kzg / ost

[PCBONEZ]

Re: kzg / ost

hah, never thought about doing that

I have.
You end up with blobs of solder stuck to the wall and floor.
If you're lucky you won't get a hot one down your sock. [Or in someone's eye.]
.

[Uranium-235]

Re: kzg / ost

nobody around, and I wear googles. maybe I should try it so it just gets stuck on my desk

then again, I have a new hakko tip coming in a few days

[c_hegge]

Re: kzg / ost

...and I wear googles...

I've never heard of anyone who actually wears google.

[PCBONEZ]

Re: kzg / ost

She does it WELL!
.

[goodpsusearch]

Re: kzg / ost

This is the best!

[CapBlown]

Re: kzg / ost

Makes me wonder if she's wearing it or not

[a23d56]

Re: kzg / ost

You end up with blobs of solder stuck to the wall and floor.

True, but in my shack that improves the decor

[Uranium-235]

Re: kzg / ost

hell yeah I wear google

[PCBONEZ]

Re: kzg / ost

https://www.badcaps.net/forum/showpo...9&postcount=26

Sorry, that was the kzg/ost thread I started a couple of days ago.

The bad caps were Nichicon HN 820uf 6.3v 8x20 mm. After removing them from my Intel D865perl board, now I can see a code A0313 too.

Are they all in the yellow circle and how many do you have?
Or, if you have mixed caps there then what are they and how many of each kind?
[Some pics would help...]
-
So many of these have been recapped if you look at 10 pics you might see ten different sets of caps.
I really don't know what you are looking at unless you show or tell me.
You say you are replacing 5 caps but it looks like you should have 9 or 10 there.

There's a BIG difference between the others being polymer vice being more lytics.
HN are 'abnormally extra good' [better than usual] caps in a skt478 with 9-10 Vcore caps.
- Is it because the other 4-5 caps are crap for specs?? [Not quality,, specs]
- Is it because they just wanted some over-kill and the other caps are extra good too.
If the others are crap for specs it's best keep close to HN specs.
If the other caps are polymer [or some other very high grade] then perhaps it's safe to downgrade from HN a bit.
[See why I'm asking now.]

Now I'm just going to give you an experience/observation thing.
Most P4 Prescott capable skt478 boards [i865 & i875 chipsets] with ~9 8mm Vcore caps used 1500-1800uF caps, usually something like MCZ, HN, or Panasonic FJ, FL.
A LOT of Intel built Dell skt 478 boards used 8x or 9x 8mm 1800uF FL or FJ in Vcore.
Any of those should work to replace the 820uF as long as you keep to the 8x20mm to keep the ESR and Ripple in spec. The extra uF THERE won't hurt anything. I know this because when 10mm caps are used on the same socket, chipset & cpu type 8x3300uF or 8-9x2200uF is common.
.
Switching to 820uF 2.5v or 4v Polymer is also an option.
Vcore actual voltage less than 2.0v so the lower voltage cap isn't a problem THERE.
https://www.badcaps.net/store/produc...products_id=76
https://www.badcaps.net/store/produc...products_id=77
The polys are also nice an short which makes heat-sinks easier and air flow better.
.

[a23d56]

Re: kzg / ost

https://www.badcaps.net/forum/showpo...9&postcount=26

Are they all in the yellow circle and how many do you have?
Or, if you have mixed caps there then what are they and how many of each kind?
[Some pics would help...]

So many of these have been recapped if you look at 10 pics you might see ten different sets of caps.
I really don't know what you are looking at unless you show or tell me.
You say you are replacing 5 caps but it looks like you should have 9 or 10 there.

I don't have a camera, but my board is just like the one in your picture with the yellow circle. I removed the 5 tall ones (they were bulged), and left the 5 short ones intact. The 5 tall ones I removed were Nichicon HN 820uf 6.3v 8x20mm.

There's a BIG difference between the others being polymer vice being more lytics.
HN are 'abnormally extra good' [better than usual] caps in a skt478 with 9-10 Vcore caps.
- Is it because the other 4-5 caps are crap for specs?? [Not quality,, specs]
- Is it because they just wanted some over-kill and the other caps are extra good too.
If the others are crap for specs it's best keep close to HN specs.
If the other caps are polymer [or some other very high grade] then perhaps it's safe to downgrade from HN a bit.
[See why I'm asking now.]

