Corrosion removal - Motherboard with leaky Ni/Cd battery

Corrosion removal - Motherboard with leaky Ni/Cd battery
Posted by CkRtech on 27 Mar, 2017 21:54
Hey all -

I am currently working on the restoration of an old computer that had its CMOS battery leak over surrounding components. Using both a soft and medium toothbrush, I have scrubbed it with ISO, added some baking soda to ISO for abrasion and scrubbed again, and finally scrubbed it with some dish soap and tap water. I have an image below that shows a before photo in the top half and an after photo in the bottom half.

You can still see corrosion at the joints on the components and especially on the large traces. I imagine I will eventually scrape off the large traces, but I would really like to get this thing squeaky clean so I can start repairs.

Would appreciate any feedback on where to go from here. I have not tried a white vinegar or vinegar/water mixture scrub yet.

With battery still in place at time of discovery:

#1 Reply
Posted by BradC on 28 Mar, 2017 01:46
Once you've got the electolyte properly neutralized and cleaned up, I've had fairly workable results painting on a layer of 100% Deoxit and leaving overnight. It works just the same for this as it does for contact oxidation and the oxide layer slowly floats to the surface. A good spray with CRC CO cleaner and another scrub with IPA and a toothbrush generally leaves copper clean enought to solder to.

I'd be concerned about via penetration and any internal board damage on that motherboard, but you can only do what you can do. Best of luck. Better you than me!

#2 Reply
Posted by VK5RC on 28 Mar, 2017 11:22
For moderately aggressive cleaning in tight spots - Sometimes use a fibreglass pencil abrasive brush - BUT don't get the splinters/dust in your eyes or skin or you will hate me for weeks! e.g. http://www.ebay.com/itm/2PC-5-Jewelers-Nylon-Fiberglass-Scratch-Brush-Pen-Type-with-extra-Refill-/222213619987. I often revisit the area a month or two later to ensure I have got most of it - I am not sure you really get rid of it all forever .
I have tended to use distilled water initially as I guessed the 'goop' would likely be water soluble more likely that lipid soluble and my IPA is 99%.
Robert

#3 Reply
Posted by tecman on 28 Mar, 2017 12:43
Based on the pictures, your chance of recovering the board are slim at best. Connectors are gone, and damage and conductive leakage paths under chips, sockets, etc will frustrate you for ever.

paul

#4 Reply
Posted by Ian.M on 28 Mar, 2017 13:12
NiCd battery electrolyte is Potassium Hydroxide, which reacts with the Carbon Dioxide in air to form Potassium Carbonate. The corrosion residue will therefore be a mixture of copper oxide and copper carbonate on the tracks, and Tin and lead oxides and carbonates on the solder joints. These corrosion products are not soluble in water, alcohol or non-polar organic solvents so will have to be removed mechanically or chemically.

You can use dilute Acetic or Citric acids to remove copper carbonate deposits - DO NOT use mineral acids and DO NOT add salt! (organic acids + salt = PCB etchant, and residual chloride contamination can be very destructive to copper).
EDTA would clean off the remaining oxides. Rinse between chemicals, and afterwards wash first with dilute detergent, then plain water, then distilled water, then shake as much water off as possible and rinse with IPA.

I think the edge connector nearest the battery is FUBARed, and you'll have to do a lot of careful cleaning round the RTC crystal, maybe even replacing components. Any PTH vias that have been contaminated are suspect and will probably need to be repaired by soldering a wire through them.

#5 Reply
Posted by carl_lab on 28 Mar, 2017 15:07
You can try to use a toothbrush and toothpaste to remove the oxidation residue.
Then use a clean toothbrush and IPA-water-mixture to clean the PCB (or use an ultrasonic bath).

#6 Reply
Posted by CkRtech on 30 Mar, 2017 01:59
Quote from: Ian.M on 28 Mar, 2017 13:12
NiCd battery electrolyte is Potassium Hydroxide, which reacts with the Carbon Dioxide in air to form Potassium Carbonate. The corrosion residue will therefore be a mixture of copper oxide and copper carbonate on the tracks, and Tin and lead oxides and carbonates on the solder joints. These corrosion products are not soluble in water, alcohol or non-polar organic solvents so will have to be removed mechanically or chemically.

You can use dilute Acetic or Citric acids to remove copper carbonate deposits - DO NOT use mineral acids and DO NOT add salt! (organic acids + salt = PCB etchant, and residual chloride contamination can be very destructive to copper).
EDTA would clean off the remaining oxides. Rinse between chemicals, and afterwards wash first with dilute detergent, then plain water, then distilled water, then shake as much water off as possible and rinse with IPA.

