Never use lower raited caps.

^ this higer voltage np not lower

Jade Falcon wrote:

Never use lower raited caps.

Richo wrote:

^ this higer voltage np not lower

That's what I thought but I wasn't sure if solid caps were different somehow. People also claim these motherboards really only need 16v.

Better safe then sorry. They used 330uf 25v caps for a reason. Sometimes they may use over rated caps to save money or increase the life span of a product. But its better to just use a direct replacement unless if you really know what your doing.

Solid caps rating wise arnt any different, they're more reliable and tend to be better overall. I.E. lower esr, ripple smaller and so on. They can however be labeled differently do to their smaller size.

While sticking with 25v caps is definitely sound advice, 16v caps would be completely fine.

Gigabyte probably excessively de-rated cheap 25v capacitors in an effort to extend their lifetime as much as possible. Modern capacitor quality has progressed dramatically since then, so any respected brand of low ESR capacitor in 16v flavor will do just fine. Even a decent modern general purpose cap is probably better than what came on that board.

Case in point: I've got a stack of hot-running rack-mount server motherboards with over 5-years (40,000 hours) on them that are all populated with 16v electrolytic caps on the 12-volt power rails not a singe one is bulged. I even reused some of them to recap the 12v/5v SMPS (which had no-name 25v caps that failed) in my Logical Devices EPROM programmer because they tested healthy on the ESR meter--in fact still exceeding their rated specifications.

16V poly caps will be OK, all they'll have to endure will be +12V from the PSU.

In the particular case of replacing CPU VRM lythics with polys you can even lower the capacitance and go away with it most of the time. But better if you can replace with polys of same or higher uF and lower ESR.

The 10mm Nichicons that Phil recommends are same 330uF, and at 7mΩ lower ESR than the original GSC crap. Much better stuff.

Man there is so much BS and misinformation surrounding capacitors. For some reason everyone seems to have an opinion 😀 I urge you to do your own research. Heaps of sites have looked into this with great detail and explain it well so it's best to form your own opinion 😀

A polymod though is a different animal, here you should know a bit more than when using electrolytic capacitors.

Thinking about it, as a matter of fact in an ancient mobo like the GA-5AX after tracing them you'd probably find that most caps are being fed +5V tops. So 10V or even 6.3V rated caps would work OK, in some positions at least.

The 25V rating of the originals tells me that Gigabyte just used general purpose high ESR crap back in the day.

As someone who has a degree in electrical engineering, I think my opinion is prudent.

In 95% of cases, you will never want to put lower voltage caps in a system. A 12v cap is rated for 12v MAXIMUM. Meaning it will have a shorter life when constantly used at that voltage. 16v is perfectly acceptable for 12v nominal systems, TODAY. We have very good caps now. The 2nd reason to use a 25v cap is HEAT. Todays caps dont have the problem as much, thanks to lower ESR and better processes.

watch this: https://www.youtube.com/watch?v=wwANKw36Mjw

Yes, you can get away with using lower voltage caps in light-duty devices, but it is NOT recommended. They used 25v for a reason, there is NO PRACTICAL ADVANTAGE to replace a 25v one with a 16v.

@luckybob: As an EE I'm sure you're familiar with the Arrhenius Equation and how it relates to estimating the operational life expectancy of an electrolytic capacitor. In short, a capacitors life expectancy is DIRECTLY related to it's temperature and for every 10C drop in temperature, life expectancy doubles. Applied voltage has practically nothing to do with with it.

Indeed, applied voltage is also factored into it but for the most part unless operated far outside the rated voltage or subjected to severe ripple, the DCL is so small the resultant dissipation is not even worth factoring. Aluminum electrolytic capacitors require very little voltage de-rating, even when operated at high temperatures and ripple current a 10-15% de-rating is sufficient except for military and space applications.

I don't know how the Dave Jones video contributes to this conversation...all he proves is that by paralleling capacitors to aggregate their ESR ratings you can reduce the amount of energy they individually dissipate. What does that have to do with this?

Anyway, you're right there's no practical reason for replacing the original capacitors with ones of a lower voltage rating--the OP can simply do the same and be happy. He probably already has. But I believe there is value in understanding WHY doing so is unnecessary. Educated decisions and Free Thinking are good things, right?

