Announcement

**PCBONEZ** · 02-23-2008, 09:34 PM

Re: BlownRecap: Ripple Current ?

In moderation going up in capacitance is fine.
In moderation means maybe +50% max.
You went to 313% of the original value and that's going way too far to be safe.

Here is what -may- have happened.

VRM output caps are in parallel with small uF caps that are either (depending on the type CPU) mounted in and/or around the CPU socket or actually built into the CPU.

The small caps take care of the higher frequency ripple.
The VRM caps take care of the lower frequency ripple.
The frequencies in between the two are *shared* between the small uF caps and the VRM output caps.

When you went from 1500uF to 4700uF you changed the frequencies the VRM caps are effective at and reduced the amount of sharing that goes on for the 'in between' frequencies.

You basically dumped a cubic butt-ton of load onto those small uF caps.
If they are in your CPU then you may have a dead CPU.
If they are in the socket either you have a dead board or you can learn about replacing SMD caps.

..

At this point you need to find out if the CPU even still works.

.

**gdement** · 02-24-2008, 12:12 AM

Re: BlownRecap: Ripple Current ?

Originally posted by PCBONEZ

In moderation going up in capacitance is fine.
In moderation means maybe +50% max.
You went to 313% of the original value and that's going way too far to be safe.

I don't think he went up in capacitance, sounds like the 4700's were to replace existing 4700uF Nichicon HD's.

I fired up my machine with replacement 6.3v Panasonic 1500uF and 4700uF FM series capacitors and after about five minutes of operation it crashed and now refuses to video or POST, instead continually resetting. Smell associated with failure however no smoke, burn or scorch marks on motherboard. No visible physical signs of failure at all. PSU checks out.

Are you sure all the capacitors were installed in the correct orientation? When you looked for scorch marks, did you look at the mosfets? The constant resetting sounds like a power problem (maybe at board level), but I don't know what to suggest any more specifically.

**PCBONEZ** · 02-24-2008, 01:39 AM

Re: BlownRecap: Ripple Current ?

gdement
I C what you mean.

Intuit
HM is a high grade comparable to Rubycon MBZ series.
HE is a WAY lower grade than HM.
*Given the same physical size* HE only handles 55% of the ripple HM does.
*Given the same physical size* HE only handles 77% of the ripple HD does.
FM is in between HM and HD. Handles 85% of HM and more than HD.
On your level (vintage?) of board HE is not suitable to replace VRM caps.
http://www.nichicon-us.com/english/p...ini/list_f.htm
-
None of that should be enough to go dead board that fast.

What gdement said is valid.
Lytics in backwards usually only live a few minutes.
If that's the problem they may even be shorted inside. (Not good to CPU.)
Also: Look for solder where it shouldn't be and check your PSU.

.

**Intuit** · 02-24-2008, 03:06 AM

Re: BlownRecap: Ripple Current ?

Thanks for the insight guy(s)/gal(s).

I double, tripple, quadruple and quintuple checked everything once the job was finished and before firing it up. The only possibility in that regard is that either one or more of the caps were bad or marked incorrectly to begin with. No scorch marks anywhere period, including the mosfets. (worked in product tweak, test & inspection for two years) I simply didn't allow it run long enough. The CPU still gets warm after a few seconds of running which is a good sign for it.

To note, before I shut it off I noticed the case power LED rapidly flickering. I have a guess as to what may cause that. What do ya'll think ?

When I desoldered the two replacement 4700 caps (near the mosfets and inductors) the smell got particularly strong with one of them which leads me to believe that it *could* be the or a source of the smell. I haven't yet touched the replacement 1500s but are considering putting the originals back in for those as well... just to see if the board is by some chance salvageable. Any chance it is ?

I have a cap tester but unfortunately it only goes up to a lousy 200uF. Any way to use the nF with these uF caps ? Any way to test these caps individually to find out which may be bad, if any ?

Once again, thanks.

**PCBONEZ** · 02-24-2008, 03:18 AM

Re: BlownRecap: Ripple Current ?

Dunno how to fix the meter problem.
You can check for shorted caps with a regular ohm meter.

.

**Intuit** · 02-24-2008, 12:51 PM

Re: BlownRecap: Ripple Current ?