Wow! Thank you for the chem specifics listed in this response. I'll line up the various chemicals and liquids needed and follow the order. The last thing I did was attempt to use CRC 2-26 (lubricant) to see if I could scrub off the corrosion. I didn't really have much luck with the scrubbing. I used contact cleaner afterward to clean the 2-26 off. I assume I can reset with the rinse, dilute detergent, plain water and then begin the citric, rinse, EDTA, rinse, clean-up.

At this exact moment, the last thing on it was contact cleaner.

Quote
I think the edge connector nearest the battery is FUBARed, and you'll have to do a lot of careful cleaning round the RTC crystal, maybe even replacing components. Any PTH vias that have been contaminated are suspect and will probably need to be repaired by soldering a wire through them.

I have desoldered the 8-bit ISA slot CN7. It isn't critical - I might have some parts boards that I can use to source a replacement. I am not sure what the speed of that crystal is - but I imagine there are some standard values (standard value with a divider for ISA bus speed, perhaps). I will do more research to see what crystal I need. Don't know if I will bother with a battery until after I reach a point where I can test.

I assume any of those vias that were used as battery leakage drain hole are worthless at this point. I have insulated kynar 30 awg for the traces, and I can jump through the vias with it or make necessary jumper wire soldering on a single side as needed. I was going to use something thicker (I have stranded and solid 22 AWG) for the big traces for the -12V, -5V, etc power delivery.

I will most likely at the very least desolder and lift one side of some of those support components next to the crystal and beyond as they have some traces underneath that will need some work.

#7 Reply
Posted by Ian.M on 30 Mar, 2017 02:26
The other issue is the probability its a multi-layer board. If it has any buried routing layers, it may well be BER due to via corrosion. If they are just power and ground planes, you may be able to patch it up well enough by jumpering to the nearest point you can pick up the lost power or ground plane connection.

#8 Reply
Posted by CkRtech on 30 Mar, 2017 02:58
Using household items, the plan is vinegar to remove copper carbonate deposits, and toothpaste (or baking soda) to remove oxidation (with the obvious rinse between). For diluted detergent - I scrubbed it with Dawn (a.k.a. Fairy liquid) when I first got started on it a few days ago. I assume I could use that as well.

Vinegar is diluted acetic acid - The concentration can apparently vary between 5%-20%. Is it necessary to dilute it more, or will straight-up vinegar be OK? Worth scrubbing immediately after adding - or OK to let it soak a bit first?

Forgive the obvious/simple questions, but I figured I might as well ask.

#9 Reply
Posted by Ian.M on 30 Mar, 2017 03:34
I'd work at a small area with white vinegar and a fibreglass brush and see how it goes. It should be dilute enough as-is. If you've got concentrated spirit vinegar, I'd start by diluting it to about the same concentration as normal vinegar (5%). You dont really need to remove well attached black oxide except from where you need to solder. Once clean, dry and lacquered, its reasonably inert

#10 Reply
Posted by Armadillo on 30 Mar, 2017 03:51
It's surprising how the leak can propagate the distance and creep up into the SIL sockets.

I would suggest that the SIL socket and the IC be desoldered. Those data vias need attention.

#11 Reply
Posted by CkRtech on 30 Mar, 2017 22:33
Tried it with a toothbrush (since that is what I had), and it wasn't good enough. Going to pick up a fibreglass brush to add to my repertoire.

Armadillo - I cropped the photo quite a bit in the original post that showed the comparison after some alcohol cleaning, but the keyboard controller IC and 8-bit ISA socket have been removed. My plan is to use a dual wipe 40 pin socket where the IC was, sand the IC pins, alcohol clean them and place it back in the socket.

Picture included here before any scrubbing.

#12 Reply
Posted by Ian.M on 30 Mar, 2017 22:45
Sanding the pins wont help reliability. If you do that, you will need to re-tin them all, wick off excess then flux, reflow and wash off the flux residue to get a smooth surface that wont fret in contact with the socket.

#13 Reply
Posted by BradC on 31 Mar, 2017 00:10
Quote from: Ian.M on 30 Mar, 2017 22:45
Sanding the pins wont help reliability. If you do that, you will need to re-tin them all, wick off excess then flux, reflow and wash off the flux residue to get a smooth surface that wont fret in contact with the socket.

