Electrolyics and Impedance/ESR

[thisguy]

Hey everyone.
I have been reading though this forum for the past week. I'm very impressed by the community. I'm still very new so I apologize for my ignorance. I just bought my first solder station and I'm eager to learn, yet my formal education is limited to a college physics course. On to my question:

I always thought lower impedance was best until I came here. I've done my best to read through the FAQs and other posts but I'm still kinda confused . Generally what I got was that when replacing a capacitor on a motherboard, replacing a cap with a lower impedance is better. However, on a PSU, you should closely match the impedance with the original or else it will "detune" and you'll get more ripple. Can anyone elaborate on this?
What about replacing caps on other devices (not the power supply portion), should I always use replacements caps with lower impedance?

Does the max ripple current rating on a cap need to match or is fine as long as its better?

Also I've seen experienced members replace caps in power supplies where the impedance is not in spec with the original (ie original is 75mOhms and the replacement is 30mOhms). So I'm confused as to when this is ok to do.


Insight to any of my questions would be greatly appreciated! Thanks!

Best,
David

[digge]

Re: Electrolyics and Impedance/ESR

I always go by the rule that when replacing, get something that is as close as possible from the data sheets of the original cap. Too good capacitor can be bad depending on how the circuit is designed. However i always try to stay on the "known good" brands.

Someone else more knowledgeable might be able to give a more in detail description on where and why its suitable to change and not change from the original values.

[PCBONEZ]

Re: Electrolyics and Impedance/ESR

Many people (including some long time members) have gotten the idea that if low ESR is good then lower ESR is better and that's not always true.

If those folks that replaced PSU 75mOhm caps with 30mOhm caps checked the output on an Oscope they would probably find they have more ripple in the output rather than less. That is because the caps work in conjunction with coils [inductors] in the PSU output to be most efficient at the expected ripple frequency. If you change one without changing the other [meaning coil & cap] it is very possible to make it inefficient at the ripple frequency of concern.
- If you are going to change ESR drastically in a PSU then you should do all the math for the filter circuit and consider changing the inductor too.
- There are a few situations were there is no choice but to change ESR because replacement caps [of good brands] with equivalent specs simply don't exist.

The ripple current rating is simply how much ripple the cap can pass without overheating. It is very much like the watt rating on a resistor or the amp rating on a wire. Changing it won't affect the operating characteristics of the circuit like changing ESR can. But, making it too small could cause the cap itself to overheat and fail. [Same idea as using too small of wire for something.]

When it comes to motherboards you first need to realize that different caps on the board have different jobs and are in different kinds of circuits. What you can 'get away with' for some caps on a mobo is a no no with caps doing other jobs.
*In general* lowering ESR on mobo caps has proven safe and problems are very rare but that doesn't mean insanely lowering ESR is a good idea.
.

[thisguy]

Re: Electrolyics and Impedance/ESR

awesome! thanks for clearing that up for me. I'll do my best to stick as close as possible when working with PSUs. I'm not sure if its best to start a new thread, but I have a OCZ 700W gamexstream PSU that has 2 bulged slim (8mmX20mm) 2200uF 6.3V Capxon KF caps. The problem is that the data sheet for capxon KF caps don't have this size cap. They only have 10x20 and 10x25, with impedance of 90 and 95 mOhm respectively. Is is safe to assume the impedance is slightly less than these as the pattern suggests? perhaps 80-90 mOhm?

The only 2200uF 6.3v 8x20 cap I found where Nichicon HMs with an impedance of 16, so that's out. I'm thinking I'll have to go with a diameter of 10. I haven't been able to find a possible replacement cap with a proper impedance .

[mockingbird]

Re: Electrolyics and Impedance/ESR

For CapXon KM use any of the following: United Chemicon LXY, LXZ, KY; Panasonic FC; Nichicon PW, PM, HE; Rubycon YXG
For CapXon GL: United Chemicon KY; Panasonic FK; Nichicon PM, HE; Rubycon PX
For CapXon KF: United Chemicon KZE; Panasonic FM; Nichicon HD; Rubycon ZL
Most of these are available on Mouser and Digikey. Rubycon ZL, PX, and YXG are available at badcaps.net only.

