Right Cap for the right job

[ELItheICEman]

Hey guys, pretty new here but I've been lurking for about a year. I am a Video Service Tech, I work on video games, jukeboxes and whatever else my boss needs fixed. I also do a lot of work from my shop at home. We work on a lot of PSU's and Motherboards for games, jukeboxes and Counter Top games. So far I have stuck with Nichicon HM or HE series, and whatever decent caps are in the shop (my Co-worker orders Panasonics caps). Should I stick to stocking the same caps for most situations (PSU's or mobo's)? Or should I get specific series for different parts of the circuits? Thanks for any feedback!

[PCBONEZ]

Re: Right Cap for the right job

I did a pretty big blurb on that earlier over here:
https://www.badcaps.net/forum/showpo...3&postcount=13
.
The first part probably won't make sense out of context but the part after 'specs issues' relates to you question.
.
And no. Caps aren't all that universal if you want them to last.
.

[ELItheICEman]

Re: Right Cap for the right job

Lot of good information there, But I am still a little confused. I'll use a few examples. I just recapped a few caps on a coworkers motherboard, for the most part the board had Panasonic caps that at least visually seem to be doing fine (don't have a reliable way to test caps yet). But they had some cheap looking 1800uf caps (5) beside the proc, 1 of which was completely blown. Should I have tried to identify the caps on the board so I don't use a cap that is too out of spec? I ended up using Nichicon HZ series caps, mostly due to their size (the original caps were a small diameter). The ripple current for the caps is 2.88A, and the other specs seemed ok to me (not much to base that on I suppose). So in the case of capacitors regulating the proc voltage what specs are most important?

[PCBONEZ]

Re: Right Cap for the right job

Capacitors don't regulate, they filter and smooth.
As a filter they short Ripple voltages to ground [hence 'filter' it out of the power].
As a smoother they slow down voltage changes caused by load current changes (transients) - like when the CPU goes from 20% to 100% - which keeps the voltage from dipping too low [or going too high] before the regulator can adjust.
.
You are asking about the Vcore caps?
The caps on the CPU side of the VRM?
- For those the ESR is #1.

- (For a given ESR 'grade'.) Physically bigger caps tend to have lower ESR.
- There are space issues near the CPU.
- And the penny pinching Bean Counters have a say too.
In Vcore when using Lytics they usually use high grades of caps and/or use the largest caps that will physically fit in the space available to get ESR low enough with as as few caps as possible.

With Lytic caps [for decades] the uF in Vcore caps has been over-kill in aid of getting the ESR low.
[Thus people are 'used to' seeing more total uF there than is actually needed.]

When Solid Poly took over the Vcore duties getting uF below .010 per cap wasn't a problem anymore and it became evident (by what is seen on actual production boards) that total uF only needs to be about 5000uF in Vcore - although quality [non junk brand] boards usually use 5500-7000 total uF.
[[Type CPU variations] Low power CPU's like the Atom or laptop cpu's [on desktop boards - MoDT] can get by with a little less because transients aren't so heavy. Socket 771 server boards tend to have 6500-7500uF in Vcore *per CPU*. The current i3/i5/i7 types and the older socket 478 don't need any more than socket 775 does. - That might change if CPU watts gets up over 150 watts again...]

HZ should be fine in any Vcore.
But replace at least 1/2 with HZ. [All is better.]
If you stick one HZ [really low ESR] in a Vcore in parallel with a bunch of higher ESR [higher than HZ] caps then the bulk of the Ripple is going to want to funnel through the HZ and you might exceed it's Ripple current rating.
[Equations used for Resistors/Resistance work using ESR instead to figure out what the Ripple Current is doing and the total ESR. There are calculators for resistors on-line.]
Lets say you have nine .022 ohm 'resistors' in parallel and you replace just one with a .008 ohm 'resistor'.
The .008 ohm is going to 'steal' a lot of current from the others.
- U C ?

When it comes to Vcore you might find it's cheaper to upgrade it to Polymer than to use HZ.
HZ are pricey as Lytics go and the common sizes of Polymers have gotten cheap.
- Just make sure you have no less than 5000uF [5500uF is a better plan] and that the total ESR is less than the total for the original caps.
[THAT mod is well tested - but only for Vcore.....]

Now that I've rambled on the details...
You basically have to look up the specs for the old caps to know what you need.
- ESR should be the same or lower.
- Ripple should be the same or higher.
With 'good brand' caps ESR and Ripple pretty much track each other in locked-step from grade to grade so it isn't as hard as it sounds at first.
And after a while you'll 'just know' what grade needs to be where. [Most of the time anyway.]

This explains what I mean by 'grades' of ESR.
https://www.badcaps.net/forum/showpo...1&postcount=17
.
.
BTW: 'Looks' aren't always an indicator of quality.
Suncon 'look' cheap but they are some of the best caps you can get.
The subsidiary of Sanyo that made their Lytics 'bought themselves out' from Sanyo and changed their name to Suncon.
Same factory, same people, same caps, different name.
Supermicro has used Sanyo/Suncon almost exclusively for their Lytic caps for ~10 years now and Supermicro boards just don't blow caps.
.