Sure, it's easy to just triple the operational voltage and use that as a spec...but if an EE worked for me tried that without additional justification, I'd fire them. I could get a trained monkey to do that.

luckybob wrote:

As someone who has a degree in electrical engineering, I think my opinion is prudent. […]
Show full quote

As someone who has a degree in electrical engineering, I think my opinion is prudent.

In 95% of cases, you will never want to put lower voltage caps in a system. A 12v cap is rated for 12v MAXIMUM. Meaning it will have a shorter life when constantly used at that voltage. 16v is perfectly acceptable for 12v nominal systems, TODAY. We have very good caps now. The 2nd reason to use a 25v cap is HEAT. Todays caps dont have the problem as much, thanks to lower ESR and better processes.

watch this: https://www.youtube.com/watch?v=wwANKw36Mjw

Yes, you can get away with using lower voltage caps in light-duty devices, but it is NOT recommended. They used 25v for a reason, there is NO PRACTICAL ADVANTAGE to replace a 25v one with a 16v.

the point I was trying to make was; Don't change what was in the board, they probably did it for a reason, just replace like for like. Granted if you have a bit of education, formal or otherwise, then you can change the design to fit your needs.

I would imagine, gigabyte had a good, cheap deal on 25v caps and using them caused no issues. overkill? most likely.

You don't need to be a MIT graduate to solve this puzzle. Gigabyte used general purpose high-ESR 25V GSC craps just because they could (ancient mobos were more tolerant of crappy caps than modern ones). The reason: they were cheaper than better quality low-ESR 10V stuff = more profit from each mobo sold.

The only possible voltages (except for the ones with valve audio) on a motherboard are:
12V - for which 16V caps of suitable quality are acceptable, if they fitted 25V, it was for ESR rather than voltage, and they had the space to
5V - which may be covered by 10V or 6.3V
3.3V - which 6.3V or 4V covers
core voltage - probably 2.5V standard electrolytic

Size & price increase significantly with voltage on solid capacitors, so it's rare to allow anything over what is actually needed. The choice of components being dictated by price and sourcing considerations in bulk... a few pence or cents saved per capacitor adds up!

Matth79 wrote:

... if they fitted 25V, it was for ESR rather than voltage ...

For the same can size (diameter x height), the higher the voltage rating, the lower the capacitance. ESR remains more or less constant.

Example, Panasonic FC 10 x 25mm.

10V: 1500uF 0.045Ω

16V: 1200uF 0.045Ω

25V: 820uF 0.045Ω

35V: 560uF 0.045Ω

Example, Rubycon ZL 10 x 20mm.

10V: 1200uF 0.023Ω

16V: 1000uF 0.023Ω

25V: 680uF 0.023Ω

35V: 470uF 0.023Ω

I call penny-pinching by using general purpose caps that happened to be 25V.

Where did you all go to university for EE? I have graduate and undergraduate degrees in electrical engineering and do not ever recall a discussion on capacitor heat, lifetime, the impacts of ESR or capacitor function, nor an arrhenius equation. Thinking I must have forgotten, I checked the indexes of 6 of my circuit design books and there is no heading in the index for either of these items. I remember school focusing WAY too much on equations and overly complicated design and less about what employeer wants. I remember seeking employment after graduating and being asked only about topics which school did not address, e.g. what multi-layer PCB boards I had to design in school. Well, we didn't do PCB design in school, nor did we discuss trace optimisation techniques. All I could tell the guy was there could be issues with how close the traces were placed and issues with induced parallel trace capacitance with a frequency dependence. Later on, when in industry, we had PCB-specific guys, who's role was mainly to optimise the PCB layout. I was talking to them about this and they never did a format EE degree, but rather, did electronic tech courses at the local community college. It seems that most people at work that I discussed with who got the more hands-on circuit component type roles did trade school certificates. Companies dealing with electric motors seemed to prefer ME degrees because the EE's dealt way too much with transistor-based designs. In conclusion, I learned way more on-the-job than I ever did in school. I wish schools would taylor their EE programs based on current industry demand.