I'd solder it into a turned pin socket and plug that into the socket on the board.

#14 Reply
Posted by james_s on 31 Mar, 2017 00:38
That crystal is going to be a 32.768kHz clock crystal for the RTC which is powered by that battery that leaked.

You're going to want to remove any ICs that have corrosion and clean out under them. Be careful because the corrosion does something to the solder that seems to raise the melting point and make it harder to remove without damaging the board.

#15 Reply
Posted by CkRtech on 31 Mar, 2017 22:57
I have a stack of 386 motherboards in the garage, so I pulled off some parts. Looking to replace the ISA slot, crystal, and most likely keyboard connector as well since I have the parts available. It would be nice to verify that this crystal is the appropriate value for the one I am looking to replace. The board I am restoring actually has two barrel style crystals on it.

Quote from: Ian.M on 30 Mar, 2017 22:45
Sanding the pins wont help reliability. If you do that, you will need to re-tin them all, wick off excess then flux, reflow and wash off the flux residue to get a smooth surface that wont fret in contact with the socket.
Ahh. Well I saw an operation like it in a YouTube video (https://youtu.be/Aj0Qbrs_3Gk?t=31m25s) with fine grit sandpaper and some q-tip alcohol cleaning afterward. I assumed I could do that just to get a little of that corrosion off the pins that were closest to the battery. (Second attached photo)

#16 Reply
Posted by CkRtech on 01 Apr, 2017 23:14
Wow. That brush really helped. I knew it would take elbow grease. What I didn't think about was how much it would destroy the brush as I went. I am sorry I didn't order replacement fiberglass brushes for the thing when I did. Still - the largest area of corrosion is (mostly?) gone. The liquid in the photo is denatured alcohol.

I will most likely remove a few more components - maybe even that entire 16 bit ISA slot (CN6 just off camera) in order to check that *one* trace that leaves pin B02 (Reset DRV) and heads toward CN7. I'll remove the resistors, diodes, jumper, keyboard connector and check the traces there.

I hope I get this thing restored. It is certainly a demanding project.

Photo is from what I will call "Round 1" of fiberglass brush scrubbing with vinegar.

#17 Reply
Posted by CkRtech on 06 Apr, 2017 02:10
I have removed most of the components from the area near the battery. I still have a bit of scrubbing to do in places, but I was wondering what components (aside from the ICs' legs) you guys would be concerned about?

I have plenty of 1/4 watt resistors to use as replacements, but I don't know about the diodes and inductors. If the diodes test properly, I can attempt to clean the legs and put them back in circuit. I have no idea about the inductors.

I also haven't addressed the legs of the ICs yet.

#18 Reply
Posted by james_s on 06 Apr, 2017 06:31
I would replace any parts you've already removed, provided new ones are available. I mean you've already got them out, may as well put new ones back in if the ones you took out are corroded.

#19 Reply
Posted by Armadillo on 06 Apr, 2017 07:37
Quote from: james_s on 06 Apr, 2017 06:31
I would replace any parts you've already removed, provided new ones are available. I mean you've already got them out, may as well put new ones back in if the ones you took out are corroded.

+1; might as well do it.

The ram sockets look new. Nice clean job.

#20 Reply
Posted by Ian.M on 06 Apr, 2017 08:43
One thing that has been bugging me: What's so special about this mid '90's 486DX ISA motherboard that it is worth heroic restoration?

Personally, I'd try to get it in runnable condition ASAP with minimal further work. You'll need to put back the keyboard controller (in a socket) and some of the SIMM sockets - the board manual should tell you which are essential - and some of the passives. If it goes to the keyboard circuit or is part of the RTC circuit you can probably omit them for now.
It should boot as far as the BIOS screen and fail with a RTC or keyboard error. If it gets that far,start rebuilding the keyboard connector area so you can fully boot it and the RTC circuit so you can test that.

Its probably not worth the risk of desoldering the next slot over to check the iffy trace if it boots and runs. The vinegar and the thorough washing will have neutralised any remaining alkali, and a week in a warm airing cupboard or an hour in a low electric oven or on a board preheater (either 95 deg C) to ensure its bone dry, followed by running in thinned clear PCB lacquer applied with a toothpick to the suspect trace with the board tilted to seal the remaining corrosion, should be good enough. You can always go back and pull it if you have to.

