Recaping how to?

[ROCKETF1]

What does a re-cap involve? I mean how do you do it? Got a couple old monitors around and lot of other sht that might need it.

[AndreiArgeanu]

Take an old broken capacitor off, replace it with a new one, done.

[minibois]

1 minute ago, ROCKETF1 said:

What does a re-cap involve? I mean how do you do it? Got a couple old monitors around and lot of other sht that might need it.

1. find suitable new caps

2. remove old caps ("twisting off" vs. "desoldering" has been a debate for years)

3. solder back in the new caps

 

Of course when finding suitable new caps, you need to take into account the physical size, in addition to the right electronic specifications.

[ROCKETF1]

Are failed cheap cap on a device more obvious. I'm recalling this from a thing read many years ago. Which is really more to my point

[Kilrah]

Often they either bulge or leak.

 

Sometimes not, and you can't know if they're dead without removing and measuring, at which point you might as well replace.

 

If you find obviously bad ones it's good to replace all the others of the same brand/type even if they aren't visibly bad.

[Hackentosher]

1 hour ago, Kilrah said:

Often they either bulge or leak.

If you find obviously bad ones it's good to replace all the others of the same brand/type even if they aren't visibly bad.

100% 

1 hour ago, Kilrah said:

Sometimes not, and you can't know if they're dead without removing and measuring, at which point you might as well replace.

I disagree here. I don't know about you, but I can hold two multimeter probes to a capacitor a lot faster than I can replace a cap. I think if you have the tools to check, it's worth your time to check if a cap is good before blindly replacing all of them.

[Unimportant]

1 hour ago, Hackentosher said:

I disagree here. I don't know about you, but I can hold two multimeter probes to a capacitor a lot faster than I can replace a cap. I think if you have the tools to check, it's worth your time to check if a cap is good before blindly replacing all of them.

Both ESR and capacitance measurements are influenced by any circuits in parallel to the capacitor under test, so it's best to remove them and measure them out of circuit. He's right in that if you've removed them anyway why not just replace them?

[Hackentosher]

2 minutes ago, Unimportant said:

Both ESR and capacitance measurements are influenced by any circuits in parallel to the capacitor under test, so it's best to remove them and measure them out of circuit. He's right in that if you've removed them anyway why not just replace them?

Yeah okay you got me there, I concede. The caps in question are almost certainly filter caps paralleled with some other caps for all of that input damping, so you would have to remove them from the circuit to properly measure. I was more thinking about individual caps that were not connected to common power rails, which would make a capacitance test more useful. OP almost certainly doesn't have the schematics for these circuits in question so there's virtually no way to know what each cap is connected to without a lot of extra work. 

[ROCKETF1]

since all i have is a tiny electric kit and an iron is size matching good enough?

[Kilrah]

What do you mean, getting one physically the same size and hoping for the best on the other characteristics? No, not at all. 

[artuc]

If recapping.

 

You need to measure the case size, diameter and height. Also check the lead spacing/pitch. 

 

Then also note down what will physically fit, sometimes there may be space for a larger part or multiple configurations, such as multiple footprint choices for say 1 large capacitor or a bank of smaller ones. Take note vertically how much room you have and how high other components are.

 

Counter-intuitively some dodgy caps are smaller than good quality ones.

 

Then note down the electrical characteristics.

 

Capacitance: Try and match or go slightly higher. Typical power bulk capacitors are not hugely critical on the value if it's similar but aim for as close as possible. Some capacitors in say a filter circuit can have very specific values and should not really be strayed from.

 

Voltage: Match or go higher, never go lower. Higher voltage rating will get bigger physically.

 

Manufacturer, Series, Temp: This can be useful for something to at least match or have an idea of specs. Like a low ESR 105C series.

 

Eaisest thing would then to be go to say Digikey/Mouser or similar and use the filter to see what you come up with, try and get the same but better brand and series and what will fit  just play with the filters.

 

I'd start with sizes that will fit and work your way back.

 

Specific knowledge of which manufacturers and series are good comes down to doing some research.

 

My go to locally is Panasonic FR/FM/FC in that order. FC often had more size variety and better availability for slightly worse specs but still very good caps.

 

If you see this chart it sums up their series.

 

https://industrial.panasonic.com/ww/products/capacitors/aluminum-capacitors/aluminum-cap-lead/lineup

 

Hopes this helps.

 

 

[Unimportant]

3 hours ago, artuc said:

1 large capacitor or a bank of smaller ones

I'd stick with the configuration as it was originally. More often then not, if a modern DC-DC converter has a bank of capacitors rather then 1 large one it is done to reduce ESR and/or spread ripple current and power dissipation across multiple capacitors, not because they need a certain amount of microfarads. In fact, at the high frequencies modern converters switch at you theoretically only need a very small amount of capacitance, in practice much more has to be used to counter above mentioned non-ideal properties.

[ROCKETF1]

ok I wanted a little hold handing and got an avalance of info. I like it

[artuc]

14 hours ago, Unimportant said:

I'd stick with the configuration as it was originally. More often then not...

Definitely. I do recall multiple smaller capacitors in parallel tended to be better than a big single cap, IIRC in specs like ESR and ripple as you've said but also could often then also get more choice of parts, flexibility on sizes and better board layout etc.

 

Just meant that if you had a board designed for multiple footprints you do have that flexibility of multiple possible options if needed.

[mariushm]

A multimeter is not enough to check capacitors. A multimeter only measures the capacitance with a low frequency signal.

 

In dc-dc converter circuits (power supplies, monitor power supplies, VRMs - switching converters that power the cpu - ) the ESR is a very important property of a capacitor. Current ripple value is also important.

You can measure this ESR value with a LCR meter that has ESR measurement function or with specially made "esr meter", most of these "esr meters" don't actually measure the ESR but rather impedance, though at high switching frequencies common in switching power supplies the ESR is more or less equal to the impedance. 

 

Capacitors are often connected in parallel in circuits, or you have an electrolytic capacitor in parallel with a ceramic capacitor, because they both have different electric properties, and each filters bits the other one can't or doesn't do it well, so they work together to produce a better result.

Also, sometimes there's other components in parallel or in series, like for example a high value resistor in parallel to the capacitor which is there for example to discharge the capacitors when the power supply is turned off. 

However, a ceramic capacitor's very low ESR will affect the reading, if one is in parallel with an electrolytic capacitor, so your measurement will be wrong. 

 

Because of this, any capacitor you test should be either desoldered, or at least one lead of the capacitor should be lifted off the circuit board.

 

So... you look at the capacitors you suspect are bad, or they're degraded and affecting the circuit. If they're swollen or the tops are cracked, or some substance is leaking from the base or top, they must be replaced.

Otherwise, you need to test with an esr meter / lcr meter with esr option or just replace if it's cheaper to mass replace.

 

You look on the label of the capacitor and figure out the voltage rating, uF value and the manufacturer and series. Unless some really crappy chinese caps, the capacitor will have the series (usually 2-4 letter) and you'll be able to find a datasheet for it.

 

For example, let's take a random motherboard with electrolytic capacitors and let's say you suspect one or several capacitors closest to cpu socket is going bad.

With some experience you would be able to see that power (12v) comes in through that 4 pin connector, and that 3 phase dc-dc converter converts 12v down to the voltage the cpu needs (on this old motherboard, maybe 1.5v or something like that).

 

So power comes into the bulk capacitors (light blue), then each phase driver (pink) pumps energy through the hi-side mosfet (orange) and the lo-side mosfets (white) and with the help of the inductors (green), you get the low voltage in the output capacitors (dark blue)

The input capacitors will have to be rated for at least 16v because they filter 12v, and they're all in parallel, so measuring one in circuit will not work, because you'll measure a higher capacity, and you'll measure a lower ESR (2 identical capacitors in paralel = half the esr measurement)

You can see there's room for extra hi-side mosfets but Asus determined it wasn't needed.

The output capacitors are two per phase, six in total, and you can see there's empty spaces for extra capacitors.  Again, if you measure a single capacitor, you'll get a bad measurement, because those six capacitors are in parallel. You really need to desolder at least a lead from the capacitor and test it like that.

Now you can see on the picture on the right, that those capacitors by the socket are 1500 uF and they're rated for 6.3v and the letters KZG are written on them ... this tells me the capacitors are made by Nippon / United Chemi Con and they're KZG series, so i can easily get the datasheet here:  http://www.nfjapan.com/datasheet/KZG.pdf

So you can see there, it says "super low" esr / impedance, designed for motherboards , blah blah, you know you got the right datasheet.

