Dr Jefyll wrote:
GARTHWILSON wrote:
His response was that the temperature cycling on the bond wires where they contact the die wasn't a problem.
What about where the bond wires connect to the lead frame, then? Seems to me I've heard thermal cycling cited in regard to
some sort of internal problem for semiconductors, but I admit the recollection is vague.
If there's any problem with the bond wires at all, I'm sure it's on the end that goes to the die. At the VHF/UHF power transistor manufacturer, we used two kinds, gold and aluminum, and I talked every day to the women who worked the machines, as well as other engineers who had more contact with what was going on in the field, and I was not aware of any bond-wire problems we had outside of infant mortality. Part of the spot-checking on production lots was to actually pull bond wires off and see how much force it took to do it. There was a machine in one of the labs with a teensy hook that did this. In the past 20+ years of making high-end aircraft intercoms, we've only had one or two bondwire problems, and they were infant mortalities.
I'm on the Bicycling forum too, and it's amazing what myths get started and take hold. The early years of carbon-fiber frames had a lot of problems, but they were solved two decades ago, and short of a manufacturing defect, they are very tough and essentially don't fatigue like the metal ones do. In fact, I've heard from lots of riders in the 250-350-pound range on carbon fiber, all without any frame problems. But with all the rumors people believe about what will damage the carbon, I used to joke that we could start another one like that cell-phone radiation would damage it, so if you make a call, you should get away from the bike. Maybe there's something like that left over from the early years of ICs. I still have occasional contact with someone who used to manage a production line at Synertek. I'll ask him if he knows of any problems they might have ever had with bond wires.
Quote:
Quote:
I guess now they can do 4GB without having to store multiple bits in one cell.
Hmmm... and 4GB is no longer leading edge. Although you weren't explicit, I think you're implying that users of SD cards that
are currently leading-edge are guinea pigs for whatever the newest process is. And that's plausible, given the dollars and the competitive forces at work.
Right. I don't know what the "leading edge" is today. Maybe 64GB? To some people, it's super important to be the first one in the office with some new thing. I'm not one of them. I'll give time for the bugs to get worked out, and to see if the new thing is even going to stick around. I'm glad I didn't get caught up in PCMCIA and zip discs and other things that lasted such short times.
Quote:
Quote:
the only things that have gone out with time are 220uF 16V capacitors that had 6V on them all the time they were powered up but their charge voltage was steady, ie, didn't go up and down which I think would have made them last longer.
Can you elaborate on that? It's not the first time I've heard of capacitors being specified with a seemingly extreme safety margin for voltage. And yet, extreme as it seems, a 166% margin isn't adequate?!
I have capacitor data books here, but they don't usually tell everything I would like to know. Capacitors' ratings are for so many volts and hours at certain temperatures though, and ripple current will affect it in many cases. Storage temperatures can affect the life too, and some of our customers' planes are parked on the tarmac in Phoenix cooking all day every day, too hot to touch most of the time.
In some linear applications especially, multilayer ceramic chip capacitors need to be derated even a lot more, not for life expectancy, but because their capacitance is reduced dramatically as you reach significant fractions of their WVDC, which can produce severe distortion. On those, I try to have the WVDC at least 5x the voltage that will normally be on the capacitor.
The socket problem is the only issue I've had with my workbench computer, and only twice in the 20 years since its original incarnation. All it took to fix them was to pry the ICs out just a tad (not the whole way out) and push them back in, for the self-cleaning effect of wiping the contacts. There probably would not have been any problem if I had used better sockets, instead of the cheapest tin-plated ones. I'm sure screw-machine ones with gold-plated contacts would have avoided the problems if I wanted to pay for them. This is the reason that military applications, especially combat ones, do not allow sockets. They need the higher reliability. In fact, they also don't want lead-free solder, for the same reason.