Now I'm just going to give you an experience/observation thing.
Most P4 Prescott capable skt478 boards [i865 & i875 chipsets] with ~9 8mm Vcore caps used 1500-1800uF caps, usually something like MCZ, HN, or Panasonic FJ, FL.
A LOT of Intel built Dell skt 478 boards used 8x or 9x 8mm 1800uF FL or FJ in Vcore.
Any of those should work to replace the 820uF as long as you keep to the 8x20mm to keep the ESR and Ripple in spec. The extra uF THERE won't hurt anything. I know this because when 10mm caps are used on the same socket, chipset & cpu type 8x3300uF or 8-9x2200uF is common.
.
Switching to 820uF 2.5v or 4v Polymer is also an option.
Vcore actual voltage less than 2.0v so the lower voltage cap isn't a problem THERE.
https://www.badcaps.net/store/produc...products_id=76
https://www.badcaps.net/store/produc...products_id=77
The polys are also nice an short which makes heat-sinks easier and air flow better.
.

I can't find any data sheet for the caps I removed, so I don't know what their ESR was. In the other thread, you said

Both of these are 'good brand' low ESR caps..

Nichicon HZ 8x20mm ,, ESR = .009 Ohms ,, Ripple = 2880 mA
Panasonic FC 8x20mm ,, ESR = .065 Ohms ,, Ripple = 995 mA

OK. But I'm looking at that Panasonic .065 ESR and wondering, why can't I use a cheap Nichicon HM that has better specs:

UHM1A102MPD6TD 1000uf 10v 8x15mm ESR=.028 ripple=1490.

Mouser sells them for $0.09 ea.

http://www.mouser.com/ProductDetail/...2fSa8Frg%3d%3d

[Scenic]

Re: kzg / ost

https://cdn.badcaps-static.com/pdfs/...dcde71fe08.pdf

[a23d56]

Re: kzg / ost

https://cdn.badcaps-static.com/pdfs/...dcde71fe08.pdf

Yes I was looking at that data sheet, but there was no 8x20 can size for the 820uf 6.3v. So I thought that meant my caps were some old version no longer made, and the old data sheet for them was no longer available.

But now I remember pcbonez said, find the same can size somewhere else in the series, because ESR and ripple are the same for any given can size within the series. OK, I got it now. I'm no EE, I just know enough to be dangerous ...

So that means the caps I removed had ESR=.012 ripple=2220.

And now the question becomes clear: is it reasonably safe to replace them with ESR=.028 ripple=1490? Are the design tolerances so tight to make that a bad idea?

[a23d56]

Re: kzg / ost

Switching to 820uF 2.5v or 4v Polymer is also an option.
Vcore actual voltage less than 2.0v so the lower voltage cap isn't a problem THERE.
https://www.badcaps.net/store/produc...products_id=76
https://www.badcaps.net/store/produc...products_id=77
The polys are also nice an short which makes heat-sinks easier and air flow better.
.

Never considered that till now, but looking at a data sheet on Mouser
https://cdn.badcaps-static.com/pdfs/...04935704c7.pdf

ESR is lower and ripple is higher than electrolytics. Is that why they work? Is lower ESR always better? Why would they mix and match polymers and elcectrolytics on the Intel D865perl VRM if they cost about the same?

[PCBONEZ]

Re: kzg / ost

Two or three years ago the cost of polymers wasn't about the same.
Still isn't except on some sizes that are popular and so in high production.
Four or five years ago polymers could easily cost 4x as much as lytics. [Or more.]
Some larger sizes of polymer are rare [low production] and cost in the range of $10-$15/cap.

They mix/match for various reasons.
Polymer are/were more expensive.
Polymer have better specs so they can 'cost down' a little on the other caps.
During the transition to polymer they weren't confident about lowering total uF in Vcore so they left some lytics for bulk capacitance.
Polymer are small so using them in some spots made installing heat-sinks easier and also they could move the Vcore caps a bit closer to the CPU.

Is lower ESR always better?
- No, but...
In Vcore it's proven. Just look at newer Core2Duo boards.
-- The ESR/Ripple/uF requirements for C2D aren't much different [if at all] than P4.
[Some circuits can depend on the ohms from ESR to work properly but that's not how Vcore caps work.]

-- [This again]
The ESR/Ripple/uF requirements for C2D aren't much different [if at all] than P4.
--
You can cap-mod both high and low side of a lytic type P4 VRM to what you see on [good stable] C2D boards with polymer and it will work fine. - That doesn't necessarily work for other areas/circuits with lytics on a mobo though.

You still haven't told me what the other caps in your Vcore are.
If you haven't figured it out yet I'm not so much into guessing when I give out advice.

With mixed caps the Ripple is going to 'tend to' go through the caps with the lowest ESR.
That means if the ESRs between the mixed caps is significantly different then to be safe [conservative] the caps with the lowest ESR should handle the entire Ripple for the whole circuit.
- So, I don't like to make 'educated guesses' on what will work if I don't know what is going on in the rest of the circuit and for some reason you won't tell me what the other Vcore caps are.

Now that I vented.
5x 820uF 2.5v R7 have Ripple 5x5600 mA = 28000 mA total.
Comparing to a common P4/C2D Vcore configuration.
9x 1800uF 8x20mm MCZ is 9x2350mA = 21150 mA total.
So - So 5x 820uF 2.5v R7 can handle all the Ripple for the whole VRM.
- But as I dunno what the other Vcore caps are I dunno what total ESR is doing or what the total uF in Vcore is and those matter too.
.

[PCBONEZ]

Re: kzg / ost

These are common configurations in [good] all poly P4/C2D boards.
7-8x820uF = 5740-6560uF
10x560uF = 5600uF
Presumably if you have at least the much total uF in Vcore you should be fine and it will be stable.
.

[a23d56]

Re: kzg / ost

You still haven't told me what the other caps in your Vcore are.
If you haven't figured it out yet I'm not so much into guessing when I give out advice.

I don't know what they are, but they look exactly like the ones in your picture, with similar blue streaks on the tops. I guess they're polymers. The only markings I can read are on the tops:

NA
560
g
343

I don't see any markings on the sides.

[a23d56]

Re: kzg / ost

Hmm, I see you mentioned 560uf poly caps. So my wild guess would be, the 560 marking on mine means 560uf.

If I understand what you've said: the ripple goes through the 5 low ESR / high ripple polys. The designer uses 5 electrolytics, to keep the cost down. No need to overbuild.

So I should be able to raise ESR somewhat on my 5 electrolytics, as long as I have enough total uF in the circuit. Sounds like the 5 cheap Nichicon HM 1000uF electrolytics I have in mind, will do the trick.

The end (I think).

[PCBONEZ]

Re: kzg / ost

>>> So I should be able to raise ESR somewhat <<<
In this case no. - That's why I ask all these seemingly pointless questions...

All polymer don't have great specs. That's a widely held misconception.
Those are polymer but they are rather sucky polymer. - Keep reading.

Yes, 560 is the uF.
- Those are Nichicon so the g = 4 volts.
Those are Nichicon NA series.
Obsolete for several years and no data sheet but NIC Components cross-references them to Sanyo OSCON SVP series - which are also rather sucky as polymers go..
(NIC Components is a lesser known Japanese 'good brand'.)

--
Also found this:
http://www.electronicstalk.com/news/nch/nch100.html
--
Can't say exactly what the 560uF NA series specs are but clearly it's no better than .025 ohms and 3700 mA.
[They always harp on the best one. That's what they harped on.]

As you can see the ESR is actually much worse than the HN Lytics.
In fact their ESR is rather crappy for Polymers.

If going polymer I would use 680uF or 820uF [with ESR at least as low as the HN] for the spots where you removed the HN to ensure enough bulk capacitance.
That should keep all of Bulk uF, Ripple and ESR in the ranges seen on factory boards.

Personally with 820uF only costing around 50 cents these days I would use them just because they are smaller and usually last longer.
.

[a23d56]

Re: kzg / ost

OK. At least I know what my options are now. I may pull out the 5 polymers too, next time I heat up my iron. Thanks for following this to the end.

The end. Again.