What would be a reasonable MTBF for a BJT or Mosfet from which I can extrapolate a lifetime prediction? This is the biggest piece of missing information.
I put those numbers into my retro-ish Casio color graphing calculator and the best regression got me this equation:
MTBF = 9.44e+10 * C^-5.09
Where C is temperature in Celsius.
The equation I was trying to use to extrapolate looks like this:
exp(((273.5+Tjnom)**-1-(273.5+V(Tj1))**-1)*-7k)
Where in SPICE language ** means power. X**-1 is just more convenient than 1/X (they are both the same).
I really have little clue what I'm doing. I'm just trying to find something that gives a reasonable estimate of instantaneous stress, lifetime and age, so I can study circuit stress in the simulator.
I put those numbers into my retro-ish Casio color graphing calculator and the best regression got me this equation:
MTBF = 9.44e+10 * C^-5.09
Where C is temperature in Celsius.
The equation I was trying to use to extrapolate looks like this:
exp(((273.5+Tjnom)**-1-(273.5+V(Tj1))**-1)*-7k)
Where in SPICE language ** means power. X**-1 is just more convenient than 1/X (they are both the same).
I really have little clue what I'm doing. I'm just trying to find something that gives a reasonable estimate of instantaneous stress, lifetime and age, so I can study circuit stress in the simulator.
I was told for reliability in commercial applications to keep semiconductor temperatures below 40C. But I have also heard that they will last indefinitely under 125C.
It is probably practical for neighboring components that transistors don't get too hot. But assuming adequate heatsinking, is there any problem with running a TO92 at say 3/4 max dissipation continuously, if temperature was stable?
It is probably practical for neighboring components that transistors don't get too hot. But assuming adequate heatsinking, is there any problem with running a TO92 at say 3/4 max dissipation continuously, if temperature was stable?
In my view, the first to go will be the electrolytic caps, the aluminium type. The metallized unts seem to keep it all together much longer.
When I pull out an electrolytic from a 30+ year old device, they typically measure at about one half their nominal value in capacitance. Most are by Elna, but that's hardly any indicator, given that most caps built into Japanese devices originated from Elna.
Thus far, I have never experienced a semicndutor failure which could not be traced back to misuse or some "operator error" while soldering, etc. What I did find quite a few times were semiconductors output devices which tended to be hotterthan I'd like them to be, let alone what they have to be in view of their use regime. This is very often a function of the thermal compound going Aarizona desert dry; cleaning up that mess is quite a task, but once done, and once new thermal compound is properly applied, the temperature difference may go as high as 15 deg Centigrade lower than previously under the same work regime.
Internal temperature is also a concern. A number of highly regarded Japanese integrated and power amps from the late 70ies and early 80ies (a good example - Sansui's 5xx, 7xx and 9xx series, but model number designations varied wildly around the world) were very tightly packed. This caused them to be service nightmares, but also to develop disturbing (for me at least) heat issues, which often resulted in bloated electrolytics, sometimes in custom format, so they were very hard to replace properly.
When I pull out an electrolytic from a 30+ year old device, they typically measure at about one half their nominal value in capacitance. Most are by Elna, but that's hardly any indicator, given that most caps built into Japanese devices originated from Elna.
Thus far, I have never experienced a semicndutor failure which could not be traced back to misuse or some "operator error" while soldering, etc. What I did find quite a few times were semiconductors output devices which tended to be hotterthan I'd like them to be, let alone what they have to be in view of their use regime. This is very often a function of the thermal compound going Aarizona desert dry; cleaning up that mess is quite a task, but once done, and once new thermal compound is properly applied, the temperature difference may go as high as 15 deg Centigrade lower than previously under the same work regime.
Internal temperature is also a concern. A number of highly regarded Japanese integrated and power amps from the late 70ies and early 80ies (a good example - Sansui's 5xx, 7xx and 9xx series, but model number designations varied wildly around the world) were very tightly packed. This caused them to be service nightmares, but also to develop disturbing (for me at least) heat issues, which often resulted in bloated electrolytics, sometimes in custom format, so they were very hard to replace properly.
As far as I know medical, avionics and space do not have to conform to ROHS. ESTEC certainly doesn't. But they do remove gold flash before soldering for the obvious reasons, which have nothing to do with lead-free.
There is a 700+ page report on the matter, I think I still have a copy if you want, not the most exciting read...
I can see if that happened ETC would stay open, but what about brake pedal override, did they explain why the brake override function never worked. Easy to verify with ETC, just try the old brake and gas technique, if it does not return to idle in 3-5 second(with ET wide open) then i would walk away from that one ..