Now you scroll down to the tables, and you see there : 1500uF 6.3v , 10x12.5, impedance at 100kHz = 0.26 ohm (or 26 mohm) and 1540 mA ripple current:

The actual diameter may be a size smaller, as motherboard and power supply manufacturers have the purchasing power to order custom batches which may be smaller diameter but taller.

So now you have the ESR (0.026) and ripple current (1540 mA) .... if you don't know how the circuit works, you want your replacement capacitor to fit, so you must have the diameter small enough for the capacitors to fit there, and you want them to be low height enough to not block coolers and fans, and you want the ESR to be lower or equal, and the current ripple must be as close as possible or bigger.  You don't want to replace with way lower esr than existing esr, as some dc-dc converter circuits will be unstable.

 

Now in this particular instance, the output capacitance is 6 capacitors x 1500uF = 9000uF ... the motherboard really doesn't need that much, it can probably work with as little as 3-4000uF. The problem is that the motherboard manufacturer was constrained by the height (couldn't put tall capacitors because big heatsinks would be blocked) and these small diameter lowish height capacitors don't have very low ESR.

The circuit actually requires much lower ESR than what was possible at that time, so they resorted to having 2 capacitors in parallel on each stage, which means that 0.026 ohm ESR becomes around 0.013 ohm

You can also see they left footprints for a third capacitor on at least 2 phases ... they probably thought maybe they can't source 1500uF 6.3v capacitors and may have to resort to using 820uF or 1200uF capacitors which have higher ESR... you can see in the chart below that 820uF capacitors had 0.036 ohm ... but if you parallel 3 of them, you get the desired 0.013 ohm ESR.

If you no longer find such very low esr electrolytic capacitors, you could go with lower capacitance and close enough esr, but populate the empty spots.

Another limitation was that the manufacturing process didn't make it easy to make capacitors with lower voltage rating. The output voltage is typically under 2v, so if they were able to they would have used 3-4v rated capacitors.

Knowing this, now you could go and look at some polymer (solid) capacitors : https://www.nichicon.co.jp/english/products/fpcap_catalog.html

I don't want to go much lower than 0.026 ohm , so I'm not gonna choose a 6 mOhm esr solid capacitor, but I can go with something like the "standard" NS series (not even considered low esr by polymer capacitor standards) : https://www.nichicon.co.jp/english/products/pdf/2013fpcap_catalog_ns.pdf - so I could use 820uF 4v NS series polymer, which has around 10 mOhm ESR and 5500 mA current ripple. The 6 polymer capacitors in parallel will give enough capacitance, and the esr is much lower.

Same deal with the input capacitors ... though you can't go lower on voltage, so if switching to polymer capacitors, 270..330uF / 16v capacitors would probably be enough.

 

 

 

 

 

 

[Mark Kaine]

Not to hijack this topic,  but let's say I have a PS3 where I think a capacitor went bad,  let's also say I find the one that's defect by some visual signs... Isn't everything I need to know to replace it written on it? 

 

I remember I soldered a couple of capacitors where I used to work and they all had pretty distinct number codes on them,  otherwise we wouldn't even have known which are the right ones.  Though I could remember wrong? 

[mariushm]

You'll have the uF rating, you'll have the voltage rating and you'll probably have the series and you can determine from series the technical properties (as in determine who makes the series, and then download datasheet and get the important technical properties from datasheet)

 

As I said in above post, the ESR and the ripple current value matter. If you install an inadequate capacitor, the device MAY work but will fail again within weeks or months due to abusing that capacitor with work it was not designed for. 

For example, a dc-dc converter circuit may be designed to be stable with capacitors that have ESR between 5 mOhm and 40 mOhm and the circuit uses a capacitor with 20 mOhm ESR due to some constraints like height and diameter. You could randomly pick a capacitor that has 35 mOhm ESR which is far from the 20mOhm of the original capacitor, but it's still in the 5-40mOhm range the circuit was designed for.

