'Historic' Motherboard with 'Bad Caps' repaired

[Dr. Frank]

Hi,

I got a failing, 'historic' PC to repair.
The MoBo is from DFI, an absolutely no-name brand from Taiwan.
It's got ridicluous 500MHz, 256MB, and still runs on Windows 98 (w/o internet connection, of course).

But I still need it urgently, because it still has ISA slots, which support an old GPIB card, and my acquisition program under Turbo Pascal is still running.
I did all those nice measurements with this PC, which are on display on the EEVBLOG forum, e.g. LTZ1000, 3458A, 5442A, VHP202Z stability.


The PC won't start up from cold, or only after several retries.

I checked the ATX PSU, but that was ok.

Inspecting the board itself, many of the 1000µF/6V electrolytics were blown, or their case were bulged out.
That is a very obvious visual root cause, I think.
The capacity of the caps also were reduced greatly.
I have no ESR meter at hand, but I assume that they are high ohmic now.

So I got  replacement caps; 1000µF/16V, 105°C, automotive grade from FROLYT (small German manufacturer).
They are bigger, and also have a different pitch.
But anyhow, their ESR should be lower, due to the higher nominal voltage.

The dis-assembly was very difficult, due to multilayer PCB with big ground areas. See photo of the not-yet-cleaned solder junctions..

I needed 2 strong solder irons, one soldering pump iron from one side of the PCB, and from the other side a 300°C iron, to disassemble the Bad Caps and clean the vias.

The assembly of the new caps was also quite difficult, as the soldering temperature was partly too low, that the vias would suck the new solder into themselves.

But anyhow, the repair was successful, the board now starts without any problems.


Afterwards, I re-discovered the old site http://www.badcaps.net/forum/showthread.php?t=425, which described exactly that problem.

So, in the beginning of the 2000s, several Asian manufacturers were hooked on a bad electrolyte technology, which was presumably "stolen" from another company but with the wrong recipe.

So I looked closer on the failing cap, manufacturer was TAYEH, which is on the list of companies, which produced those failing caps.
And the MoBo is from 2001, or so.. Bingo!

A very interesting industrial story also, I think.

Frank

[wraper]

Frankly saying those capacitors were not too bad if managed to survive 13 years.

[XOIIO]

Agreed 13 years isn't bad.

But, if it's a no name motherboard how is it historic in any way?

[Dr. Frank]

Agreed 13 years isn't bad.

But, if it's a no name motherboard how is it historic in any way?

13 years old- W98 - 500MHz.. if you don't call THAT historic in the fast progressing PC business..

Forgot to mention, that in this PC I still have 5 1/4" and 3 1/2" floppy drives running, for the case that I ever have to access those old formats again. (Perish the thought!)


Well, 13 years endurance for 105°C electrolyte capacitors is not that good either, especially as they obviously fail systematically (9 off 13).
And we've got a real root cause for that failure..
 
I have old instruments here from HP and Fluke, with electrolyte caps ok after 20... 35 years.

Frank

[XOIIO]

Agreed 13 years isn't bad.

But, if it's a no name motherboard how is it historic in any way?

13 years old- W98 - 500MHz.. if you don't call THAT historic in the fast progressing PC business..

Forgot to mention, that in this PC I still have 5 1/4" and 3 1/2" floppy drives running, for the case that I ever have to access those old formats again. (Perish the thought!)


Well, 13 years endurance for 105°C electrolyte capacitors is not that good either, especially as they obviously fail systematically (9 off 13).
And we've got a real root cause for that failure..
 
I have old instruments here from HP and Fluke, with electrolyte caps ok after 20... 35 years.

Frank

Historic implies it had some sort of meaningful purpose, or something noteworthy. I have an 88 or so year old typewriter but it isn't historic. If it had written some important document maybe, but that's doubtful.

[mariush]

I don't know if you were sarcastic or making a joke, but DFI wasn't really a no-name brand, they actually did some quite excellent boards at a time.. look up DFI Lanparty.

They've stopped making them around 2009 and pretty much exited the desktop boards and focus on industrial stuff.

Yeah, old board, old capacitors, it happens. Good on you for managing to fix it with some crappy soldering irons.

[nctnico]

For these kind of jobs you really need hot air to pre-heat the PCB.

