[Fascia]

Hey Guys,

I've had this O2 cam for a long time, but only now thought I'd give it a go. Shortly after plugging it in, the O2's speakers start making a sputtering static noise and I detected the distinct scent of magic smoke escaping the camera, so I unplugged it as fast as I could.

Upon examination, it appears one of the SMD parts has died, and potentially damaged some nearby components as well. Attached is a close-up of the affected area. Does anyone know what the damaged component is? Unfortunately, the melted face of it has removed most markings to identify it. It looks like it might be one of the "4.7-16 F4" components seen elsewhere on it. But I'm afraid I don't know enough about these components to say what that means.

[robespierre]

It's a tantalum solid chip capacitor. They ignite if overstressed. Sometimes the flames can scorch nearby components, but in each case I have seen, the only damage was a coating of soot that needed to be cleaned.

The main suppliers of these devices are Kemet, Vishay, and AVX.

The markings indicate 4.7 uF, 16 V max, and F4 is the case size.
I can check later to see if the one that blew is the same value.

[weblacky]

Those appear to be capacitors - 4.7uF 16V, I would GUESS they are tantalums. You can try to find out by using a multimeter and testing which "side" the white stripe is on for the other "like" 4.7-16 parts above it. If the stripe is on the negative side (ground) than it's not a tantalum, if it's on the positive side then it is a tantalum. From what I've read you can feel free to replace older Magnesium Tants with newer Polymer Tants in all situations.

Guess rebuilding these modules and cards is going to be slightly more involved than I thought, scanning caps is fine for me, but there are a lot of them, even in this camera (not to mention how many are in the IO module).

I'm kind of surprised this happen, but I'm curious if this was a cap simply failing short...or if this was a power source inside the the IO module (or PSU) going high?

[Fascia]

I'm curious if this was a cap simply failing short...or if this was a power source inside the the IO module (or PSU) going high?

Hmm, a good question. The O2 hardware reference says for the camera there's +5v on pins 1 and 35, -12v on pin 30, and +12v on pin 64. The connector on the AV module is pretty thin for multimeter probes, hopefully the connector in the camera itself will be more accessible.

F4 is the case size.

Is this an old standard? I can find 4.7uf 16v tantalum capacitors on the manufacturer websites, but they all give case sizes as four digit numbers.

[weblacky]

I'm not sure about the size, I'd use a pair of calipers and lookup SMD standard sizes for tant caps, hopefully you'll recognize the footprint.

Also please test the the burnt cap for being open or closed short. The smoke leads me to think it shorted closed. After removal of the cap, check if one side was on the ground side, check for short using Ohm mode and diode mode on multimeter on two pads of the (now removed) bad cap.

Hopefully this was a "sputter" failure (repair surges and closures (trying to heal) but didn't. You could also (if you're feeling adventurous), remove the dead/shorted cap and hook it back up to the O2 module and try to measure the two pads with a multimeter set to DC voltage. The voltage across the pads is likely going to be one of those numbers, given what you said, I'd assume 5V DC, since derating the cap from 16v to 5v makes much more sense.

[robespierre]

OK, from my Moosehead camera, I can see that C329 is identical to C271, a 4.7 uF 16 V tantalum with length 0.130" x width 0.115".
It looks like this is the "1311 Imperial" case size.
You can do a parametric search like this to find replacements.

Tantalums are sensitive to reverse voltage (below -1 V) and spikes or ripple AC+DC exceeding their rating. One thing that happens in older computers is that the power supply filtering goes south as wet electrolytic caps dry up, letting through more ripple. If the tantalums were marginal to begin with, this is their call to burst forth. This is one reason it might be wise to use a higher rated part: if the DC in the circuit is 12 V with a tolerance of +/- 10%, and designers spec'd a 16 V tantalum with +/- 10% tolerance, that's only 1.2 V of headroom in the worst case. If a 25 V rated cap is used, that becomes 9.3 V of headroom: much more comfortable. There are polymer substitutes, but not many in this size, and polymers can still fail—they just don't explode. They are also 3x the price.

