosions.

=> "electrolysis", hence *electrolytic* capacitor.

[ I really don't know where this came from but I remember it as a thumb rule from way WAY back in electronics school. I've heard it now and then in other places. - Not really the sort of thing that comes up much, ya know? ]
-
The working voltage (circuit voltage) needs to be no less than (approximately) 1/3 of the rated voltage of the cap.
- Because:
If you don't have at least that much voltage the DC current isn't strong enough to rebuild the oxide layer. Since it isn't being replaced, erosion 1 and erosion 2 eventually destroy the oxide layer to a point of no return.

The oxide layer is an insulator and -IS- part of the dielectric.

As it becomes thinner and thinner:
(dielectric) the capacitance value is going to go down, eventually drastically.
(insulator) the current is going to rise, eventually drastically. This is going to overheat the internals of the cap and eventually the internal materials (including the electrolyte) will break down. The cap may form gas and vent, it may fail open, it may even short under these conditions.
~~~~~~
All of this assumes the failing cap doesn't cause another component failure somewhere else in the circuit.

[ Somewhere in the forums someone disassembled a cap that failed this way and posted pics. You can SEE that the 'paper' between the foil layers is brown and burned from the heat. It looks very much like paper that was left on a stove long enough to get brown, going for black, but didn't quite make it to flames.]

Your 25v cap should not be used in a circuit with less than about an 8.3v working voltage. (So that voltage would be just dandy on a 12v rail but no where else on a MoBo.)


~~~~~~~~~~~~

Capacitance too high is a different animal.
It really depends on what the rest of the circuit does.

Throwing the circuit out of balance have been covered.
[Adding to that if you go too far off the design value you may inadvertantly create a resonate circuit by the interactions with other components. Having a component intended to remove ripple voltages behave as a part time oscillator is not generally considered effective.]

Excessive charge current has been covered.

