calibre 8500gt blown cap

[holmes222]

hi..need help here..
have a calibre 8500gt lying aroung ..closer inspection, there are 4 lytic caps that looks bulged and broke ...list :
- 6.3v 1500uf KZJ (1)
- 16v 680uf KZG (2)
- 16v 470uf KZG (1)

the plan is to replace 4 lytic caps with solid caps
question :
- do the voltage and capacitance must be same as before?
- if not, how much value of voltage and capacitance (solid caps) is suitable to use ?

thans..

[Th3_uN1Qu3]

Re: calibre 8500gt blown cap

Half capacitance is generally the rule for solid caps. Voltage stays the same obviously.

[Agent24]

Re: calibre 8500gt blown cap

Half capacitance is generally the rule for solid caps.

I thought that depended on the type of circuit.

[PCBONEZ]

Re: calibre 8500gt blown cap

That half capacitance 'rule' originally came from a conversation between myself and KC8 a few years ago and it was SPECIFIC TO VRMs.
.
Then some twit read it out of context and started spreading the idea that 1/2 is always okay. - IT'S NOT.
And now we have a whole passel of 'Repeater Twits' that don't know where it came from spreading it even further...
.
It comes from an application guide for a VRM Controller Chip.
It DOES NOT AND NEVER DID apply to anything except a VRM.
.
If it's not a VRM cap then use the original uF or draw out the schematic and do the math.
.

[c_hegge]

Re: calibre 8500gt blown cap

On a video card, though, the rule does apply, since they are are in the GPU (and some cases VRAM) VRM.

[holmes222]

Re: calibre 8500gt blown cap

with this picture @attachment
looks like its on the vrm line ? or just voltage smoth filtering ?
so ..whats rules apply ? half capacitance ? same capacitance ? other ?
still find suitable replacement for tht lytic here ..

[PCBONEZ]

Re: calibre 8500gt blown cap

On a video card, though, the rule does apply, since they are are in the GPU (and some cases VRAM) VRM.

No. - There is no 'rule' for video cards.

Too many different circuits on video cards.
No standard layout.
Some parts of video cards ARE frequency sensitive - and so uF sensitive.

If you want to mod a video card then trace the circuits.
.
.
-----
Beyond that - 'any ole polymer' won't do. [As suggested by the Post that started me ranting.]
.
Back to my and KC8's conversation that the twit squad skimmed over, took parts out of context, and ran with,
.. Even in a VRM you can only lower uF if you are significantly lowering ESR.
.
There are Polymer that have WORSE ESR than some Lytics.
.
.
There is also that some circuits [especially VRs and VRMs] require some amount of bulk capacitance.
[That quality of uF that acts like a battery back-up for transients.]
For -CPU- VRMs we know what that is by Intel docs, but GPU manufacturers don't publish that info.
.
.
So a blanket statements like:
""'any' Lytic can be replaced with 'any' Polymer""
-or-
""'any' Lytic can be replaced with 'any' Polymer - if you reduce uF by half.""
Are complete bullshit and Bad Advice.
It's okay - sometimes....
.
.
.
-------
'Rules' apply to -all- of 'whatever':
There are far too many video card designs to say -anything- is 'always' true on video cards.
- I don't care how many times you 'got lucky'.
- There are too many video card designs to make a blanket statement like that.
.
And any suggestion that implies 'any ole polymer will do' - is bogus.
Even when switching lytic to polymer you have to check specs on both new and old caps or you don't know a thing.
.
Any suggestion of lowering uF by using polys - that does not include lowering ESR - is incomplete and gives people the wrong idea.
.
.

[Agent24]

Re: calibre 8500gt blown cap

I think you're better off just to forget polymer and go with a high quality electrolytic, they last long enough!

Personally I would only do a Poly-mod on something where it has been proven to work already...

[PCBONEZ]

Re: calibre 8500gt blown cap

with this picture @attachment
looks like its on the vrm line ? or just voltage smoth filtering ?
so ..whats rules apply ? half capacitance ? same capacitance ? other ?
still find suitable replacement for tht lytic here ..