#21 Reply
Posted by CkRtech on 06 Apr, 2017 09:18
Ha. So I bought this system new in 1993. My plan was to rebuild the 486 as it had been plucked of drives and other components, handed down to a sibling in 1996, and then placed in the attic/storage for a couple of decades.

I didn't think I would be combining my electronics hobby with my vintage computing hobby, but the battery took its toll.

I suppose I could clean off the components I removed as best I could and put them back in circuit, but the plan is to add quite a bit to this system and use it on a semi-regular basis.

If I did want to replace the components, I am not sure about the diodes and inductors. If you think they'll probably be fine, I suppose I could move on. I do want this to continue to work for years into the future (if that makes a difference)

#22 Reply
Posted by Ian.M on 06 Apr, 2017 09:23
If you can ID and obtain an equivalent component, even if you have to pull it from your pile of scrap boards - *REPLACE*. This is especially important for small inductors due to the fine wire inside them.

#23 Reply
Posted by CkRtech on 09 Apr, 2017 03:23
OK. I got my replacement fiberglass brushes in, removed a few extra components, went to town on the traces & scrubbed - vinegar, dawn, alcohol, oven @ 95 C. Pics included. All the through-hole stuff that was a bit dark got a single pass for the inevitable resolder of fresh components. I was happy to see that shinny copper (he said somewhat sheepishly after a vigorous scrub).

Hopefully it wasn't a big deal for me to go that far with cleaning. You even mentioned minimal further work. That was before I said I wanted to run on a semi-regular basis. If I created more work for myself, that is fine...more in a bit.

I was happy to see that the traces underneath it all were in better shape than I thought.

So now my plan/questions are -
1: The top trace (-5V) is mostly gone rather than a retainer of (ideally) neutralized corrosion. My plan was to take some solid strand 22 awg and bridge the top trace from under the BT1 label over the destroyed section and solder the other side to a point just prior to the mounting screw hole - join the strongest points of what remains of the trace. Should I still add PCB lacquer to the destroyed area running underneath the wire, or does it look clean enough now?
2: Since I have shinny traces exposed now, should I use PCB lacquer as the new protective coat or use something else/take another route to protect it?

I imagine I could have stopped and left the blackened traces as they were, but I honestly wanted to get it as clean as possible.

3: What sort of timeline am I on to cover the traces, and should I? (I assume yes because of oxidation)
4: Side note regarding battery - probably going to solder a coin battery holder to BT1 and leave out diode CR5 which is used to provide a charge to the old barrel battery. I'll take alternative suggestions into consideration, but that is not the main point of my original thread topic.

#24 Reply
Posted by Ian.M on 09 Apr, 2017 04:21
1. I'd use copper foil to repair the wide trace - you can do a neater looking job of it than with wire. The other traces can be repaired with appropriate thickness strands of tinned copper wire. Tin all the traces you need to repair first while yo've got good access, before doing any other repair work.

2,3. Yes, PCB lacquer would be good, applied *AFTER* all other repair work is complete and the board has been de-fluxed. You will need to glue some of the repair wires down anyway and in addition to protecting the tracks, the lacquer will also help bond the repairs in place. If you are going to put this project aside, a light film of liquid Rosin R type flux (not RMA or RA, or gel flux) or pure 'white' Rosin dissolved in alcohol applied locally to the tracks you have cleaned to bare metal will dry to a coating that retains solderability for months or even years, but you'll need to wash it off again with alcohol of flux remover before lacquering it.

4. I wouldn't use a board mounted coin cell - there's no guarantee it will even work with your board as the original battery was 3.6V, and if it does work, you may find it needs a new CR2032 every three moths!

I don't know if the four pads I see the other side of C31 from the 14069 footprint are for an off-board clock battery - if they are an off-board 3xAA holder with a series diode to provide anti-charging protection and drop the voltage a bit would be the best option. Alkaline AA batteries will keep the CMOS data and RTC running for *YEARS* and with their holder Velcroed to the case, its very easy to keep any potential leakage well away from the circuit boards. Worst case, you could reinstall a new 3.6V NiCd, but if yo end up doing that, fit it after lacquering the board, stick a rectangle of Kapton tape over the battery footprint extending a half inch outside it, , smear a little lacquer over the tape edges to seal any small gaps where the tape doesn't lie perfectly flat due to tracks and/or repair work, then pierce the battery mounting holes with a needle so the pins are a tight push fit through the tape before you solder them. With a drop of lacquer applied after soldering where each pin goes through the tape to seal them, the board will be well protected against any future leakage.