There is another one.
-
Higher capacitance caps take longer to charge and discharge and that can affect timings.
-
That bigger bucket from the past post isn't taller it's bigger around.
How high (deep) the water is in the bucket represents the voltage.
- Let's say that bucket needs to be 3/4 full to be at the circuit's working 'voltage'.
When you first start the computer that bucket starts filling and the BIOS runs the motherboard through POST checks. The BIOS gets to the check for the 'voltage' in that bucket (or rather the attached circuit) and it's not filled to 3/4 full yet.
-> POST Error -> lock-up or shut down. - NO GO!
----- And that's not going to change until you replace that cap.

~~~~~~~~

If you know what the working voltage in the circuit is, then you can make voltage changes safely by following the 1/3 rule.

As they age the capacitance of electrolytic caps goes down...
.. and .. they have a 20% tolerance in the first place.
---
So: If you raise the capacitance only one step you are (almost) always safe because as the cap ages it will be moving INTO the original operating range of the original cap anyway. [This assumes the original cap was a good choice in the first place -and- that it wasn't already raised once during a prior re-cap job.]

.

[starfury1]

Re: What happens when you seriously increase capacitance?

Gee thats a long one PCbonez and from the quick read fairly well covers quite a few aspects.

It also does answer another question that arises from time to time "why so many voltages with caps."

and why people tend to say "never stray to far from the original values as a rule"
1 step or maybe 2 max
(unless you know the design and can predict the behavior of the circuit, but still the voltage aspect above is something I suppose would be either not known or probably forgotten)

Yep I talk about bigger bucket I mean a fatter bucket not a taller one

To be honest I mostly forgotten the real theory behind caps and technically more accurate answers are important to a total understanding so thanks for that input

post this link as another question came up about the chemicals used for the electrolyte(and I'll know were to find the link) and it somewhat covers some of what you talked about.

http://www.elna-america.com/tech_al_principles.php



cheers

[PCBONEZ]

Re: What happens when you seriously increase capacitance?

That's an interesting page. - Thank you.

-
Yes,
If a higher voltage cap was -always- 'okay' to use then every cap on a motherboard would probably be standardized to some value like 16v or 25v because that would make designing motherboards (and buying parts for assembly lines) easier.

-
Because techs are 'supposed' to use the original value caps and 'know' it's usually okay to go up one value on a cap if they HAVE to, the minimum voltage to make the cap work doesn't come up very often.
-
I think most of the people that are even aware there is a minimum voltage are electronics design engineers of the kind that do all their work on paper (or a computer) and possibly don't even know how to solder. (I'm not 'talking them down' or anything. Physically working on circuits is just not their job.)
-
Those folks don't end up in places like this forum very often because repair is not what they are interested in and so that kind of theory information rarely makes it's way into the knowledge pool here.

-
I was an electronics tech (retired from) but because of what I worked on I got much more theory training than most techs do. - But once trained component level repair was only a small part of my job and that training was a long LONG time ago.
-
I too have forgotten most of the theory behind things.
Most of the time all I remember is some thumb-rule or limit or formula (calculation) exists, but not what it IS.
-
Google is my friend! - LOL
.

[999999999]

Re: What happens when you seriously increase capacitance?

That's an interesting page. - Thank you.

-
Yes,
If a higher voltage cap was -always- 'okay' to use then every cap on a motherboard would probably be standardized to some value like 16v or 25v because that would make designing motherboards (and buying parts for assembly lines) easier.

It is not really significant. A motherboard is populated with so many parts that whether 16V were used everywhere or lower becomes a matter of cost per part. Saving a few cents is the reason, and when there is no cost difference we do actually see capacitors used that are more than 3X the voltage rating. For example 50V 10uF is a very common value used extensively in 12V and lower circuits in many electronics including PSU and some motherboards.

Because techs are 'supposed' to use the original value caps and 'know' it's usually okay to go up one value on a cap if they HAVE to, the minimum voltage to make the cap work doesn't come up very often.

This is not really true, you're not necessarily supposed to use the original value. Original values are not always chosen because they are optimal in some way (other than cost). We see the same thing with caps not having low enough ESR, you don't try to match that higher ESR either, rather the tech is not constrained by a bean-counter to pick based on a few cents cost savings and really does not want to pick a functionally identical part, because that part failed once already.

I think most of the people that are even aware there is a minimum voltage are electronics design engineers of the kind that do all their work on paper (or a computer) and possibly don't even know how to solder. (I'm not 'talking them down' or anything. Physically working on circuits is just not their job.)

There isn't a minimum voltage in an absolute sense. Like everything else it's a tradeoff. There is negligible difference in lifespan caused by using (for example) 25 or 35V cap due to oxide wear, but more often the other factors in cap lifespan become more important like the surface area, ESR, higher ripple current tolerance and better filtration of that ripple (to the powered parts).

Would it interest you to know that when power supplies aren't built to penny-pinching stanards, more often industrial and expensive equipment supplies, it is far more common to see caps of (for example,) 35V rating for a 5V and 12V output? I've dealt with several (Power General for example).

Those folks don't end up in places like this forum very often because repair is not what they are interested in and so that kind of theory information rarely makes it's way into the knowledge pool here.

The primary limitation in choosing the best replacement part is lack of information (about circuit stability and current limits upstream of the cap), and lack of real-estate. Beyond these two, a higher capacitance and voltage is functionally superior and longer lasting for our purposes.

I'll elaborate more in another reply.

[999999999]

Re: What happens when you seriously increase capacitance?

Drop in capacitance using a cap at only 1/3rd the rated voltage is negligible, and 50V caps are used in 5V circuits, I would consider that the upper limit. These aren't circuits where an exact capacitance value is critical. We don't have to care much if the lower voltage causes a thinner oxide film because it's running at this lower voltage, not the higher voltage.