KZJ and KZG are high failure rate caps and they don't like heat.
Not surprised they blew on a video card.
.
Those look to be in VR or VRM OP filters.
Check if one leg of each cap is attached to a MOSFET.
If so they are VR or VRM OP filters.
.
Regardless, you can use poly without issue as long as you leave uF the same and make sure you you get poly with ESR at least as low as the old caps.
[There are Polys out now that have really crappy ESR so pay attention to what you are getting.]
Lowering uF ~might~ have issues. 'Probably' get lucky, it's 'usually' okay, but no guarantees on untried mods.
The only one that can say for sure is someone that has done so with THAT card.
.
Poly are readily available up to 1500uF [in 6.3v] these days so there's no legitimate reason to be lowing uF for caps in that uF range anyway.
Few years ago a ploy with uF as high as 1000uF was rare.
Not so rare anymore.
.
Going Lytic:
KZJ need caps like Nichicon HN or Rubycon MCZ or better. [Not many options anymore.]
KZG are spec'ed like MBZ, WG, HM, FJ. Those or better will do.
.
.
.
Where is 'INA'?
.

[holmes222]

Re: calibre 8500gt blown cap

KZJ and KZG are high failure rate caps and they don't like heat.
Not surprised they blew on a video card.
.
.
.
Where is 'INA'?
.

so the safe option is stay with electrolytic caps (quality for sure) with same voltage and capacitance ..seems a good option
still sort out a replacement candidate here

INA = INDONESIA

[Th3_uN1Qu3]

Re: calibre 8500gt blown cap

Voltage, capacitance, ESR and ripple have to be the same. You'll want a ripple current rating that is the same or higher, and +/-50% variation is alright when it comes to ESR. Electrolytic capacitors have pretty loose tolerances. I have successfully replaced both KZJ and KZG with KZE on a motherboard and it works just fine.

When polymer caps were first introduced they were always very low ESR compared to lytics, so you could decrease the capacitance and keep the same output ripple voltage, since ESR was lower. But from what i'm reading now it seems there have been more "general purpose" polies released, which have specs comparable with electrolytics - so if you're not careful what you're buying, the ESR advantage of polies is taken away and you'll end up using the same number of caps as if you were using lytics, which makes the job more expensive.

[PCBONEZ]

Re: calibre 8500gt blown cap

Voltage, capacitance, ESR and ripple have to be the same. You'll want a ripple current rating that is the same or higher, and +/-50% variation is alright when it comes to ESR. Electrolytic capacitors have pretty loose tolerances. I have successfully replaced both KZJ and KZG with KZE on a motherboard and it works just fine.

When polymer caps were first introduced they were always very low ESR compared to lytics, so you could decrease the capacitance and keep the same output ripple voltage, since ESR was lower. But from what i'm reading now it seems there have been more "general purpose" polies released, which have specs comparable with electrolytics - so if you're not careful what you're buying, the ESR advantage of polies is taken away and you'll end up using the same number of caps as if you were using lytics, which makes the job more expensive.

Th3_uN1Qu3 when you showed up I was impressed with how bright you are but now you taken the habit of passing out bad advice.
You got lucky in some isolated cases and are passing out bad advice based on a very few one-off experiences.
The fact that something works in a few specific situations doesn't make it 'rule' that applies to everything.
-
You are smart but young and reckless.
Not everyone is like you and here for just a hobby and kicks.
-
For all you know the person at the other end's LIVELIHOOD depends on what they learn when they ask questions here - so if you aren't going to hand out complete and accurate advice - then just don't.
-

[PCBONEZ]