I know almost nothing about capacitor selection and ESR optimisastion and am constantly learning stuff school failed in. With that in mind, I would like to ask you capacitor pros what you think about using 100 uF, 220 uF, and 470 uF niobium oxide capacitors as replacements for electrolytic caps in motherboards. The ESR is not nearly as low as the values a poster just presented, but the spec sheet claims they are "low ESR" - 100 m-ohm, 400 m-ohm, 75 m-ohm respectively. I hate wearable components and products with a short shelf-life like electrolytic capacitors. For sake of argument, ignore the price of these guys.

These go up to 1000 uF, but then the rated voltage drops below 6.3 V.

http://www.digikey.com/product-detail/en/avx- … 64-1-ND/4561730
https://www.digikey.com/product-detail/en/avx … 829-1-ND/564861
https://www.digikey.com/product-detail/en/avx … 253-1-ND/930045

There's also up to 220 uF capacitors for 6.3 V, but I don't see an ESR listed in the table. e.g. https://www.digikey.com/product-detail/en/tai … 80-1-ND/4950534

I recently ordered a few of these higher capacitance niobium oxide capacitors for use with a switching voltage regulator. I've seen designs which use four 47 uF niobium oxide caps in parallel as well.

feipoa wrote:

... what you think about using 100 uF, 220 uF, and 470 uF niobium oxide capacitors as replacements for electrolytic caps in motherboards. The ESR is not nearly as low as the values a poster just presented, but the spec sheet claims they are "low ESR" - 100 m-ohm, 400 m-ohm, 75 m-ohm respectively ...

"Low ESR" is just a vague mostly meaningless marketing term (how much low is low?). IMHO 75~400 mΩ is way too high ESR for motherboard VRM lythics replacement. Also those are SMD, you'd need extra tinkering to replace thru hole lythics with them.

feipoa wrote:

... I hate wearable components and products with a short shelf-life like electrolytic capacitors ...

Polys are excellent replacement for lythics on motherboards.

Not so much on PSUs, where you still need lythics for at least the larger primary bulk caps (no polys of such uF & V), and where fully replacing lythics with polys on the secondary can trigger unwelcome effects.

Even this 2017 made in Japan $1000 engineering jewel still sports lythics on the secondary: http://www.jonnyguru.com/modules.php?name=NDR … Story5&reid=508

To maximize lythic longevity, use quality jap caps of some long-life series like Chemicon KY, Panasonic FR, Rubycon ZLH, etc.

TELVM wrote:

"Low ESR" is just a vague mostly meaningless marketing term (how much low is low?). IMHO 75~400 mΩ is way too high ESR for motherboard VRM lythics replacement. Also those are SMD, you'd need extra tinkering to replace thru hole lythics with them.

Motherboard VRM's from which vintage? Any vintage including 486? On many VRM upgrade modules which sport an AMD Am5x86-133 and Cyrix 5x86-120, they use tantalum polymer or niobium oxide capacitors, which have ESR of a similar range 70-400 m-ohm. See this image, download/file.php?id=32744&mode=view Originally, I was thinking that they added the four 47 uF SMD caps in parallel because either the 220 uF SMD cap was too expensive or unavailable at the time, but now I'm wondering if they did this to reduce the ESR. AVX Corp has 45, 70, 200, and 250 m-ohm ESR options for their 16 V tantalum polymer caps, but I don't really know what was available in 1996 when this module was produced.

The spacing of the 220 uF niobium oxide capacitors I have in hand contain the exact same lead spacing as many electrolytic capacitor lead spacings; thus I think tweaking would not be necessary if the SMD caps were ordered with their specific spacing in mind.

For the 100-1500 uF range, and with motherboards ranging from 386 to Tualatin, do you know for which vintage motherboards I can use cermaic, niobium oxide, tantalum polymer, or regular tantalums as a replacement? Which circuits on the motherboard could use which? Any general rules and caveats to watch out for?

From your commentary, it sounds like aluminum polymer caps are fine for just about everything non-PSU related?

Concerning the other comments concerning using lower voltage rated electrolytic caps - the PSU doesn't generate anything above 12 Vdc. It was commented that using larger voltage-rated electrolytic capacitors disipate heat more readily and thus will likely last longer than lower rated caps. Does this same rule apply to these solid dielectric caps (ceramic, aluminum polymer, tantalum, tantalum polymer, and niobium oxide)?

