Amiga motherboard (A3000) cleanup with Isopropyl alcohol: weird results

[Wol]

da9000

Looks like a Lead compound to me, possibly
some form of lead sulfide or oxide or nitrate.

It will have formed due to chemical reactions with
the battery electorlyte and or the Isopropyl.
just remove it with a Nylon brush eg: Nail brush or
hard toothbrush.


Wol.

[Framiga]

"Couldn't find any How is it going to be different?"

a 30% of "only Jesus knows" remaining stuff!

Anyway... i can provide a RS-Components RoHS IPA 400ml can code ID

cod: 693-775

but not knowing where you are from, i don't know if it could help.

[da9000]

Wol wrote:
da9000

Looks like a Lead compound to me, possibly
some form of lead sulfide or oxide or nitrate.

It will have formed due to chemical reactions with
the battery electorlyte and or the Isopropyl.
just remove it with a Nylon brush eg: Nail brush or
hard toothbrush.


Wol.

I have no idea what it is, but you seem to know better I took it off with exactly that: hard toothbrush
No improvement in my A3000 But that's probably because some trace is broken somewhere or something...

What was interesting is how the white stuff appeared around various pins/holes... as if there was acid inside the motherboard and the alcohol got in there, reacted and pushed the stuff out. I doubt that's the case, as the battery leak was cleaned almost immediately (about 5-6 years ago), but I was a neophyte then so I think I only used some Q-tips with plain water, and not isopropyl. After having sat unused for 5-6 years, it seems the acid kept on eating away at places. I _believe_ it booted one last time (after the 5-6 year hiatus), and then it just started green-screening ever since :~(  ...My first 'migy ...

[da9000]

Framiga wrote:
"Couldn't find any How is it going to be different?"

a 30% of "only Jesus knows" remaining stuff!

Anyway... i can provide a RS-Components RoHS IPA 400ml can code ID

cod: 693-775

but not knowing where you are from, i don't know if it could help.

Heh... If you insist that 30% more pure is better, I'll look for it. I'm in the US so I don't know if I can find that specific one, but I know some electronics shops, so I'll see what they have. Thanks

[Framiga]

the problem now is not the quality of the alcohol anymore.

Now you have to find anf fix the damaged parts. Look at the damaged zone with a magnifier lens (or what you prefer)

I would suggest you to reflow the solders also (obviously if you have the right tools)

PS- for the "next" cleaning work: some Isopropyl alcohol flavour in USA

[da9000]

Framiga wrote:
the problem now is not the quality of the alchool anymore.

Now you have to find anf fix the damaged parts. Look at the damaged zone with a magnifier lens (or what you prefer)

I would suggest you to reflow the solders also (obviously if you have the right tools)

Yes, I've been looking as hard as I can, but tomorrow I will add a high-quality camera with a monitor so I can get even closer and more detail on the board.

I'm not sure if sure if "reflow" means to use one of those "heat guns", but I've re-soldered a lot of the pins/joints on key components around the damaged area: all Agnus, Denise and Paula pins, and some other ICs and components around them. This was all done from the back side of the motherboard. I've also replaced all the soldered Chip RAM (1MB) with sockets and put new (and tested) DIPs. No difference  I've also tested all diodes (zener and not) by removing from the motherboard. Also tested all fuses and they are good. I don't know how to test caps, inductors, ferrite beads and that "EMI" stuff... I also tested all resistors _without_ disconnecting them, and none seemed to be shorted (didn't test exact values to be honest).

I've started looking at the schematics, but there are too many lines and I'm getting confused
My next plan (as far as testing) is to find out how to test certain chips that are related to the Chip RAM (since I get a green screen, and then a black screen after a couple of seconds, and immediately a green screen again), so I can see if they are getting the right voltages at least. I don't have a scope so I can't test to see if they are also getting the right signals :/

Wish me more luck!

[amiga92570]

Use a commercial board cleaner available from an electronics supply. It is usually provided in an aerosol can. THe white stuff is probably oxidation. If you use a acid core solder and do not clean the board thoroughly after repairs this will happen every time. Battery leakage can have the same result. If you have a schematic I would check the affected traces with a ohm meter to verify continuity of the circuit. To start with though I would verify voltages at different points on the board, and check to be sure the oscillators are working and then verify data, address, etc line are active. This will pinpoint what area you should be attending to.

[da9000]

amiga92570 wrote:
Use a commercial board cleaner available from an...

I will try to get some. Any suggestion for places/stores/brands?

I don't know if my solder is "acid core". How do I find out?

I've tested all suspicious looking tracks and they seem ok with continuity testing. Any suggestions at which points I should be testing voltages at? And how can I test the oscillators (I don't have replacements) ? Finally when you say "verify data, address, etc line", are you talking about the ECS chips or the Chip RAM itself?

Thanks

[da9000]

Just wanted to say thanks for all the input guys. I'm going to get some sleep now... been up all night trying to figure this thing out. I'll update this thread as I discover new things.

In the mean time, can someone who's a hardware expert tell us a bit more about the multi-layerd Amiga boards? After looking at the A3000 rev. 9 board with a bright light, at an angle, I could see the middle layers ONLY under the CPU area. Nowhere else. So, if the area under the battery was corroded (as it is in my case), do you really think it has affected something under there that's important? Or not? (from what I can see there's nothing under the battery section)

Thanks in advance!

[X-ray]

I have seen this happen before.
There is a small highly venomous spider called a violin spider, and it has tiny eggs that lie dormant for ages, in certain conditions. It looks like these eggs were in the solder joints and the alcohol has rendered them active.
See those webs there? In about a week or so you'll see tiny juvenile violin spiders emerging from there. They will seek out the one who used the alcohol on them, and bite him in his sleep.
The effects of these bites are tissue necrosis, hallucinations, searing pains and convulsions. Death is usually unavoidable.

All because you bought the wrong A3000 mobo  :devildance:

[amiga_3k]

Now here's just an idea...

You've de-soldered the chip-RAM and replaced it by sockets and socketed versions of the RAM. Now, if you've poured the alcohol (any good drugstore (that's the kind where you can get medicals, right?) should be able to deliver near 100% alcohol) on the component side of the mainboard then normally it should not escape at the underside of it. If it does, there's some leakage - indicating possibly defective soldering. So... I'd say de-solder the chip-RAM sockets, clean the soldering holes to make sure they're free from acid and corrosion. Then solder the sockets back in place, making sure the soldering is 100% okay AND 'rest' on the PCB. That last part is important because while pressing in the Chip-RAM you don't want to add forces to the solder-joints.

[Framiga]

look carefully to VIAs also (perform a continuity check from the bottom side of the VIAs to the top one).

IMHO is theres a damage, is on the PCB itself.

Good luck :-)

[Castellen]

@da9000
Some of these non-booting problems can be really tricky to solve, even with all the right equipment and knowledge.

If you're fairly sure it's related to a leaking battery, then as already mentioned, good approaches are to measure nearby vias and replace ICs which have been heavily corroded.

Don't let something else throw you off, such as a dodgy socket contact elsewhere on the board.  If you haven't done so already, re-seat the socketed devices, particularly the CIAs.  Socket contact problems there can also cause green screen.

If the battery has leaked significantly, it's highly likely to have damaged the U400/U450 (Paula/Denise) sockets.  If the contacts appear green/blue instead of shiny, they need to be removed/replaced.  Attempting to clean the sockets results in intermittent operation at best.  Don't waste your time trying.

To remove the sockets, by far the easiest way and to risk the least PCB damage is to remove the pins one at a time.
Basically you do this:
1. Use a scriber/small knife to bend the outside most "leaf" of every contact sharply towards the centre of the socket.
2. Once every contact is bent away from the plastic edge, it helps to use small sidecuters to trip away the socket's thin plastic lip.
3. Arrange the board so it sits vertical, and one at a time, grip the contact leaft you bent inwards (I find low pressure on small sidecutters works best) while heating the contact on the solder side.  Yeah, you need two hands for this, so don't try it if you're an amputee
4. After a few seconds of heating, the entire contact and leg will pull up out of the top of the plastic socket.  Pull gently on the pin, it doesn't take much force to extract it once the solder has melted.
5. Repeat until all contacts are removed, the plastic body will just fall away by itself.
6. Clean all holes with vacuum desoldering tool/solder wick.

The only tricky part is the ground and VCC pins (2 or 3 of them on each device). Since they're soldered directly into the ground/VCC plane with no thermal relief, you need a *lot* of heat to melt the solder sufficiently.

Sounds time consuming and a lot of work, but once you get the hang of it, the entire operation only takes five minutes.


Assuming you've tested the ICs, clean any obvious corrosion from the legs and just solder directly into the PCB.  If you insist on using one of those stupid sockets, at least use a decent machine pin one, and always use a new one.

It's near impossible to create a "bad" solder joint with plated through hole technology.  Usually the worst thing you can do is have solder shorts, but you've got to be pretty bad to do that with component technology this big!


As for that isopropyl alcohol residue.  I see that a lot as well, no idea what it is exactly, probably lead oxide from the solder.  Don't worry about it too much, it appears to be inert and doesn't seem to cause any long term problems.  Seems to be more cosmetic.
Hard to tell from your photo, but it does seem to be a bit worse than normal?  You can brush it off if it bothers you.

Don't be too concerned about the purity of the IPA.  If you're also using it in your crystal meth lab, then yeah, it needs to be pure :-)  But for cleaning big chunky electronics, who cares?  That's why pure IPA is hard to buy over the counter.  At least here in NZ anyway.
Most of the time I just use methylated spirits (the purple stuff).  Cheap, easy to get, and works just as well.


To answer your PCB questions:
The A3000 board is 4 layers.  95% of the tracking is all on the top and bottom (external layers).  As you noted already, just about all of this is under the 68030.
Elsewhere, the internal layers are basically solid copper ground and +5V planes.

I don't have a copper mask of the board, but I'm fairly certain there is no internal tracking around the RTC area.
Tell-tale signs of internal tracking is often a via hole visible by itself with no tracks running out from it in one or both sides of the PCB.

This diagram shows PCB construction for the A4000, though the 3000 is much the same.  It also shows how you can repair open circuit vias.


Hope that helps.  I can say from a lot of experience that this kind of thing can sometimes be very time consuming and problematic to repair.

[Edit] Damn, that was a big post.  Apologies if Wayne's server runs out of disk space, you can blame it on me :-)

[da9000]

@amiga_3k:
You're correct in that if the solder job is not well done, the alcohol or other liquids might sip through. But what I think happened was that the alcohol didn't sip through to the other side, but because I put so much of it, it just overflowed and went around the board (the board wasn't horizontal the whole time). My solder job is pretty good too, so I had made sure there were no holes or anything of the sort. Of course I can't be 100% sure on that, but I think it's ok (based on other jobs I've done). But right now I can't take that stuff apart, because I have no more solder wick, and it was such a pain in the butt taking the Chip RAM out, to begin with. But even when I took the Chip RAM out, I didn't find ANY bad chips. So the problem was not the Chip RAM. It's someplace else...
Thanks for the idea though


@Framiga:
Yes, I've done a little 'via testing' but not much, because it's hard to get the board to stand up... I'm not a pro, and I don't have the right equipment. But yes, the problem is definitely with the motherboard or the chips I could not test, but I don't think it's the chips because the computer was working solidly. It's the battery the killed it, I'm 99.9999% certain.
"Multo grazzie", I'll need lots of luck!


@Castellen: (<-- this guy is my hero! without him or his site my A4000 would be dead or trashed by now!)
I'll take a paragraph approach on this one:


History:
--------
I believe the problem is due to the leak, as might be evident from the history so far: the machine worked fine for a decade almost. Around 2000 I noticed the battery get fuzzy, but no real spillage. No scars, no corrosion on the mobo. Removed it, cleaned the area with, I believe, wet with water Q-tips and the like. I don't think I used any sort of alcohol, but it's been too long to remember. Left it sitting until 2005 (I know... now I'm paying for it). Turned it on around May 2005, and I _believe_ it booted to some sorta ok state (Kickstart screen or even WorkBench, I forget). Turned it off and then on again and it green screened. As soon as I saw that, I opened it up again and saw that inside it had a few nasty scars, not only where the battery was, but specs of scars around the surrounding area. But they were small scars, like it had drizzled acid. No big scars or corrosion, except where battery negative terminal was. Also, no corroded ICs or sockets, or pins. Just a bit of green/blue (tirquise?) on some pins, like the resistor and cap north of the negative terminal and a couple of the ICs near there (U480, U477, U472).


Typical things:
---------------
I did reseating of all these ICs: 2xCIA, SCSI, Agnus, Paula, Denise, Chip RAM, Kickstarts (2.04), RTC, 4xGAL chips on the Chip RAM side, disabled FPU, removed all Fast RAM for now. Maybe I should do _all_ socketed ICs like Amber, Gary, Ramsey etc? Would any of the ones I didn't reseat cause a green screen? (I don't know which other chips access Chip RAM). I also tested the following on my A500: 2xCIA, Agnus, Paula, Denise. All worked flawlessly.

[EDIT: I resat all socketed chips. I read someplace that Gary could possibly cause green screen].

Socket repair:
--------------
I will use your method, but I prefer to use sockets, so I will go buy some good ones (machine pinned as you said), and remove the existing and stick the new ones in. And yes, the ground/VCC pins sucked big time because of the 'heatsink' effect. Especially when trying to remove the Chip RAM. But for informational reasons: I did re-heat/re-solder (what's the right term?) the actual socket joints for many of the ICs in the area (Agnus, Paula, Denise, part of the video circuitry, RTC and a couple of the surrounding chips). Nothing changed


Advanced terms:
---------------

It's near impossible to create a "bad" solder joint with plated through hole technology. Usually the worst thing you can do is have solder shorts, but you've got to be pretty bad to do that with component technology this big!

This went a bit over my head Can you explain the "plated through hole tech" ? And as for "solder shorts", do you mean because the holes are too close? I've been very careful, testing with my DMM to make sure there are no shorts.

[EDIT: I understand now, and I agree 100%]

Isopropyl:
----------
Yes it looks a bit worse than expected, and at some point I thought it might have to do with the soldering job. I'm not so sure because of one thing: the Chip RAM I desoldered were the last 2 banks, 8 chips. The farthestmost bank (4 chips, 0-512KB) did not have any of the white gunk left afterwards. But the 2nd from farthest bank (4 chips, high 512-1024KB) had tons of it! And it's not like I did anything special when desoldering/soldering the 2nd set... Also of note: the next 2 banks (1024KB-2048KB) didn't have any gunk. Of course these were soldered from the factory and I didn't really touch them.

As for the purity, I agree with ya: I'll keep the high quality stuff for my meth experiments and use the 70% for 17 year old motherboards And the damn alcohol is so expensive for such a common "ancient-tech" thing...


PCB layers:
-----------
Thanks for the "lonely vias" tip! That is really really helpful, so now I can track and see if there is any inner-layer usage near the scars! And yes, of course I've been using your "via repair" page to guide me along this Herculean task

Thanks again for all the info/tips/help Castellen!


I think Wayne's server drives must be chocking right about now!

[da9000]

Just an update:

I checked the A3000 power supply, just to be certain, and I got the following readings. In general they were pretty stable, only the 2nd decimal digit (0.0x) changing once in a while.

Pin  Voltage
---  -------
1    +4.92v
2-6  +5.12v
11   -4.92v
12   +4.93v
14   -11.90v
15   +11.72v

Pins referenced from: http://pinouts.ru/Power/amiga3000power_pinout.shtml

BUT while measuring voltages on the motherboard I found something suspicious. _Most_ (I didn't test all) of the ICs on the Agnus side of the board were getting 5.02v to VCC, except U452 and U451. Those two were getting 4.76v. Could this be a problem?

Another general question: would an A3000 work with 1x256Kbit chips for Chip RAM? Or must it have 4x256Kbit chips? I'm just curious, but I've been using 424258 NECs (details here: http://www.amiga-stuff.com/hardware/256kx4-dram.html), which were there to begin with...

[Castellen]

The voltages mentioned all look OK.  Should be fine for U451/U452 as it's well within operating limits.

I'm too lazy to look at the manual at the moment, but from memory, U541/U452 are just buffering video output data, so nothing too critical.  They'll be running from the video supply, which is just a separate +5V supply from the rest of the logic.

As for the chip memory, just use what's listed to work correctly.
Just because you're getting a green screen labeled as "chip memory error", there are a number of other things which can cause the same problem which is not related to chip memory.  A bad CIA, for example, can produce the same thing.


Quote:
-----
This went a bit over my head Can you explain the "plated through hole tech" ? And as for "solder shorts", do you mean because the holes are too close? I've been very careful, testing with my DMM to make sure there are no shorts.
----

Plated through hole (PTH) simply means that the walls of the holes drilled in the PCB for component legs to be soldered in, are plated with copper.  So instead of having just a top and bottom pad with a hole in the fibreglass, the top and bottom pad are joined together with a tube shape of copper around the outside of the hole.  The tube is chemically plated in during PCB fabrication.

They're good at creating incredibly strong, reliable and vibration proof joints for leaded components.  Connections which are conveniently compromised by using crappy dual-wipe IC sockets.
The down side to PTH is it's a lot harder to unsolder components from.

Solder shorts can sometimes be caused from a "hair" of solder bridging two pads together.  Sometimes by applying too much solder as well.  For some reason, people have the idea that using an entire roll of solder for each joint will make it more reliable or something?!  You only need enough to just fill the hole.

Luckily, the component lead pitch used in the A3000's board technology (2.54mm) is so large (far apart) it's quite hard to cause accidental shorting. You could probably still get reliable soldered joints from a blind gorilla weilding a flame thrower.

Actually, I've just been working on some boards which appear to have been soldered by such a beast, but that's another story...


Good luck with the A3000 repair.  The only thing you can easily do is measure/repair any damaged tracks/vias around the battery area, and replace any corroded sockets and ICs.

Make sure you haven't done anything silly in the process.  I've seen it before where people pull out ICs to reseat them, but end up bending pins when re-inserting them into sockets, so the legs bend up or get broken off.