Re: calibre 8500gt blown cap

[Back to the 'room' as it were.]
-
The vast majority of experienced people here would call replacing KZG or KZJ with KZE stupid.
It will -always- lead to more Ripple in the circuit [if there was Ripple there to start with] and no one past the level of a novice would ever recommend it.
-
A +/-50% variation in ESR is bogus.
It stems from not understanding meter readings.
A typical ESR meter can only read to +/- .01 ohms.
So:
If you have a batch of caps that are supposed to have a .010 ohm ESR then meter readings of .00 to .02 [+/- .01] should be expected.
- That's not a variation in the cap's ESRs - it's meter capability.
Same-same: If you have a batch of caps that are supposed to have a .030 ohm ESR then meter readings of .02 to .04 [+/- .01] should be expected.
- That's still not a variation in the cap's ESRs - it's still meter issues.
.
If you have some caps where their ESR is supposed to be .044 ohms [not uncommon in PSU caps or even on motherboards] then this +/-50% idea would mean .022-.066 ohms for -actual- ESR.
[Even without meter error that would be .02-.07 ohms on a meter.]
If the readings vary by that much in a batch then some of those caps are bad.
.
+/-50% = okay .. is bogus for another reason.
In data sheets the stated ESR is the MAX value.
That means they are +0% on the 'plus side'.
Any ESR reading that is higher than the data sheet by more than .01 ohms [meter error] indicates a bad cap.
That means if that .044 ohm ESR rated cap reads .06 ohms on an ESR meter it's -definitely- bad.
~ And that's only 36% higher.
In things like LCD screens caps with .058 or so ESR are not uncommon so reading .07 would mean it's -definitely- bad.
~ That's only 21%.
.
.
[General FYI]
High accuracy ESR meter cost 10 to 100 times what a typical trouble shooting ESR meter costs. Only one person I know even has one. The ones that are only 10x more $ aren't that great for ranges [don't have 100kHz at all or are only good fro small uF at 100kHz].
The 'real' good ones are usually large bench devices and a PITA to use.
.

~~~~~ Moving on to specs issues:

~ Voltage:
The Voltage rating is a limit of the cap.
- It just has to be more than the actual circuit voltage.
Changing it 'up' [within reason] is not a problem.
[There has to be enough DC BIAS (and thus DC leakage) for the self-healing effect to work. A rated voltage 10x or more than the actual voltage is pushing it for anything that handles a lot of ripple or large/fast signals. (It's voltage reversals that degrade the oxide layer such that it needs to be healed.)]
Changing it 'down' you -have- to know the actual circuit voltage and don't go below that.
Actual Voltage exceeding the Rated Voltage will cause excessive DC leakage currents which can lead to a permanently degraded or even a shorted cap.

~ Capacitance:
[Not counting the non-typical case of Poly mods here...]
Capacitance (uF) is an application consideration.
Changing uF is not recommended unless you can't get around it.
The capacitance [uF] often works in conjunction with other components (which the uF is matched to) or the bulk uF is important for smoothing transients.
If you MUST change then it's preferred to go up one standard value [which is roughly 20%] as the typical uF tolerance [on the caps] is +/-20% and uF goes down as caps age.
[So as it ages it will be getting -closer- to the original value instead of further from it.]
Changing uF won't hurt the cap but can screw up the circuit operation in several different ways - depends on what the circuit does.

~ Ripple:
The Ripple rating is a limit of the cap.
It's basically it's amps limit for AC [typically rated at 100kHz for low ESR caps.]
Exceeding it for any length of time will cause the cap to overheat internally and die young.

~ ESR:
ESR is an application consideration.
The ESR is in essence the cap's 'resistance' to passing AC [Ripple] and is usually stated at 100 kHz for low ESR caps.
[Data sheets usually have conversion factors for other frequencies.]
100kHz is used because:
- 1: It's near [usually dead-on] the low point of the ESR curve.
- 2: It's near the frequency of the Ripple from most SMPS circuits. [Including single chip regulators.]
If the ESR is higher than it should be then less Ripple is being removed from the power distribution and any IC chips that use that power are subjected to Ripple Voltages in their supplied DC power.
IC chips don't work well with noise in their power source and are inclined to have noise in their output. Their output is data.
In 'PC Speak' [vice 'Electronics Speak'] that means high ESR = corrupted data.
On a mobo IC chips affected include: the CPU, Chipset, RAM, Drive Controllers, Drive PCB's, LAN chips and ~all~ the other chips they connect to.
~~
Even in a system that 'appears' to be working perfectly fine, too high of ESR can "present as": corrupted/ruined/missing files (including system files that halt booting on an error) on hard drives (the drive will test good on another board), RAM errors, bad RAM (that tests good on another board), no boot, boot loop, BSOD, non-working LAN/Sound/USB, and anything else that can be caused by a bad IC chip. [The IC chip probably isn't actually bad if you have high ESR.]