Looks like you also have some OST RLS there. I just replaced like 15 of them on an Asus nForce motherboard. I used Nichicon PW to do it. Anything in that class would be good too.

[PCBONEZ]

Re: Electrolyics and Impedance/ESR

awesome! thanks for clearing that up for me. I'll do my best to stick as close as possible when working with PSUs. I'm not sure if its best to start a new thread, but I have a OCZ 700W gamexstream PSU that has 2 bulged slim (8mmX20mm) 2200uF 6.3V Capxon KF caps. The problem is that the data sheet for capxon KF caps don't have this size cap. They only have 10x20 and 10x25, with impedance of 90 and 95 mOhm respectively. Is is safe to assume the impedance is slightly less than these as the pattern suggests? perhaps 80-90 mOhm?

The only 2200uF 6.3v 8x20 cap I found where Nichicon HMs with an impedance of 16, so that's out. I'm thinking I'll have to go with a diameter of 10. I haven't been able to find a possible replacement cap with a proper impedance .

[Within a series] the ESR and Ripple specs follow & change with the can size, so when you come across a custom ordered size look in the data sheet to find the closest uF/volts cap with the same can size you actually have, then use that cap's specs. It will be close if not dead-on.

I've run into those 8x20mm 2200uF KF before.
I ended up using HM because there was no way to use bigger than 8mm in that PSU.
Some of the 8x20 were actually heat-shrinked to prevent grounding.
.

[PCBONEZ]

Re: Electrolyics and Impedance/ESR

I'm dog tired right now but I'll TRY to explain how to work around phantom caps sizes.

The KF has no 8x20 so look at the 8x16....
Find another brand/series with an 8x16 with the same or close specs to the KF 8x16.
I will call it the ZIT series to be silly.
8x16 KF specs = 8x16 ZIT specs.
Look at the ZIT 8x20mm and use those numbers for the KF 8x20mm.
.

[PCBONEZ]

Re: Electrolyics and Impedance/ESR

Looks like you also have some OST RLS there. I just replaced like 15 of them on an Asus nForce motherboard. I used Nichicon PW to do it. Anything in that class would be good too.

Yes.
And as to the quote you quoted, Rubycon PX and not low ESR caps. They are GP.
.

[mockingbird]

Re: Electrolyics and Impedance/ESR

Yes, you're right, but I used the "near GP" low-ESR caps like you suggested because of the better spec they usually have over GP.

[thisguy]

Re: Electrolyics and Impedance/ESR

Thanks for the explanation. So if I must use 8x20 HM series caps, how significant will it affect the ripple and affect my computer? Will there be risks to my computer?

Also I made a mistake initially on the size of the Capxon 2200uF 6.3v cap, the size is actually 8x25mm (not 8x20). Since the diameter was the limiting factoring in choosing another cap I don't think it really changes much right?

One of the 2200uF caps is indeed heat-shrinked. When you say to prevent grounding you just mean so it doesn't come into contact with any nearby coils/heatsinks and short right?

Mockingbird,
The OST cap next to it is actually a 10x20mm RLP series cap, with a higher impedance than the RLS series.

Thanks again for replying so late when you're all tired

[Th3_uN1Qu3]

Re: Electrolyics and Impedance/ESR

Thanks for the explanation. So if I must use 8x20 HM series caps, how significant will it affect the ripple and affect my computer? Will there be risks to my computer?

Usually not. Why ripple may sometimes go higher is because it's not ripple - it's control loop oscillation. If they power supply's control loop wasn't compensated properly it will oscillate when parts shift values or are changed with different ones. When we say that ripple could be higher, it could go maybe 10-20mV higher... which isn't a risk. Oscillation brings a tell-tale hiss from the transformer - if the transformer is quiet the control loop is likely okay.

One of the 2200uF caps is indeed heat-shrinked. When you say to prevent grounding you just mean so it doesn't come into contact with any nearby coils/heatsinks and short right?

Correct.

[mockingbird]

Re: Electrolyics and Impedance/ESR

Thanks for the explanation. So if I must use 8x20 HM series caps, how significant will it affect the ripple and affect my computer? Will there be risks to my computer?

From what I understand, using too low ESR caps in place of higher ESR caps will cause the coil to "whine".