Equilibrium is reached and it would not account for the failure mentioned so long as there isn't an extreme (closer to an order of magnitude) difference between working and rated voltage. That failure would come from a defect, or if it were sitting uncharged for so long that it needed the oxide layer reformed before use, or if the bung had shifted and there was new oxide-less area (Like leads) exposed.

As for lesser voltage difference, it won't cause excess leakage current to any significant extent, because the oxide layer thickness (which effects dielectric function at the working voltage) still corresponds to the working voltage until there is a much larger difference than 1/3. One exception to this might be if the cap is a very poor brand and has poor construction or materials, but I feel we can rule these out as that would also tend to lead to premature failure if a closer match of working to rated voltage is used.

An exception is if one were to take that cap (cannibalize it off a used piece of equipment for reuse) and then place it in a different circuit having significantly higher working voltage without reforming the oxide layer (supply with gradual voltage increase if it's bad enough).

The supposed 1/3 rule flies out the window when looking in side many different pieces of electronic equipment. (Most?) manufacturers commonly use little 35-50V caps sprinkled all over equipment that uses these small lytics instead of tantalums, in circuits operating at or below 12V.

Whether capacitance increase matters depends a lot on the subcircuit. A bios does not have the effect PCBonz suggests, the PSU would turn off or fail to issue Pwr-Good for initilization, there won't ever be a case where the PSU had stayed on and yet not stabilized by the time the PROM had been read, bios expanded and then executed.

If a board did not meet Intel's design guide specs and ignores Power-Good, or the PSU is faulty (or deliberately non-adherant to ATX specs as with some really low end generics), in this case there could be a voltage problem and it would cause a lockup pre-bios-initialization, but any problem occuring after that point would tend to be instability not caused, maybe even partially lessened by the increased capacitance.

You can generally use a 35V cap anywhere in a computer except of course the few areas where working voltage > 35V, except that to have adequate capacitance it may easily be too large to fit in the alotted space. Real-estate is precious in many PSU or most motherboards, there is this addt'l incentive to use caps with only enough voltage margin besides the lower cost.

If it is not too large to fit without lowering the capacitance value, and all else is equal (same family of cap - temp rating, etc), in the circuits most prone to failure (output of switching supply circuits) the higher voltage rated cap will last substantially longer, unless the circuit is so undemanding that either cap alternative would have dried out from quite a few years of aging. The oxide layer factor just isn't so significant because these are uses where the cap wouldn't be expected to last decades.

I urge anyone who disagrees to just try it. My philosophy is simple: Forget about the one chance in 5,000 where there is some hypothetical possibility of a resonant loop, or where going up more than one notch in capacitance would cause a problem, until you actually see that causing a problem in a specific circuit. In practice these things are very rare in our uses, and the far more important factor is that the ESR of the larger part (moderately higher voltage part in same family, etc) is significantly lower, the lifespan significantly higher. These very desirable attributes are offset by the larger size and higher cost.

I'm not suggesting to use 50V caps in <12V circuits but 35V? I've used them and 25V when out of <=16V caps and never a problem (years later). Switching power supply manufacturers like Power General routinely do this as well. Right now I'm looking at a DC-DC supply with 5V/12V output with all UCC LXV 35V out the output LC filter. The main reason why a manufacturer will only use a little above the working voltage is that in a given cost and size, increasing capacitance is a more important goal up to the reasonable limit of the supply. They're not dealing with the same locational supply issues or sometimes unreasonable small-order costs a DIY repairer might encounter when repairing only a couple of PSU. Further, why would you trust the original design spec when you're having to repair it?

There is one notible exception in that when a cap's oxide layer is reduced through lower working voltage than rated, it becomes more susceptible to voltage surge -> leakage current increases. This is less likely in our uses because these circuits have at least moderately fast, controllers and feedback. Any peak ripple effects are regularly recurring events that of not of great magnitude, but do also build a correspondingly thicker oxide layer.

Don't be afraid to experiment with this and you'll see for yourself, there's more to gain than lose if you have the space to fit a larger cap and that by increasing both voltage and capacitance within reason. Those who worry over it, I don't think they try it in switching supply circuits. I've never had a problem doing so.

[gonzo0815]

Re: What happens when you seriously increase capacitance?

I agree to that, minimum working voltage is not that critical, and a conservative design practice for analog circuits would never be a 16v part at a 12v location.

Not quite sure, but IMHO even this topic was already mentioned and discuses here, so i think those veterans of us are well aware of those issues and implications.
But instead of explaining all this repeatedly in all the details, there are some rule of thumbs, which are for many people all they need or want to know and which doe well in real world apps.


The problem with increasing capacitance though is a real one, but most PC components are designed for a lot more, thus there is in many cases a large upper margin possible.
But for those, who only want to fix something, i think the on value bigger is a good rule of thumb.

[PCBONEZ]

Re: What happens when you seriously increase capacitance?

Many 9's

My I ask where you were formally trained in Electronics?
Any Electronics program I've heard of teaches Electronics Techs to replace components with the same value when ever possible.
When it's not possible you draw out the circuit and do the math.
(Which is beyond the knowledge level of many people you are making these recommendations to.)
- I doubt that just sticking whatever in there and crossing your fingers is taught by any professional organization anywhere.

>> 'supposed' << = assumed

So lets change >> Because techs are 'supposed' to use the original value caps <<
to
"Because techs are 'assumed' to be responsible and to follow good engineering practice by using the original value caps"

Now do you understand?