The ruling on the sudden acceleration, was driver error, panicking due to the other problems like sudden breaking.
"The chances that the same dumbness occurred in two totally separate environments, to give identical results, is of course dramatically lower ... "
The next step is to ask each team to try to 'break' the others code.
I've seen FMEA's on discrete semi devices - just the silicon, and not the packaging - go into many A4 pages. So, for SW, I can imagine the effort required for a really good secure job . . .
The next step is to ask each team to try to 'break' the others code.
I've seen FMEA's on discrete semi devices - just the silicon, and not the packaging - go into many A4 pages. So, for SW, I can imagine the effort required for a really good secure job . . .
Your First Language Is English ?......
Breaking is what happens when the car crashes.
My father has been a passenger witness in Denmark to two UA events in a Camry on the same afternoon.
Try convincing him that the UA events were due to driver error.
Dan.
I think you mean braking.
Breaking is what happens when the car crashes.
My father has been a passenger witness in Denmark to two UA events in a Camry on the same afternoon.
Try convincing him that the UA events were due to driver error.
Dan.
Last edited:
No they use lead free, but also a lot use ENIG as a surface finish (not gold flash) as this does not cause gold embrittlement of the joint due to the small amount of gold used (4% of gold in a joint leads to problems with AuSn4 intermetallic layer being formed, I believe). What is more of a concern is pure tin component finishes, these not being fully wetted during the soldering process, leading to possible tin whiskers. Lead free BGAs have to be re-balled, and the preference would be to have discrete's pre-wetted with a lead solder but with so many discrete's on boards the cost makes this impossible except for designs where failure is an absolute no no.
No panicking is required. At sudden breaking, due to inertia, the driver's body including his right foot will suddenly move forward...
A clear case of positive feedback leading to oscillations.
Jan
That was the judges ruling in the cases brought against Toyota. As well as NASA, Keith Armstrong was brought in to do a report on EMC. The amount of paper work generated was impressive. I was involved in a study of component finishes and the risks of going lead free for a USA/UK defence project at the time so ended up trying to read as much of it as I could mentally stand, we decided that lead free solder was to risky and ditched the idea (some European customers insisted of RoSH though).
The increase in solder temperatures due to lead free has also made people aware of another obscure failure mechanism CAF (conductive anodic filament), though I haven't heard of any actual field failures due to this phenomena. Its surprising how exciting PCB design and assembly can get!
Last edited:
Judges Technical Ignorance......Or Graft ?....
Those with tin whiskers knowledge/experience/understanding would I think agree that the judges finding is flawed/plain wrong.
Sure, panic may have exacerbated the outcome, but I do not believe that to be the root cause.
Lead free solder, tin whiskers and dodgy software/firmware/hardware....looks like we are in for more of the same into the future.
I will stick with my old 'Drive By bowden cable Wire ' throttle control for a good while to come......that and the key operated physical start/ign/acc/kill switch.
Dan.
Those with tin whiskers knowledge/experience/understanding would I think agree that the judges finding is flawed/plain wrong.
Sure, panic may have exacerbated the outcome, but I do not believe that to be the root cause.
Lead free solder, tin whiskers and dodgy software/firmware/hardware....looks like we are in for more of the same into the future.
I will stick with my old 'Drive By bowden cable Wire ' throttle control for a good while to come......that and the key operated physical start/ign/acc/kill switch.
Dan.
They don't, if you look at either the NASA tin whisker site or the NPL site over here, they are pretty pictures of them growing through the coating.
http://www.npl.co.uk/upload/pdf/mitigation-of-tin-whiskers-using-conformal-coatings.pdf
Picture on page 3.
3%+ lead is the only way.
http://www.npl.co.uk/upload/pdf/mitigation-of-tin-whiskers-using-conformal-coatings.pdf
Picture on page 3.
3%+ lead is the only way.
Last edited:
You state the obvious, everybody already knows this stuff..
Exemptions were typically for safety, flight, weapons, industrial control. Now, the exemption list is a horrid mess. Internet servers have their own exclusion. Go figure. It appears that lobbying groups rule the exemption committee. And yet, lead batteries are not covered.
I couldn't find that.
We designed and had over 10 thousand pieces of hi tek equipment built by outside contractors, required lead free, had to use ENIG. Don't know what they did about BGA's.
I would assume that the judge can only rule on the merits of the arguments presented. It's always very difficult introducing engineering/science into the courtroom, "experts" can always be found on both sides of the coin.
If it were easy to push science in the courts, things like near superconductivity and quantum teleportation could be legally embraced. After all, prove it doesn't exist...
jn
- Status
- Not open for further replies.