[David Hess]

So I got  replacement caps; 1000µF/16V, 105°C, automotive grade from FROLYT (small German manufacturer).
They are bigger, and also have a different pitch.
But anyhow, their ESR should be lower, due to the higher nominal voltage.

Higher voltage will not necessarily make their ESR lower but should increase their ripple current rating.

The dis-assembly was very difficult, due to multilayer PCB with big ground areas. See photo of the not-yet-cleaned solder junctions..

I needed 2 strong solder irons, one soldering pump iron from one side of the PCB, and from the other side a 300°C iron, to disassemble the Bad Caps and clean the vias.

The assembly of the new caps was also quite difficult, as the soldering temperature was partly too low, that the vias would suck the new solder into themselves.

This is a common problem.  Even with my high power temperature controlled soldering iron and largest soldering iron tip for high heat capacity and thermal conductivity I still end up "walking" the capacitors out.  Then I use a small drill to remove any solder from the hole without damaging the plating.

[rob77]

i think i wrote this several times in other threads.. but here is the method i'm using.

1. clamp the board with capacitors facing down
2. hot air pre-heat to just below the melting point of the solder
3. while hot air still on - touch the leads of the cap with the soldering iron
4. the cap will fall off the board by gravity alone - leaving nice clean holes

to mount the new caps - just stick them in while the board is hot (hot air not needed at this stage) and solder them with your iron - the vias will suck-in the solder like crazy

and regarding using the word historic for an 13 year old computer... i think it's ok even if the computer doesn't have any historic value   for example i'm using the word "ancient" and actually not a single piece of electronic could be "ancient" by definition

[babysitter]

But I still need it urgently, because it still has ISA slots, which support an old GPIB card, and my acquisition program under Turbo Pascal is still running.

Ah how I miss TP, no longer having a system that runs it naturally. Times got bad when one had to start patching the delay-bug afflicted files.

Also, points for still liking ISA, where one could really watch what is going on. 8 or 16 bit ?

(And gratulations for the MoBo fix - I only had a DFI b/w hand scanner, of course also with a ISA card...)

[Dr. Frank]

Ok, placed some quotation marks around 'historic', so that some deadly serious people are not disturbed any more... 

I don't use crappy solder irons,  ,  they are from Weller, but I really had to pick the big one. And I don't own a hot air station, yet.

Well, in the datasheets for the same electrolytic capacitor type (EKS in this case), higher rated voltages usually are specified with lower ESR (@100Hz), and lower tan delta.
Lower ESR in turn leads to higher ripple current ratings, as the ripple current heats the capacitor by the ESR.
The 2nd effect is, that higher voltages imply a bigger case size, therefore bigger volume & surface for better power dissipation.

Things may be different at higher frequencies, where the property of the electrolytic fluid comes into play.

Frank

[free_electron]

sheesh. it has the old AT style and ATX power connectors. it is ancient !
probably has a 5 pin din keyboard connector too...

[David Hess]

I use a good Weller and also have a big 100 watt temperature controlled one for really tough jobs.  My Weller vacuum desoldering tool lacks the heat capacity for this but works great on DIP packages where there is no copper ground plane.

I think it just depends on the capacitor series.  Usually there is a sweet spot where the ESR is lowest at a given rated voltage.  160 V and higher rated capacitors usually or always have a jump to a higher ESR.

When replacing capacitors in older equipment, I usually select higher rated voltage ones just to get an equal case size, lead spacing, and higher ripple current rating.

[nukie]

Soldering multilayer PC motherboard is like soldering a huge sheet of coppers. You have a soldering tip with a small contact area pumping lots of heat and spreading out rapidly, it just won't happen. The temperature drops as soon as your iron is off. The heat will never get to the otherside of the board. Sometimes you get lucky with two iron and a pair of clamping pliers as a weight, let gravity do the job as the solder melts but mind the feet below.

Preheating to 150C is a good practice. It takes less time and reduce load on the soldering iron. The ceramic capacitors around the area also get a good dose of de-aging to regain full capacitance.

Well you can thank me later after google, if you dont already know that ceramics age too.

If the electrolytic caps live in a happy cool environment, spec according to handle the ripple current and voltage derated, i bet you they will last 30 years no problems. I have a few in my old Fluke.

[coppice]

I don't use crappy solder irons,  ,  they are from Weller

If its less than 5 years old I'd say a Weller is crappy. The handles seem to crack if you sneeze on them these days.