~~~

The discussion here is about low ESR capacitors used to remove ripple voltages in DC power supply circuits.
- All of your counter examples are out of that range.
Capacitors used for blocking, coupling, to absorb an inductive kick, or that are not polarized are not the same kind of capacitor, don't have the same chemistry, and either don't see ripple voltage because of the application or because of their chemistry (or uF) they don't have the same problems.

Even your PSU examples are the wrong kind of cap.
- If you will refer to a schematic of an ATX PSU you will see the only high volt caps after the last transformer are used for feedback to the IC in the control circuit. Those don't have anything to do with removing ripple.
- Now if you will now look at those same caps inside a PSU you will see they don't have vents. They don't have vents because they use a different electrolyte because they are built for a different application than those we are talking about.
- You will also notice that all the low ESR caps (with vents) in that last stage are 16v or less.
- The higher voltage caps with vents are before that stage and ARE on higher voltages.
(Unless maybe you are looking at some cheapie PSU where they slapped 'whatever' in there.)

~~~

I found the formula (math) used to calculate the operating life of aluminum electrolytic capacitors.
I don't want to explain the whole thing so I'll just show the part I'm talking about.

Mv = Va/Vr
Va = Applied Voltage
Vr = Rated Voltage

Operating Life = (the other variables) x Mv

So as Va goes down the life of the capacitor is shortened.

- End of story.

~~~

1/3 is the value I learned and have heard in other places. (Including in this forum from someone else.)
- As I said I don't know where it came from or what it was based on.
I worked on equipment that could kill people if it failed so I may have been taugh' an over conservative margin.
- But thanks. I'm gonna stick with it.. It works, it's obviously safe, and I don't like fixing things twice.

~~~

(Related)
There is some discussion now that the methods of endurance testing low ESR caps needs to be changed from testing them at their max voltage to testing them at some minimum voltage. (Some % below the rated voltage.) Testing them at max voltage builds the oxide layer and so is reducing the stress the test puts on the cap. (Extending it's life.) The test is intended to test the caps in their worse operating condition. Max voltage is not their worst operating condition, minimum voltage is.
--- Remember these tests were created before low ESR caps even existed.

~~~

My discussion about POST and the BIOS is exactly correct in the situation I was think of when I wrote it.
I wasn't talking about the PSU being stable I was talking about voltages on the motherboard being stable.
The CPU is not the only thing on a motherboard that gets power through a DC-DC converter.
Stick a huge cap in one of those other outputs and the system may not start.
One of the first POST checks is for video.
A video card slot may well have a DC-DC converter and a video card that's not powered up yet would halt POST.
Same-same with memory slots if they aren't powered off vsb.
---
The results of using a cap too far out of range would be the same (or nearly) as having a bad caps with certain failures.
I've seen bad cap motherboards fail to post and the PSU stayed on.
~~
Anyone else?

~~~

Experimentation is fine... In fact it's fun... Good way to learn new things.
But if you are going to do it you must accept the fact that you may completely destroy what you are working on.
-
Not everyone reading these forums is willing to (or can afford to) write-off a motherboard.
Since you don't know who will read this, blanket statements and recommendations to 'experiment' with or change component values far past the original should be written with that in mind and followed with some kind of warning as a disclaimer.

~~~

Back to the poster's original question.

Assuming the 4700 and 1000 have the same operating lifespan.
(Which is reasonable if you stay in the same series of caps.)

We'll say the Va is 3.3v
(Common for a 6.3v.)

The Mv for the 6.3 would be 0.524
The Mv for the 25v would be 0.132

The life of the 25v would be about 25% of the life of the 6.3v

Many 9's and a few others seem to think that's an insignificant difference.
I suppose that's a personal decision.

~~~

[999999999]

Re: What happens when you seriously increase capacitance?

"Because techs are 'assumed' to be responsible and to follow good engineering practice by using the original value caps"

Now do you understand?

I understand how you are drawing your conclusion, but I feel you are only thinking in idealistic terms, not realistic terms.

In reality, good enngineering practices would prevent most of the cap failures we see, but rather you are dealing with a unit that has failed in one of it's critical parameters. Rarely a cap is purely defective, but more often it is inappropriate for the environment and/or subcircuit it is placed in.

The reality is, it isn't hard for an engineer to spec out a good PSU, the harder part is making it all fit in available space at the target price, and still remain as stable and long-lived as possible. Most often when we do a repair, we are countering the original flaw, else potentially reproducing the same scenario.

The discussion here is about low ESR capacitors used to remove ripple voltages in DC power supply circuits.
- All of your counter examples are out of that range.
Capacitors used for blocking, coupling, to absorb an inductive kick, or that are not polarized are not the same kind of capacitor, don't have the same chemistry, and either don't see ripple voltage because of the application or because of their chemistry (or uF) they don't have the same problems.

I'm taking about the same circuits, and same cap family. Entirely applicable.

Even your PSU examples are the wrong kind of cap.

No, I never suggested using an inappropriate kind of cap. Same make/model/series just the larger value so long as it fits, without going to extremes.

- If you will refer to a schematic of an ATX PSU you will see the only high volt caps after the last transformer are used for feedback to the IC in the control circuit. Those don't have anything to do with removing ripple.

What you just wrote doesn't make sense. No cap is "used for feedback". It may be that some relatively small value caps reduce ripple in the feedback but these aren't normally going to fail. It may be that the controller IC uses a very small value film cap to set frequency but this also is (even more) unlikely to fail.

- Now if you will now look at those same caps inside a PSU you will see they don't have vents. They don't have vents because they use a different electrolyte because they are built for a different application than those we are talking about.

You're going to have to refer to a specific PSU and specific caps to qualify this claim. I suspect it is non-applicable to the discussion about the typical LC subcircuit caps failing. Further, caps in the 50V range as I'd previously mentioned, are also susceptible to failure if subjected to as much ripple current.

- You will also notice that all the low ESR caps (with vents) in that last stage are 16v or less.

Yes, this is where cap failures are most common and where PCB real-estate and budget dictates the part used. It is no proof of a moderately higher voltage part not working, but given the space (diameter) constraint, it would be undesirable to reduce capacitance so for practical reasons the best match may be the of similar capacitance and voltage, rather than the reason you claimed.

- The higher voltage caps with vents are before that stage and ARE on higher voltages.
(Unless maybe you are looking at some cheapie PSU where they slapped 'whatever' in there.)

No, that stage is practically always using caps with vents.

I found the formula (math) used to calculate the operating life of aluminum electrolytic capacitors.
I don't want to explain the whole thing so I'll just show the part I'm talking about.

Mv = Va/Vr
Va = Applied Voltage
Vr = Rated Voltage

Operating Life = (the other variables) x Mv

So as Va goes down the life of the capacitor is shortened.

So long as the cap rated voltage is not extremely different, the difference in lifespan from this factor is small, often negligible by itself, but there are other factors that mitigate and supercede this such as lower ESR, better smoothing, better heat dissipation. Those factors killed the original cap in many cases, and are the factors that should be addressed when choosing a replacement part for a longer tour of duty than the first, failed part saw.

1/3 is the value I learned and have heard in other places. (Including in this forum from someone else.)
- As I said I don't know where it came from or what it was based on.
I worked on equipment that could kill people if it failed so I may have been taugh' an over conservative margin.
- But thanks. I'm gonna stick with it.. It works, it's obviously safe, and I don't like fixing things twice.

You've overgeneralized and used hearsay. The PSU or motherboard is more likely to fail, sooner, using your strategy. Remember I'm not talking about some crazy substitution like 100V higher ESR cap where there was low ESR 5V cap previously, rather we were talking about the issues in using a 35V cap if all else were equal, if it were otherwise a suitable part. Key in addressing and preventing recurring failures is to recognize why the failure occurred.

It's certainly possible that one only needs to use a better brand and same value of cap to replace the one that failed, that the repair would produce an acceptible remaining lifespan for the unit, but that does not necessarily make it problematic to use another value, depending on the specifics.

(Related)
There is some discussion now that the methods of endurance testing low ESR caps needs to be changed from testing them at their max voltage to testing them at some minimum voltage. (Some % below the rated voltage.) Testing them at max voltage builds the oxide layer and so is reducing the stress the test puts on the cap. (Extending it's life.)

