10

I am trying to replace a bad 820 µF, 6.3 V capacitor on my PC's motherboard. Right now I don't have a 820 µF capacitor in my home and also don't have enough time to buy a new capacitor. I have two 470 µF capacitors rated at 16 volts. The damaged capacitor is located near a RAM socket of the motherboard. Can I use a combination of two 470 µF capacitors that is almost 940 µF? Is it safe to use higher value capacitors in computer systems?

Dave Tweed
  • 168,369
  • 17
  • 228
  • 393
Farid-ur-Rahman
  • 1,011
  • 6
  • 22
  • 34

4 Answers4

12

Most likely the two 470 µF capacitors in parallel will be fine. At these values, they are almost certainly power supply caps. About the only drawback of more capacitance is higher inrush current when the supply comes up, but the 15% extra capacitance is very very unlikely to cause any problem whatsoever.

jusaca
  • 8,749
  • 4
  • 36
  • 58
Olin Lathrop
  • 310,974
  • 36
  • 428
  • 915
  • 1
    Also, those electrolytics are often spec'd at something like +80, -20% tolerance. In other words, the original capacitance could be way off what is marked and the circuit still works fine. – markrages Jan 03 '13 at 20:00
5

When it comes to capacitors, apart from capacity and voltage you also need to ensure that the ripple handling capability and ESR (equivalent series resistance) are well-matched.

In any application with high-frequency switching (i.e. DC/DC converters on motherboards) the capacitors will be exposed to high frequency (100s of kHz or higher) ripple current. A general-purpose capacitor (one that's only rated at 120Hz) will die a painful death within hours or days in this sort of application. There are capacitors branded as 'low ESR' and/or 'high ripple current' which are rated for operation at 100kHz - these are the sorts of capacitors you need to use in motherboard applications.

Another factor is operating temperature and rated life. A 105C capacitor can be swapped in where an 85C capacitor was found with little risk. The opposite case is not so safe to do.

Also don't substitute a 1000 hour rated capacitor in a circuit where a 5000 hour rated capacitor has failed.

Another issue is that if you drastically change the capacitance value and/or ESR by changing capacitors, you're changing the characteristic of the LC filter in the DC/DC converter. This can be a problem if the feedback compensation of the converter is aggressively tuned, as you may render the converter unstable by the capacitor change. If there are some ceramics in the circuit, their very low ESR tend to swamp out the influence of the electrolytics, making the swap somewhat safer.

All that being said, paralleling capacitors to match capacity is generally fine, so long as you're not substituting two GP caps (or a GP in parallel with a low-ESR), you're not going with a lower temperature part or drastically cutting the rated hours, and you're not radically changing the overall ESR of the circuit (usually not an issue if there are ceramic capacitors in parallel with the electrolytic caps).

Adam Lawrence
  • 32,921
  • 3
  • 58
  • 110
3

Capacitors should be matched for capacity (in farads), esr, and resonance. Changing this on a delicate electronic circuit like a motherboard is asking for trouble. Caps arn't chosen at random by the manufacturer after all. You can blow coils, fry chips, maybe even plague the other caps. The Voltage rating is a safe limit, so increasing that is fine.

Doom and gloom aside, if your not willing to wait for a replacement capacitor, and are willing to buy a new computer if you screw up, you want to wire the two capacitors in parallel. The slight difference in capacitance might not be enough to cause problems. What might cause problems is a high esr rating. Computer circuits arn't big fans of noisy caps and interference.

Certainly, it does depend on where in computer the cap comes from, but if it's near the ram, it's for the cpu, or the ram itself. Data corruption is possible. A power supply cap wouldn't be as risky.

Passerby
  • 72,580
  • 7
  • 90
  • 202
  • you'd be surprised about what manufacturers choose for components. They are not engineering gods :D – Gustavo Litovsky Jan 03 '13 at 16:51
  • @GustavoLitovsky well, yea, Dell cap plague problem for example, or fly by night china no-name manufacturers, but generally, the parts arn't chosen by reaching in a bin and picking what feels nice. At least they would stick close to the ic reference design. They might be cheap, but I assume they arn't completely incompetent. – Passerby Jan 03 '13 at 17:02
  • Usually guys like Dell will stick to the reference design from someone (unless they design their own which I doubt in most cases). Sometimes OEMs will go for the lower cost components just to lower the cost of the design, because reference designs usually tend to use high quality components that are known to work well. These things tend to fall through the cracks and into the lap of users. – Gustavo Litovsky Jan 03 '13 at 17:11
  • 6
    Large-value electrolytics are never used in tuned circuits, so this answer is unrealistically pessimistic. – markrages Jan 03 '13 at 20:01
1

Just letting you guys know, after seeing this site and real world testing it myself...

I'm not an electronics expert, but I have gotten an older (not functioning) ASUS mainboard where 15 caps of the same capacitance and voltage (6.3 V 820 µF) were (visibly) bad.

I decided to swap the caps from an even older mainboard with higher capacitance caps (6.3 V 1000 µF).

Installed OS and software, and after 2 days of using it, I think I can say it did the job.

Greenonline
  • 2,064
  • 7
  • 23
  • 38