Before the transistors give up or pucker out is a long time depending on thermal stress. A power supply cap's value and performance will go over time. Most of the are rated in hours , like 20000 Hrs. Operating a PS cap close to it maximum voltage and current limits its life. How many electrons and pass through are atom lose an electron to cause failure is unknown, I believe.
Germanium (and some early silicon from the same era) are a special case. The tin used in their construction can grow dendrites from even the smallest potentials - even static buildup sitting on a shelf. That can cause internal shorts. It’s possible to clear the short by fusing it but that can cause both mechanical and electrical damage so don’t count on it working each and every time. Don’t try it in circuit.
If the package isn’t hermetically sealed (ie, TO-1) all bets are off too. They will either work or they won’t. Non-hermetic packages of today are far better than the ones made in the 60’s, and the metallurgy of todays transistors (even 1970’s transistors) is less sensitive. When glass passivated diodes came out it was a game changer for reliability.
Not at all concerned about transistors seeing brief excursions to 250C. When something royally effs up and outputs get blown it may very well be closer to 400C. That kills in minutes or seconds.
Whether thermal cycling causes slow of fast wear out depends on construction. Aluminum TO-3’s 5000 cycles. Steel 100,000+. Lead Free DPAKs aren’t good for more than 5000 cycles either - and class D amp makers love them.
Plastic packages vary in how moisture sensitive they are. New TO-92’s are pretty damn good. Old ones could be very bad. SMD packages tend toward suck. DPAKs aren’t as good as TO-220. SOIC is horrible compared to PDIP.
TO-3‘s are supposed to be hermetic. Fakes aren’t.
Electro-migration is an issue. It is basically very slow fusing caused by accumulated DC current. It literally thins the conductor out in the middle over time and current density. Although you generally need to be above current ratings before it gets fast enough to “blow”. There are generally design rules concerning it. Thermal fusing itself may happen faster, if there is little or no heat path for the I squared R self heating.
Since thermal cycling was mentioned, I thought I'd also mention another related failure mode called 'bond-wire liftoff', which is common in power transistors. The wire is aluminium, the die is silicon (or SiC) and the package is ceramic / plastic. The coefficients of thermal expansion of these materials are somewhat close but still far enough to cause failure if sufficiently cycled (thermally).
The following paper deals with IGBT modules, but since IGBTs are MOS-gated BJTs, the same should apply to BJTs as well.
https://ieeexplore.ieee.org/document/6254052
The abstract in the above link says it all, but anyone interested in further details may read the paper here (download PDF).
https://www.researchgate.net/public...odules_due_to_thermomechanical_induced_stress
The following paper deals with IGBT modules, but since IGBTs are MOS-gated BJTs, the same should apply to BJTs as well.
https://ieeexplore.ieee.org/document/6254052
The abstract in the above link says it all, but anyone interested in further details may read the paper here (download PDF).
https://www.researchgate.net/public...odules_due_to_thermomechanical_induced_stress
I don't work in a fab, and simply use ready-made transistors / modules like anyone else.
This kind of failure usually occurs when the operating frequency is low and its time-period is long when compared to the thermal time constant of the die. Fortunately, this failure mode is 'open circuit' and doesn't do much damage (as opposed to short circuit failures).
This kind of failure usually occurs when the operating frequency is low and its time-period is long when compared to the thermal time constant of the die. Fortunately, this failure mode is 'open circuit' and doesn't do much damage (as opposed to short circuit failures).
There are two ways of looking at old components which have not failed:
1. They've worked this long, so they'll probably go on working for ever.
2. They've worked this long, it must be time for them to fail.
If we could predict this, we'd win the lottery every week too.
My advice would be to go for it!
1. They've worked this long, so they'll probably go on working for ever.
2. They've worked this long, it must be time for them to fail.
If we could predict this, we'd win the lottery every week too.
My advice would be to go for it!
But it is exactly the failure mechanism in the power supply IGBTS and the IRFS4227’s in the Behringer iNukes. There is nowhere near enough bulk heat sinking and the semiconductors are sized to barely handle the average load, so every single BOOM BOOM BOOM at war volume runs them through one of those thermal cycles. Maybe not the full 25 to 200C so you will get more than 5000, but the number isn’t a million.
And the die attach failure from thermal cycling happens at the same time. The joint doesn’t just outright fail. But the thermal resistance goes up from internal micro-cracking due to CTE mismatch, and eventually power dissipation that once resulted in safe operation are no longer safe. And the thermal cycling temperature excursion gets progressively larger. Even the solder joint to the PCB exhibits this to some degree. I hate SMDs, I hate SMDs, I hate SMDs and this is why. Even if I could see worth a damn without glasses these days.
Every time I get an unfamiliar unit in for repair I open it up on the spot, if I see SMD I close it right back up, not going to deal with that crap. Many times there is no Service Manual for these items as they were never intended to be repaired. I stopped working on Behringer equpment years ago.
Craig
Craig
You mean “repair” = replace entire PCB. With a list price approximately the same as the street price of a new unit.
If you’re ambitious, you could test the transistors using one of the configurations Nelson Pass uses to match and test MOSFETs. And if you have the specs for those part, you can measure them against the data sheets to determine the quality of the part. Of course, the parts will not be exactly as the spec, but in arrange of the specs.
Depends on how they were stored. Some failure mechanisms have been mentioned already. I'll add ESD damage as a potential issue. I cringe when I see parts offered for sale in plastic tubs.
Tom
That s a good price even if there were no 2kva transformer, the 2SJ50/2SK135 are the best laterals you can find,
and the big advantage is that there s a lot of them here as a lateral fet amp require several parraleled devices
to get good perfs, i can only advise to jump on the occasion if the amp is functional.
and the big advantage is that there s a lot of them here as a lateral fet amp require several parraleled devices
to get good perfs, i can only advise to jump on the occasion if the amp is functional.
My priority is transistors if i would be able to get the amp. 2kva transformer looks good at first but i doubt i would use it. Mentioned amp is rated 400W at 8 ohms, so its working 90 V rails or so. I don't think i would ever build an amp working at that voltages. Although there is an option to re vinding the secondaries or maybe i would just snip some part of the secondary vinding 🙂
Btw, some decades ago I was given these goodies:
Sixteen pairs of 2SJ50/2SK135's, plus two extra 2SJ50's. According to the marks at the mounting holes, all of them are pulls,. Anyway, all of them are fully functional 😊 👍 !
Best regards!
Sixteen pairs of 2SJ50/2SK135's, plus two extra 2SJ50's. According to the marks at the mounting holes, all of them are pulls,. Anyway, all of them are fully functional 😊 👍 !
Best regards!