The other thing that makes them fail is excessive heat during soldering, so make sure to get in and out fast and don't dwell on it.

edit:

F4 is the case size.

Is this an old standard? I can find 4.7uf 16v tantalum capacitors on the manufacturer websites, but they all give case sizes as four digit numbers.

Yeah, I was wrong about that. Case size is not printed as it can be measured. It may be "B case" or "1210 case".

[weblacky]

I'm sort of thinking PSU too, in that we've seen a lot of that lately and it's something I'm trying to work towards to tackle. I'm unsure of the whole thing, but the O2 PSU does produce those voltages directly. So I'd assume there would be no need to any kind of DC/DC converter in the audio/video IO card to power the camera. Just straight through power from the PSU would be believable.

O2 PSUs are second on my list to tackle...but that will still be awhile.

[robespierre]

The Moosehead camera was made by Logitech—their name for it is "Z EYE":


The Logitech logo is on the back:


Note that there are 10 tantalums on the back of the board.

As to preventative maintenance, I highly do not recommend it. This device is not really serviceable at all: there are pieces of Scotch tape strategically stuck on to prevent shorts, many loose plastic pieces and wires without any slack at all, and the whole thing fits together tighter than a preacher's ass. Interestingly, the cable is double shielded and double sheathed.

[weblacky]

While I won't be doing this for the foreseeable future, I'll have to tell you the hard truth.

The advice we've all been giving ourselves of "I wouldn't recommend it". Well it's time to put take off our training pants and put on big-boy pants now. The days of "escaping" to another used/NOS unit is nearly gone. Can you still get stock yes, prices...getting higher...stock...getting older.

I'm going to start breaking these rules slowly and systematically in the near future for the good of all of us. It's time to fight back on component aging. While many SGI systems were made with high-quality parts and some rugged power protections in mind ,that doesn't prevent age from eventually catching up.

Right now, power is the first-level threat. Bad or sloppy power regulation/generation is a #1 killer, thats needs to be stopped like 5 years ago (It's top priority right now). However cap ESR is the next (nearly equal) priority in nearly ALL SGI models. It's a fight which to start with. Likely I'm going to start concentrating on specific systems, tackle PSUs, then move to mainboards, then specialized cards (maybe) like the ones in the O2, to perform 101-level rebuilds on an entire system at once, before moving on.

Some SGI don't utilize that many electrolytics (or nearly at all), some use A LOT. If nothing more we need to get base testpoint readings for diagnostics and then move on to preventative changes/upgrades. Polymer tants are very low-hanging fruit (with current SGI tant caps, now). In terms of affective upgrades to other chemistries...I don't have the knowledge to do that safely...so upgrading existing (but same style) components is on my list.

The problem with Tants is while you can check ESR with several methods, most of the sneaky failure checks can only be done live with an oscilloscope to check for spurts.

Not to mention MLCC cap aging/de-aging ideas that need to be done every couple decades or so.

Yeah, unless we want to wait until components like the O2cam are $300+ a pop on eBay, we'll need to start getting daring and doing both preventative changes and investigative recordings now while working examples are around freely. Of course, I'd start with O2 PSU first, then mainboard/backplanes, then IO cards (lots of stuff there), and finally maybe Cams. I mean with the voltages they are talking about and not current limited in any way, IndyCam and O2Cam might be time bombs? We don't really know and don't have those answers.

[Fascia]

Yeah, sadly, that's a big part of the reason I'm nowhere near as active as I once was. I'm damned near afraid to turn these systems on, wanting to do so sparingly so as to eke as much longevity as I can from them.

I had a bad SCSI cable in my Fuel, which is notoriously hard to find. I bought one from Ian Mapleson and he told me it was the last one he had that wasn't in already in a working system. Bit by bit these out of production parts will cease to be available at all :(