Based upon the final 6.3v comparisons can you see the apparent cause of the issue, minus possible faulty component(s) ? Combined effect from having so many "higher grade" capacitors ? (same uF rating)

Nichicon 4700 HD Series [ORIGINALs]
Max Impediance 20°C/100kHz: 0.016
Rated Ripple (mArms) 105°C/100kHz: 3290

Nichicon 4700 HE Series
Max Impediance 20°C/100kHz: 0.024
Rated Ripple (mArms) 105°C/100kHz: 2650

Panasonic 4700 FM Series [REPLACEMENTs]
Max Impediance 20°C/100kHz: 0.013
Rated Ripple (mA) 105°C/100kHz: 3630

.

Nichicon 1500 HM Series (12.5mm/16mm) [ORIGINALs 25mm]
Max Impediance 20°C/100kHz: 0.025/0.018
Rated Ripple (mArms) 105°C/100kHz: 1540/2000

Nichicon 1500 HE Series (20mm)
Max Impediance 20°C/100kHz: 0.046
Rated Ripple (mArms) 105°C/100kHz: 1400

Panasonic 1500 FM series capacitors [REPLACEMENTs]
Max Impediance 20°C/100kHz: 0.019
Rated Ripple (mA) 105°C/100kHz: 2180

re Ohm meter, yeah thought of that too. Even the known bad/leaky ones, out of close to fifty capacitors aren't shorted. There was some difference in the behavior of each same-value capacitor in relation to how they accepted and dispersed the charge to the ohm meter, which indicates that some were bad. Some that were determined bad by this makeshift method hadn't leaked or bulged at all.

This answers my question about using the nF setting to test the uF setting...
http://www.mikroe.com/en/books/keu/02.htm
>>... 1500pF is the same as 1.5nF, 100nF is 0.1uF ... 1F=10^6uF=10^9nF=10^12pF ...<<
Using that logic a 1500uF would be 1500000nF and 4700uF would be 4700000nF. Micro versus Nano in other words. Meter isn't going to handle that needless to say.

**Intuit** · 02-24-2008, 01:08 PM

Re: BlownRecap: Ripple Current ?

By the way, last time I recapped this board the first thing I did was go into the BIOS and monitor the voltages for a bit. Since no anomolies were noted before I neglected to do that this time around.

To others that may read this... don't make the same mistake. Always take an extra ten minutes or so to monitor the voltages and temperatures in the BIOS before doing anything else.

**PCBONEZ** · 02-24-2008, 03:00 PM

Re: BlownRecap: Ripple Current ?

Not sure your point.
.
The HD or HM to FM should be fine (not perfect but close enough).
The HD or HM to HE doesn't give you enough ESR.
[For ESR a smaller value is better.]

When talking about/comparing grades it compares the exact same can size [diameter and width] because ESR is tied to the physical size and that is the only practical way to compare cap grades.
Within any series the ESR and ripple will vary with the can size.
This makes things a little tricky because a 10x20mm in one series(grade) may be the same (or nearly) specs as say a 10x30mm in another series(grade).
-
The bottom line is you really need to look up the ESR/Ripple (and not just uF/volts) of original caps if you want to find appropriate replacements.
And, if that information is not available (as for some crap brands that don't publish data sheets) then you go by what is usually seen in that particular position on similar boards.