No matter how many times you repeat it, again you overlook that the minimal effect of this is overshadowed by the other reason the cap failed. Let's face it, the majority of caps that failed did adhere to your idea about having working voltage closer to rated voltage. It was the OTHER parameters that were insufficient, not voltage rating.

[999999999]

Re: What happens when you seriously increase capacitance?

My discussion about POST and the BIOS is exactly correct in the situation I was think of when I wrote it.
I wasn't talking about the PSU being stable I was talking about voltages on the motherboard being stable.

Yes, and you're wrong. The voltages have stabilized to an acceptible level simultaneous to, or at least within ms after Power-Good initiates system initialization, before the PROM is read, before bios is decompressed, before bios is executed at all.

The CPU is not the only thing on a motherboard that gets power through a DC-DC converter.
Stick a huge cap in one of those other outputs and the system may not start.

Yes, if the total capacitance is raised quite a lot, but that excessive measure is no argument against lesser values that work fine. You are writing this as if you made some general assumption and have not tried it. Theory is not fact, you must use experimentation next.

One of the first POST checks is for video.
A video card slot may well have a DC-DC converter and a video card that's not powered up yet would halt POST.
Same-same with memory slots if they aren't powered off vsb.

The results of using a cap too far out of range would be the same (or nearly) as having a bad caps with certain failures.

If by result you mean system wouldn't work, yes, but key is your vague comment about "too far out of range". You have not demonstrated that anything above same value or next notch up is too far out of range, and in fact others have used caps more than one notch up and it was not too far out of range. So far, all evidence is that the theory about matching value or next-up rating is wrong.

I've seen bad cap motherboards fail to post and the PSU stayed on.
~~
Anyone else?

Scientific method is what you lack. Anyone can memorize an overgeneralized idea but that does not make it appropriate to try to randomly apply the idea to specific scenarios without testing, and consideration of other variables and mitigating factors.

I don't just write this stuff, I do it.

[starfury1]

Re: What happens when you seriously increase capacitance?

well I think the basic theory is agreed upon just some difference of opinion on the cause Effects, both in the reality and theoretical.
which is beyond my realm of experience so finer detail of that I can't really comment on, cause I have no real idea or practical experience of dealing with at that level of operation.
( on some aspects I suppose you may have to agree to disagree thought, we are all in title to our opinions based on our sphere of experience and theoretical understanding)

I guess its a matter of selecting the right type of capacitor for the job.

Could the situation be improved upon from the manufactured unit as supplied, this is most certainly the case.

Regardless of what the design engineers intended the bean counters and stock availability and costs all go toward delivering something other then what they may have intended.

It is true too that fully understanding the circuit you are in a much better position to decide how far you can change something to improve the situation of circuit operation and reliability.

Since mostly these days, a lot of the information is not known (or finer points of it)
it does make it a lot harder.

True too that the final analysis is in the practical reality of doing it.

hypothetically speaking;

one aspect of the original post is what long term effect may arise?
it may work fine.

But like sticking General Purpose caps in a VRM circuit it might work fine too...but not for long.
(or the 100A nail in the fuse box)

So is a possible failure going to occur cause of increased inrush current via the supply rails to that cap.

I guess in the days of Valve and Transistor circuits a lot of the basic designs were known and used and techs could at that point understand and design a lot of them them selfs where as these days with the complexity circuits and use of VLSI chips the design aspect is somewhat unknown.
Probably a lot of a parameters have to be taken into account with regards to the chips used.

So these days its probably safer especially if you don't know the basic parameters of the circuit to stick close to the original device values
This does not mean you cant improve on it by at lest using a decent quality device (capacitor) in its place.

There is a but here...and that is sometimes some aspect may cause issues with the circuit, granted its rare but it does happen.

There is a review over at jonny G's on 2 psu's that had been recapped with quality caps which resulted in a somewhat worsening of specs rather the improvement.

specific page here

Article Here

All I can guess at (and no design engineer here) is, I think from memory the output stage operates as a tuned circuit
(think its PI filtering thing or some variation of it)
so it was optimized for the original junk caps.

Thats pure guess work

Anyway the short answer is I wouldn't do what was stated in the original post that started this thread.

It was asked as a hypothetical question.

So to me it really depends on what they were used for and what possible circuit variations may exist for it, plus the construction of the pcb for it.

But one thing that I would be concerned about, even if it did work without issues is possible long term upstream damage from inrush current

I think the idea was to have a "fatter" bucket for the RAM to operate from
not that I can see a problem conceptually with that...within reason
(Thats pure guess work again)

Well just my thoughts

Cheers