I would like PCBONEZ to legitimize my suggestion before you take it at face value... But since a 8mm 2200uF cap is impossible to find except for HM, why not use a 1800uF cap in its place that matches the ESR, and then replace another cap that is in parallel with it with a higher uF value to compensate for the deficiency?

I had to do this recently. I needed a short cap that was 2200uF, I ultimately ended up putting a 1800uF Panasonic FR (Miniaturized FM) instead.

Also I made a mistake initially on the size of the Capxon 2200uF 6.3v cap, the size is actually 8x25mm (not 8x20). Since the diameter was the limiting factoring in choosing another cap I don't think it really changes much right?

Don't worry about height. As long as it fits.

One of the 2200uF caps is indeed heat-shrinked. When you say to prevent grounding you just mean so it doesn't come into contact with any nearby coils/heatsinks and short right?

I am wondering the same thing myself. That 1800uF FR I mentioned... The top barely clears a heatsink. I'm wondering if I should paint some nail polish on top.

The OST cap next to it is actually a 10x20mm RLP series cap, with a higher impedance than the RLS series.

According to PCBONEZ' great reference sheet, you'll require one of these series to replace RLP:

------------------------------------------------------
HE, KY, YXG, YXH -- [.069][.046]
------------------------------------------------------
JXA, FC, PW, NRSX, NRSZ, and others --[.065-.080][.044-.054]
------------------------------------------------------

(The first bracket denotes a 8mm part, so the specs *almost* match for a 10mm part. RLP spec sheet shows .041, but that exact ESR can only be found with 8mm parts. Panasonic FC and Nichicon PW are probably the best match here).

[PCBONEZ]

Re: Electrolyics and Impedance/ESR

Usually not. Why ripple may sometimes go higher is because it's not ripple - it's control loop oscillation. If they power supply's control loop wasn't compensated properly it will oscillate when parts shift values or are changed with different ones. When we say that ripple could be higher, it could go maybe 10-20mV higher... which isn't a risk. Oscillation brings a tell-tale hiss from the transformer - if the transformer is quiet the control loop is likely okay.

That too but what I said is accurate.
Switcher Ripple in the OP can also go up simply because the filter is no longer 'tuned' to optimize filtering at the ripple frequency.
-
It can be more than a 20mV increase and if the PSU is already border-line it can easily exceed the ATX ripple spec.
-
Oklahoma Wolf [from JonnyGURU site] inadvertently proved this several years ago when he did a PSU recap intended to be used as an article for the site. The ripple after the recap [with good Jap caps] was worse than the ripple before the recap [with cheap CH/TW caps] largely because the ESR of the before/after caps wasn't even checked. [What he did was the 'normal thing' at the time but we 'the community of recappers' [does that sound cool or what?] know more now than we did back then.] The 'story' play-by-play was in the old [now defunct] forum over there and I don't think it was ever published as an actual article because the outcome was a bit embarrassing.

[This is not aimed at anyone, just an observation by someone that became an ET 30 years ago...]
When they teach ET's all about filters and band-pass and bandwidth and all the related cool stuff [that I learned, and forgot, and am trying to relearn] they do it at RF frequencies where ESR has negligible effect, so they tell [or used to] the students to disregard ESR completely. Well, the students learn how to do 'filter math' without looking at ESR. They get used to not looking at ESR. They teach other people how to do it and never mention ESR. There are even Electronics instructors & teachers that don't know what ESR is. Problem is down at PSU ripple frequencies the ESR is a really big deal and it needs to be included in the equations.

For me the #1 reason to stick to original values is:
I just want to get the recap done without tracing out circuits or practicing my math skills.
IOW: Lazy... LOL
.

[PCBONEZ]

Re: Electrolyics and Impedance/ESR

I am wondering the same thing myself. That 1800uF FR I mentioned... The top barely clears a heatsink. I'm wondering if I should paint some nail polish on top.

I don't remember right now were I saw it [and so don't remember the numbers right off] but there is actually a minimum clearance required over the top of vented caps.
The reason is the caps have to have room to bloat -and- to still allow the vent to open after bloating to release pressure if something goes wrong.
If the clearance won't let it vent then the cap may explode vice just venting.
.

[PCBONEZ]

Re: Electrolyics and Impedance/ESR

Thanks for the explanation. So if I must use 8x20 HM series caps, how significant will it affect the ripple and affect my computer? Will there be risks to my computer?

Depends on the specific circuit.
If the filter is just 1-cap & 1-coil it will be different than if the filter is a full pi filter.
If a full pi filter it will be different between changing the value on only one side of the filter as opposed to changing both caps.
Also if the cap isn't in the final OP [output] filter it might not affect ripple at all.
-
If you really want to know the equations are all over on-line.
You will have to figure out the values of the inductors in there first.
-
The HM suggestion is because this is one of those times I mentioned earlier where availability of replacements gives you no other choice.
Keep in mind you don't have to blanketly use HM to replace all these caps.
You could use HM in positions where there is no other option and something else in positions where you have more wiggle room for the can size.

Also I made a mistake initially on the size of the Capxon 2200uF 6.3v cap, the size is actually 8x25mm (not 8x20). Since the diameter was the limiting factoring in choosing another cap I don't think it really changes much right?

It can but you can just modify my ZIT equivalency scenario and look for a ZIT that is 8x25mm instead of 8x20mm
That's a really big ZIT.
.

[goodpsusearch]

Re: Electrolyics and Impedance/ESR

I replaced a bulging 2200uF 6.3V Capxon KF 8mm with a Nippon Chemicon KZE 6.3V 1500uF 10mm in a power supply similar with yours and it is working without a problem since then:
https://www.badcaps.net/forum/attach...8&d=1287917331


thread: https://www.badcaps.net/forum/showthread.php?t=11662

[PCBONEZ]

Re: Electrolyics and Impedance/ESR

Gezz,,,
KF is such a bastard series and 8x25mm is such a bastard size even I can't get a good cross reference.
I had to revert to using the KF 8x11.5mm because 8x16 is so uncommon.
I can more or less match the 8x11.5 but then those series that match don't have an 8x25mm.

You will find that a lot of the crap brand data sheets have data that seems randomly scattered all over the place. [Capxon KF is one of those.]
- Whereas "Good Brand" data sheets follow sensible patterns.

From what I can tell an 8mm KF would be a slight downgrade from a Chemicon KY but a Panasonic FC wouldn't be adequate.

All 8x11.5mm
KF 580/.250 or 590/.156 [used 2]
FC 555/.117 [too low on Ripple]
KY 640/.13
.

goodpsusearch's work-around is another option that works at times although I'd probably look hard for a 1800uF 8mm instead if it was in an OP filter.
.

[TheLaw]

Re: Electrolyics and Impedance/ESR

8mm caps aren't too common, for sure. Usually it's like 5-6mm and then the next interval is at like 10mm. That's what I've found after opening maybe 4-5 PSUs.

[thisguy]

Re: Electrolyics and Impedance/ESR

As a novice, I may just start with replacing the bad 2200uF Capxons as I'm afraid I'm gonna screw something up if I start recapping everything. As I get more time and knowledge I'll consider replacing those OSTs everyone hates. I'll be honest, I should be studying for an upcoming MCAT exam lol . I'll probably just try the 8mm HM as there so little wiggle room on the PCB and anything larger in diameter would lead to a very awkwardly placed caps (especially since its flush against another cap and I would have to have long legs on the cap and bend them parallel to the axis of the leads). Out of curiosity, does placing a cap off the PCB with long legs cause any problems like interference or anything (assuming the legs aren't at risk from shorting anywhere).

Looks like the HM is only in stock on Mouser. It seems like every time look something up Digikey it's out of stock. Wish Mouser had a cheaper First class USPS shipping option like Digikey.

On a side note, when you guys remove the PCB of the power supply from the case, do you guys usually desolder the input wires? On my PSU, it seems like the only way to unclip the socket for the incoming AC line and the power switch is to pull them outwards (which is a problem because they are soldered to wires from the PCB). It seems a little awkward to desolder the joints where the lines meet the PCB because the limited space. And if I desolder the joint at the socket, I worried that all the X and Y filter caps soldered to the same joint will fall off and it would be a bitch to put back on. How do the pro's usually handle this??

[TheLaw]

Re: Electrolyics and Impedance/ESR

Be confident! Power supplies are pretty easy because they are kind of "heavy duty". Motherboards are a little more fragile.