.
Using HE on your board may have made it permanently unstable but I don't think it would create your scenario where it worked fine for a while then went dead board all in one instant.
-
That sounds more like a cap (or something else) failed hard than problems with under rated caps.

~~~
Something I am curious about.
All I see is 6.3v caps mentioned.
What happened to the 16v caps?
Did you change those?

.

**kc8adu** · 02-24-2008, 05:04 PM

Re: BlownRecap: Ripple Current ?

look at your cpu.
if the hsf wasnt seated right it may run for a short time then burn out/stink.
had a mobo mailed back after a recap where the customer had the hsf cocked on the back/hinge side of the socket.cpu was burnt.amd chips quickly burn up if you goof up installing the hsf.

**Intuit** · 02-24-2008, 05:31 PM

Re: BlownRecap: Ripple Current ?

I think we're confusing eachother. The originals, as stated, are the Nichicon 1500 HM and the Nichicon 4700 HD. The replacements are the Panasonic 1500 FM and Panasonic 4700 FM.

HE were never used in this board. You said (that guy in the link was wrong,) the HE series weren't close to the HM series and so I just threw those specs in there too which confirms your statement. Didn't aim to confuse you with that extra bit. The HE series were never actually used, but originally pulled their specs because I did not (until you found them) locate the specs on the HM series; so used them in lieu of more precise data.

Okay I'm thouroughly confused. HE at 1400 versus ~2000 with HM doesn't give me "enough" ESR _BUT_ with ESR smaller values are better ? If both are true than logically I must be confusing the terms of ESR and Ripple Current and they aren't the same thing.

The difference from ~2000(HM) to 2180(FM) is only 8%, but what I'm asking is, is it possible that an accumulation of that difference accrossed several capacitors (if in series) have caused this failure ?

The HM series that I have measure 10x25mm and the specs given in the HM datasheet maxes out at only 10x16mm for the 6.3v line. Based on what you're saying my 10x25mm cap would have a higher ESR than the "2000" spec given for the 10x16mm ? If I understand right this would put it at an even closer match to the FM series which confirms what you're saying.

The (non-Nichicon brand) Teapo A3 series caps were all changed early-mid last year and had resumed 24/7 operation since then without issue. Re specifically the 16v caps, if I remember right there are six 100uF, two 470uF, and two 1000uF. The 100 and 470 were both replaced with FM series however due to availability, settled on FC for the two 1000uF.

I chose the FM series as replacements mostly because their rated life spans were the longest while maintaining higher Ripple handling capacity versus other series. Trade off was they're much larger than the originals which made installation more difficult than imagined.

SelfNote: https://www.badcaps.net/forum/showthread.php?t=1458

**Intuit** · 02-24-2008, 05:39 PM

Re: BlownRecap: Ripple Current ?

Originally posted by kc8adu

look at your cpu.
if the hsf wasnt seated right it may run for a short time then burn out/stink.
had a mobo mailed back after a recap where the customer had the hsf cocked on the back/hinge side of the socket.cpu was burnt.amd chips quickly burn up if you goof up installing the hsf.

The only thing I had noticed before was some of the black sealer surrounding the CPU core was flacking-off (it's old) but nothing to indicate burning there. The CPU heatsink does get warm so that's a good sign I assume. In addition the fact that the power light rapidly flickered upon failure does indicate that it is a power-related issue as opposed to overheating. But maybe you're right... and honestly I hope you are. Thanks for the comment Kc8adu.

**PCBONEZ** · 02-24-2008, 08:30 PM

Re: BlownRecap: Ripple Current ?

ESR and ripple are concerned with AC (volts at a frequency).
--
If they were about DC volts then:
ESR would be 'resistance'
Ripple would be 'current' or 'amps'
-
Filter caps are used to 'short' AC ripple to ground.

Both ESR and ripple ratings have to do with how well they 'pass' ripple (to ground).

Lower ESR means less resistance to passing the ripple voltage to ground.
Less=better

Higher ripple means it can pass more (current/amps) of ripple to ground.
(without overheating)
More=better

Changing the uF affects what frequency of AC they do this best at.
They will be less effective at frequencies above and below the optimal frequency.
(The over all circuit determines the freq,,, not just the caps.)

--

Physically bigger caps have lower ESR than smaller.
[--Within the same series that is.]
IOW in any series you'd expect a 10x16 to have higher (poorer) ESR than a 10x20.

ESR and Ripple are not directly related to uF.
ESR and Ripple are directly related to physical size, the lead spacing, the distance between the aluminum foils, the length of the foil roll (as rolled), and the chemistry of the electrolyte. - Most of these things are the same within a series but not from one series to another. (Usually) The only thing that changes within a series is the size of the completed roll.
-
I said 'usually' because some series are actually two or three different designs grouped into one series. This is common in crap brands but not too much in reputable brands.
- Sanyo WG for instance actually contains 3 or 4 sub-series in the one data table for WG.

.

**kc8adu** · 02-24-2008, 11:52 PM

Re: BlownRecap: Ripple Current ?

btw your choice of caps is not your problem.
i would double check they are installed correctly.
normally they bulge and pop quickly if reversed.
interested in hearing about that cpu.

**Intuit** · 04-01-2008, 04:46 PM

Re: BlownRecap: Ripple Current ?