BUT, the capacitor will degrade over time, so that ESR value will slowly go up as the capacitor is going through heating up (during use of device) and cooling down, so in a year or so it may be barely within 40mOhm. Also, capacitors are affected by temperature, so you may try to start the device one day when it's cold (let's say 5-10 degrees) and the dc-dc circuit will fail to start because the capacitor's ESR is higher due to the temperature. After around 10-20 minutes, as the capacitor warms up, its ESR may go down within the acceptable range where the circuit starts working. 

 

The opposite can happen as well. For example, there's a very common and very generic linear regulator chip called 1117 (you'll find it with various letters in front AMS1117, LM1117 etc) which requires an output capacitor with ESR between 0.1 ohm and 1 ohm on the output in order to be stable. 

You maybe see a swollen capacitor on the board (most likely due to being close to the chip which produced heat) and you'll jump to replacing it with a good quality low esr capacitor which may have 0.05 ohm ESR or even better.... and probably 80% of the time, the circuit will be stable when you test the fix.  But, the regulator will probably be unstable between certain temperature ranges or between certain current output ranges ... for example, when you start and the chip is cold, between 20c and 50c it will be stable, but it could glitch between 50c and 90c ONLY when current goes above 300mA.

 

So with such chips you'll often see manufacturers use a more "general purpose" capacitor, that's not designed for very low ESR, and you may look at the datasheet and say "oh, look at this shit capacitor, the manufacturer cheaped out and went for a 1000h @ 105c capacitor with 100mOhm ESR, no wonder it failed... I'll replace it with my fancy 8000h @ 105c capacitor with 10mOhm ESR "  ... no, don't do that. Replace with something more modern but with conservative technical properties... and also pay attention when replacing a capacitor like this with one rated for higher voltage (the bigger volume , the better the properties of a capacitor, so a capacitor that's 5 mm taller , same uF but higher voltage rating, may have better esr and current rating)

 

With the 1117 regulators, manufacturer will often simply add a 0.1-0.5 ohm resistor in series with a low esr capacitor (or a ceramic capacitor) just to keep that minimum resistance above 0.1 ohm at all times, while allowing cheaper or surface mount ceramic capacitors to be used. Adding a resistor can be cheaper than replacing that regulator with one that's 5-10 cents more expensive but doesn't become unstable.  

 

 

In lots of circuits especially vrms there's capacitors in parallel and people often make mistake of replacing only ones that are visibly swollen or leaking electrolyte... you kinda have to replace all in parallel even if those are not looking bad.  basically the other capacitors took over for the failing capacitor and were probably working for a long time above what they were rated for, they were "overworked", so even those those caps. are not leaking they're most likely at some amount damaged, just not visible yet. 

If you add a new capacitor, that new capacitor will be better than all the others and will get more of the "work" done, until it degrades to the level of the other capacitors. 

[Unimportant]

17 hours ago, Mark Kaine said:

Not to hijack this topic,  but let's say I have a PS3 where I think a capacitor went bad,  let's also say I find the one that's defect by some visual signs... Isn't everything I need to know to replace it written on it?

In the power supply or the motherboard? Because from what I can see in online pictures the PS3 motherboard does not contain any electrolytics. The other types of capacitors typically don't show visual signs of failure (except for tantalum, which can be overly visual on failure).

[artuc]

Also @mariushm has touched on polymer capacitors.

 

Aluminium polymer caps are much more reliable and have longer lifetimes than wet electrolytics.

 

You can often pre-emptively recap electrolytics if there are symptoms that may suggest it beyond obvious bulging, venting or exploding or just when restoring old gear due to long term degredation.

 

Polymers generally you'd want to only replace if known faulty, just mentioning it as it is possible to confuse the types, so you should check the series to confirm what you're dealing with as mariushm has already recommended.

 

AFAIK polymers always tend to be in the raw aluminium can finish, whereas electrolytics tend to be similar in SMD but otherwise will have a wrap and vent on top in through hole flavours.

 

 

[Unimportant]

@artuc He also mentioned a DC-DC converter can rely on the capacitor ESR for loop stability. (The ESR introduces a zero which provides phase boost at higher frequencies). Polymers have much lower ESR then plain electrolytics. The result might not be stable or have such low phase margin that it rings and overshoots.

 

Edit: nvm, I tought you meant replacing plain electrolytics with polymers.