~~
- Thus ESR is probably ~THE~ most critical spec in low ESR applications.
Raising ESR - BAD!
Lowering ESR - In most cases it's okay - but there are exceptions.
- 1: Lowering ESR drastically but only far from the power source [for example far from the PSU connector on a mobo] means you are encouraging Ripple to pass through all the traces on the board to get to the lowest ESR cap. That not only means you are heating the traces with ripple, you are also exposing the IC chips that tap off between the power source and the cap to more ripple because the trace resistance will cause some to pass through the IC's instead. [Think like resistance dividers. The trace, the cap, and the IC are resistors as far as the ripple 'knows'..] If the lowest ESR is near the power source then that won't happen because less Ripple gets to the traces to start with. So: If you are only going to lower ESR on a few caps then do the ones closest to the power source. Or: change -all- the caps the same amount. - The caps farther from the power [that aren't for other local regulators] are actually to prevent ripple from 'bouncing' back in a similar way to what happens with RF in antennas. Their job it to keep the DC nice and 'flat' for the entire length of the traces. Ideally they shouldn't 'see' that much ripple but they should pass all they 'see'.
- 2: Too low of ESR can cause MOSFET 'ringing' in some VRM type circuits. In other words is creates high frequency noise in bursts. It is rather rare but it's a good reason not to go psycho lowering ESR a huge amount below the original value 'just because'.
- 3: In PI Filters the value of ESR works in conjunction with the the value of the coil to 'optimize' the over-all circuit to remove the expected ripple frequency. (Which will vary by circuit design.) Xc is -not- the predominate factor at 100KHz because ESL cancels it's effects out in that range leaving ESR as the only factor doing much of anything.
.
.

[Agent24]

Re: calibre 8500gt blown cap

I think that post should go into Ratdude's FAQ document thing, if he's still working on it...

[PCBONEZ]

Re: calibre 8500gt blown cap

hi..need help here..
have a calibre 8500gt lying aroung ..closer inspection, there are 4 lytic caps that looks bulged and broke ...list :
- 6.3v 1500uf KZJ (1)
- 16v 680uf KZG (2)
- 16v 470uf KZG (1)

INDONESIA - I dunno where you can get good caps there.

KZJ and KZG are both out of production.
Here are the data sheets.
.

[Th3_uN1Qu3]

Re: calibre 8500gt blown cap

If you ask me where i came up with 50% variation in ESR just read *any* cap datasheet. Along the life of the cap the ESR is allowed to double for the cap to be considered bad.

The advice i give is due to where i live. Do you know how repair shops here handle cap replacements? They just pull random caps from random boards that have failed for other reasons. No ESR meter, no time to check anyway. And they work. I bet this isn't the only place where these things happen. Since you said you don't know where to get good caps in Indonesia, why do you send him chasing unobtanium?

As for livelyhood, please. This is a cheap video card not an industrial or medical device.

[PCBONEZ]

Re: calibre 8500gt blown cap

If you ask me where i came up with 50% variation in ESR just read *any* cap datasheet. Along the life of the cap the ESR is allowed to double for the cap to be considered bad.

No, they don't.
.

[PCBONEZ]

Re: calibre 8500gt blown cap

The advice i give is due to where i live. Do you know how repair shops here handle cap replacements? They just pull random caps from random boards that have failed for other reasons. No ESR meter, no time to check anyway. And they work. I bet this isn't the only place where these things happen. Since you said you don't know where to get good caps in Indonesia, why do you send him chasing unobtanium?

As for livelyhood, please. This is a cheap video card not an industrial or medical device.

But the OP is FAR from the only person that will -ever- read this and blanket statements that are WRONG can affect -numerous- people in different situations.
That includes people that HAVE to do this to eat.

You've been here long enough to KNOW when it's bad advice.

So, if you are going to be irresponsible and hand out bad advice then I'm going to call you on it.
.

[Th3_uN1Qu3]

Re: calibre 8500gt blown cap

Mkay, i'll do my best to only post my own projects here, and let everyone else get GOOD advice.

[holmes222]

Re: calibre 8500gt blown cap

sory for the late comment..busy time here
...here in badcaps i have enlighten with the bright view of suggestion and knowledge..
i know its very hard to find high quality caps in my country here ..and the video card is just plain old card for now ...but,hey..if that work its still good for test or just office/browsing pc
for project update :
- find 6.3v 1500uf LXV here (consider to replacing KZJ) (its a good option ?? or not ??)
- the caps with similiar rating to KZG is hard to find here
picture added soon