Hello KC8ADU. Last time I tried to beat around the bush and explain that I am an experienced and careful individual. This time I will be a little more direct: I am not an idiot or newbie. No, there weren't any capacitors or heatsinks or otherwise that were installed incorrectly. Never-the-less I appreciate your time & thoughts.

It's still preliminary (only ran for five seconds with nothing but the PSU, CPU, RAM installed) but the system now runs.

What did I do ? Reinstalled the original two Nichicon 4700 HD Series capacitors. That was all, that was it.

Now what I need to do is to insure that a repeat of the previous session doesn't happen again and the original caps aren't burned-out (and possibly the board with it)... so still need to precisely understand what went wrong the first time.

PCBonze, your help has been invaluable. Once I have a theory as to what happened the first time, I will run it by you.

**PCBONEZ** · 04-01-2008, 11:10 PM

Re: BlownRecap: Ripple Current ?

KC8ADU wasn't trying to insult you. Everyone makes mistakes.
People in here remind each other to check the 'obvious' things all the time.
- It's easy to get the polarity backwards on a cap. He was just saying check-it.

Besides, it's MY job to annoy people.
I don't do it on purpose, it's just my noisy sarcastic attitude.

Given what you described I think maybe one of your new caps is a dud.

.

**linuxguru** · 04-02-2008, 03:16 AM

Re: BlownRecap: Ripple Current ?

I don't think the 4700uF FMs were the problem - however, it appears that you have managed to revert to a working configuration by reinstalling the HDs, so there was something about the recap that was probably dodgy. I'd suspect a hairline solder bridge from one of the FM cap pads to ground, but the point is moot.

Just FYI, Nichicon HDs are generally good - it's almost never necessary to replace them unless there's a visible sign of failure. All failures reported so far are *Taicon* HDs, which are a different beast, even if they have identical specs and formulations on paper.

**PCBONEZ** · 04-02-2008, 01:07 PM

Re: BlownRecap: Ripple Current ?

I agree.
Either a bad new cap or a soldering goof. (Bridge or bad joint.)

And also Nichicon HD rarely go bad and so usually don't need to be replaced.
If it's working with them then just leave them in and call it good.

.

**Intuit** · 04-02-2008, 09:37 PM

Re: BlownRecap: Ripple Current ?

Thanks for the responses guys/gals. Sure it runs, but the question is, for how long ? Last time there was no indication that there was any problem until it ran for several minutes and suddenly failed. That's what I need to guard against and why I fired it up for only a short time.

I've been reviewing the data over & over and had initially came to the bad-cap liklihood as well. But science doesn't operate by simply going with the most likely conclusion, but instead, by eliminating most other possibilities. For the time being, that's what I simply don't have enough knowledge (and tools) to do.

Would be helpful to be able to test them individually to find out which one of the two, or if both have failed; but the needed tester over at Grainger.Com runs around $60 and above. May purchase it later, but not in the budget now.

The going theory or observation here is that the cap(s) overheated. Question is, what would've caused it/them to overheat ? What would feed them too much current, or draw too much current from them ?

There are around 50 capacitors total that have been replaced with higher grade capacitors that offer not just initially less resistance, but increasingly less resistance as they heat up. Does less resistance automatically mean increased current ? They also have an ability to discharge faster, have more capacity but have greater stability over temperature differences. Individually they would be well within tolerance, but what about cumulatively ? 50 are a lot of caps. Have the last seven, put it over the edge or pushed it too far from design spec ? Would the coils on the coils on the adjacent inductors need to be adjusted ?

What are the consequences of too little resistance (or too much) in the typical DC or filter setup ?
What happens when too much current is run through a cap ?
What do higher frequencies do to cap capacity ?
How does decreased resistance affect cap capacity ?

One capacitor sits in the way of the RAM slot latch and had to be setup beside it's normal location, extending the leads to proper holes. Confirming that longer leads wouldn't take that cap out of spec ?

===============
Don't read too much into what I stated... not foaming at the mouth or anything. Thouroughness, that these things have been mentioned, is not annoying, but repetition can be. Again, worked in an electonics plant soldering, tweaking and QC-inspection for two years. Trust me I'm not a newb when it comes to soldering and inspection. I know how to maintain a clean iron and how not to overheat components, leave solder balls, etcetera. True everyone makes mistakes which is why an experienced professional always knows to look for them. ...and look for them again and again. "Measure twice, cut once." to borrow an old carpentry saying. Mistakes are very expensive in an electronics manufacturing facility. This board has close to 50 capacitors and had replaced all but the last 8, which 7 were Nichicon caps, better than a year ago with zero issues. Wire brushes are also very handy with solder balls and don't need a magnifying glass due to my myopia. I explained part of that and still I get the same non-progressive responses. Repetition, not thouroughness, is what generally annoys people.
==============

**Intuit** · 04-02-2008, 10:00 PM

Re: BlownRecap: Ripple Current ?

...and no there weren't any surface mount components knocked-off, etcetera... looked for that too...

Announcement

BlownRecap: Ripple Current ?

BlownRecap: Ripple Current ?

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment