LCIII Recap - Apple design fault -47uF reversed!

[paul.gaastra]

Uniserver said

Can some one please take a nice picture so people can visually see how it should be and post it in this LC-III Caps! thread?

It's no use me taking a photograph of the board I recapped correctly because I put the tantalums on upside down and soldered fuse wire to the pads. So you can't see the capacitor markings. I do that because I am afraid of shorting the pads to the ground plane. Generally the terminals of SMT tantalums are wider than the pad of the original electolytics. It makes for a lot of work and really makes a mess.

 

[RickNel]

how come everybody's LC III logic boards haven't blown up?

I've been thinking about that. This rail serves the RS422 ports, driving the negative half of the TX and RX differential circuits. The positive sides of each signal are correctly filtered. So I'm guessing that the filter isn't essential but is there as standard design practice to protect the signal when it is close to limit of specs - say with a very long cable to a printer or modem. It's fairly low current and short duty cycle and the overvoltage on the reversed cap might even be within, or not far outside, the tolerance of the original electrolytics. The ESR generates heat, but the aluminium cans might dissipate this better than the replacement poly tantalums, and so last indefinitely with moderate or low usage.

It could be interesting to test a machine in the original state, just by finger touch on the cap, to see whether the reversed cap is noticeably warmer than its neighbours when in use, say with a printer attached to the port. I don't have one of the relevant machines to test myself.

I'm an amateur at this, and would be happy to hear opinion of a proper EE. :-/

Rick

 

[olePigeon]

I think it'd be fun if one of the original LC engineers stumbled on this thread and commented on it. I wonder if they knew about it and didn't bother fixing it, or if this is something new that no one noticed before? Certainly news to me.

 

[Gorgonops]

Someone mentioned before on this thread that it's common for Electrolytic and Tantalum capacitors to have opposite polarity markings, are we *absolutely positive* that doesn't explain the confusion here?

I *seriously* doubt these machines shipped with an electrolytic capacitor that was put in backwards. Electrolytics have this nasty habit of blowing up like bombs when run in reverse (seen it in the flesh several times), and even if by some miracle that didn't happen it'd still heat up and act like a (partially) shorted circuit. This is the sort of thing the manufacturer would have noticed after shipping a few thousand units.

 

[dougg3]

Someone mentioned before on this thread that it's common for Electrolytic and Tantalum capacitors to have opposite polarity markings, are we *absolutely positive* that doesn't explain the confusion here?
I *seriously* doubt these machines shipped with an electrolytic capacitor that was put in backwards. Electrolytics have this nasty habit of blowing up like bombs when run in reverse (seen it in the flesh several times), and even if by some miracle that didn't happen it'd still heat up and act like a (partially) shorted circuit. This is the sort of thing the manufacturer would have noticed after shipping a few thousand units.

I thought the same way at first -- we have to be thinking something wrong -- but look at the multimeter picture closely. There's -5V across that capacitor. And bbraun confirmed the same capacitor orientation on his LC III.

It's backwards. Unless Apple had a special batch of electrolytics made that had the black mark on the positive side...and used it only on that capacitor...

I don't understand why it hasn't popped either. My understanding is that electrolytic capacitors in reverse can explode nastily. Is there any explanation? Not enough current available from the -5V rail to do any harm? Maybe the original capacitor, even though it has a polarization mark on it, happens to be bipolar? I'm just trying to think of any rational explanation other than "it's backwards" and I'm coming up blank.

 

[techknight]

-5v isnt really high enough to blow a lytic. at least I havent blown one yet in reverse at such a low voltage. Especially if the working voltage of the capacitor is rated alot higher than 5V.

 

[RickNel]

Yes, I think we may be seeing a situation where an electrolytic is more tolerant than a tantalum. Still waiting for any report of this reversed electrolytic actually failing in normal use. If there are cases of it failing before the others on the positive rails, we could infer that it had been carrying a marginally greater heat load because of the ESR from the reversed polarity.

Rick

 

[techknight]

And that is one thing I never learned in electronics, why electrolytics caps have a polarity. While some do not.

 

[uniserver]

could this be a solution?

http://www.mouser.com/ProductDetail/Taiyo-Yuden/EMK325BJ476MM-T/?qs=sGAEpiMZZMuMW9TJLBQkXnpMKdRe6%2fORaxesfjtxlnQ%3d

100: $0.735

Multilayer Ceramic Capacitors MLCC - SMD/SMT CAP MLCC 1210 16V X5R 47uF 20%

 

[paul.gaastra]

Brilliant, uniserver! A non polarised replacement!

I'm no expert on whether the esr is OK on those, but it does say it's low. I am continually amazed at how technology is advancing. We would never have thought of getting a 47uF capacitor in ceramic just a few years ago.

However, it's no better solution than putting a polarised electrolytic or tantalum in with the correct polarity.

Maybe another solution is to leave C22 off the pcb.

Maybe someone like the techknight has an oscilloscope and can put a probe on the -5v supply when two of his Macs are transfering a file on Appletalk and see if the supply is degraded at all with no C22. I've seen his YouTube videos. He knows what he's talking about.

 

[James1095]

I noticed the same thing when I re-capped mine. I happened to be looking at the traces and thought something seemed odd, so I measured it and sure enough, the marking on the board is backward. I had already removed the old capacitor so I'm not sure whether it was installed correctly or not.

There is a similar error on the Electrohome G07 monitor used in a lot of 80s arcade games. One of the capacitors is marked backwards on one side of the board and it was never corrected throughout the production run.

 

[armadillu]

I noticed the same thing when I re-capped mine. I happened to be looking at the traces and thought something seemed odd, so I measured it and sure enough, the marking on the board is backward. I had already removed the old capacitor so I'm not sure whether it was installed correctly or not.

There is a similar error on the Electrohome G07 monitor used in a lot of 80s arcade games. One of the capacitors is marked backwards on one side of the board and it was never corrected throughout the production run.

Well, brining some life back to this thread. I just recapped an LCIII with tantalum capacitors (blindly following what's marked on the board); and guess what... The machine wouldn't start, so I started testing voltages and my C22 tantalum caught on fire!

 

[Callan]

I can confirm you have to change the polarity of c22. It's -5 rail and for some reason the mask has you putting it on backwards. They must have installed the cap reverse of what the mask says or used a bipolar cap at the factory. It would have never worked otherwise. Saying that... it's strange your cap caught on fire. Normally it just pull the -5 down to -2.2 (roughly - and yes... i learnt this the same way). Did you test the supply out before installing it? It needs a load to test properly, but you can test to see if your in the ballpark. Also... what was your tantalum voltage rating?

(As everyone has stated before me - just started reading the thread)

 

[David Cook]

... The machine wouldn't start, so I started testing voltages and my C22 tantalum caught on fire!

Yup. There is a popular YouTube recapper that I like very much, but unfortunately still believes Apple didn't make a mistake. That would be fine, except he often recaps with tantalums. That's a fire hazard.

Saying that... it's strange your cap caught on fire.

That's very common with reversed tantalums on 5V rails so long as the power supply can produce more power than the shorted tantalum can dissipate. I guess what you're pointing out is that the LC III's power supply -5V rail is only rated at 75 mA = 0.375 W, so normally it doesn't overload the component (only drops the voltage). I suspect that some of the LC III power supplies are significantly more capable of their rating (and some tantalum's short more completely).

- David

 

[armadillu]

I am using a Meanwell RPT-60A to power my LCII (which works great); this may explain why it caught fire for me. I got the idea from here. And I used this same power supply (borrowed from my LCII) to test this new LCIII I just recapped... I replaced the burnt cap with a similar one (don't have an extra with the same specs) and now it won't boot any more now. Either way, yes the tantallum caught fire and my room still smells awful because of it! I cut the power as soon as I saw it. With the new capacitor voltages read correctly all around the board, so I think it's funny corrupt PRAM business again with this LCIII, as I got it to work before the recap (after some messing around with the reset headers + PRAM clear booting).

 

[armadillu]

Either that, or it somehow something got toast during the incident... The first tell something was amiss was that the fan started turning faster than usual, and immediately after the fire happened. The fan going faster at least indicates an overvoltage on the 12v line... So maybe that's the end for this LCIII, sigh...

 

[Phipli]

Yes, reverse polarity tantalums often go up in smoke.

Also, yes, the -5V cap on the LC III is fitted backwards and labelled backwards from factory. Easy to confirm with a meter. The +ve should go to 0v because it is higher than -5v. Use the continuity tester function to confirm that the + pad on the footprint is directly connected to the -5v pin on the power connector... I.e. wrong. Ooooops.

The other place you'll see originals fitted backwards is on eMachines Nubus cards! More often than not, they fitted the electrolytic capacitor next to the left side of the Nubus connector backwards. The reason seems to be that it was the only electrolytic in their design and someone decides to fit it with the band facing the same orientation as all their tants... I fitted a tant as a replacement and carefully orientated it the same way... Flash and a dead short. Sigh. Thankfully I didn't kill anything other than the cap and my pride.

It would have never worked otherwise

Electrolytic are more resilient and don't usually fail instantly in this senario. On my eMachines card, the cap had bulged over the years, but no kind of catastrophic failure. "Not working" isn't the end of the world with a capacitor that is just doing power smoothing, there is likely capacitance elsewhere in the circuit, and it just... well... works a little less well. Perhaps a bit more noise on the rail.

 

[Autumn]

Thanks for bumping this decade-old thread, armadillu. If I hadn't stumbled across this, I wouldn't have realized that I just installed that capacitor backward in my machine this week. It's been swapped around now and voltages measuring good.

Here's an extra infographic in case anyone else stumbles across this conversation later.

 

[ymk]

Yup. There is a popular YouTube recapper that I like very much, but unfortunately still believes Apple didn't make a mistake. That would be fine, except he often recaps with tantalums. That's a fire hazard.

The mistake is easy to confirm with a DMM, in seconds.

 

[Phipli]

The mistake is easy to confirm with a DMM, in seconds.

If it is who I think it is, they have a record of just mocking people who raise issues like this. Discovering the truth sometimes needs you to open your ears first.