Lastly, I did a quick tally on digikey for solid dielectric caps and ESR. The Aluminum, Tantalum polymer, Tantalum, and Niobium Oxide caps are all specified as "low ESR".

Aluminum Polymer Capacitors the max capacitance for

16 V is 2200 uF | ESR = 12 m-ohm @ 100 KHz
6.3 V is 2700 uF | ESR = 8 m-ohm @ 100 KHz

Tantalum Polymer Capacitors the max capacitance for

16 V is 470 uF | ESR = 25 m-ohm @ 100 KHz
6.3 V is 1500 uF | ESR = 55 m-ohm @ 100 KHz

Tantalum Capacitor

16 V is 330 uF | ESR = 55 m-ohm @ 100 KHz
6.3 V is 1000 uF | ESR = 30 m-ohm @ 100 KHz

Niobium Oxide Capacitor

6.3 V is 470 uF | ESR = 75 m-ohm @ 100 KHz

Ceramic Capacitor
16 V is 470 uF, though there is an obsolete listing for 680 uF. No ESR spec. I assume it is high?

So the highest capacitance we can use non-electrolytic caps for is 2700 uF. This covers nearly all capacitance in my 386-to-Tualatin boards. There may be a strange 3300 uF in there. Do I need to alter any of the capacitance when changing the capacitor type?

From this list, it seems that tantalum polymer is the next closest in ESR to aluminum polymer.

Part of when I think is psychologically turning me off from aluminum polymer capacitors is that they are cylindrical, and my mind associates this with leaky/bulgy electrolytic capacitors. Also, my Macintosh SE/30 had what looks like aluminum polymer capacitors on it. A few years back, they leaked some corrosive substance and caused trace damage. Maybe they weren't aluminum polymer caps?

feipoa wrote:

Motherboard VRM's from which vintage? Any vintage including 486? On many VRM upgrade modules which sport an AMD Am5x86-133 and Cyrix 5x86-120, they use tantalum polymer or niobium oxide capacitors, which have ESR of a similar range 70-400 m-ohm. See this image, download/file.php?id=32744&mode=view Originally, I was thinking that they added the four 47 uF SMD caps in parallel because either the 220 uF SMD cap was too expensive or unavailable at the time, but now I'm wondering if they did this to reduce the ESR. AVX Corp has 45, 70, 200, and 250 m-ohm ESR options for their 16 V tantalum polymer caps, but I don't really know what was available in 1996 when this module was produced.

Don't forget 486-era VRMs are probably almost exclusively linear designs, with far more relaxed capacitor requirements. I didn't start seeing switch-mode VRMs appearing on motherboards until towards the end of the Socket 7 days.

In fact there's actually such a thing as having TOO low of an ESR when considering linear voltage regulator design-specifically those incorporating LDOs such as those common on 486 motherboards and upgrade modules. In these cases having too low of an ESR can prevent the LDO's output loop from establishing a sufficient amount of phase lead necessary for stable operation.

This capacitor discussion is sort of akin to my days building hot-rodded small-block V8s for myself and a few of my friends. A fallacy I commonly witnessed in that circle was, "if a bigger than stock camshaft is good, than a HUGE one must be even better!" It most certainly is not. Like a camshaft, a capacitor is just one component of a complex system. Rather than focusing on only a few favorable specifications, it's important to understand how ALL of its attributes effect the system as a whole and choose the best overall fit for that specific application.

feipoa wrote:

... From your commentary, it sounds like aluminum polymer caps are fine for just about everything non-PSU related? ...

Nope, just in the context you asked for, i.e. replacing lythics on motherboards.

feipoa wrote:

... It was commented that using larger voltage-rated electrolytic capacitors disipate heat more readily and thus will likely last longer than lower rated caps ...

IMHO it's larger can size, not larger voltage rating, what increases heat dissipation in lythics. That said, at the same capacitance a cap with higher voltage rating will most probably be of a larger can size, so will have better heat dissipation.

feipoa wrote:

... my Macintosh SE/30 had what looks like aluminum polymer capacitors on it. A few years back, they leaked some corrosive substance and caused trace damage. Maybe they weren't aluminum polymer caps? ...

Not familiar with the hardware but if they leak they're lythics. Beware some lythics may look like polys from a distance, like this Sacon FZ crap: