Does transistor functioning deteriorate (change from spec?) over time and use? I have some 30 year old solid state electronics for which I'm wondering if changing some of the transistors might offer improved or closer to spec performance.
A related question concerns whether current production of said transistors might operate better (closer or more consistently closer to spec?) than the 30 year old dudes. TIA.
A related question concerns whether current production of said transistors might operate better (closer or more consistently closer to spec?) than the 30 year old dudes. TIA.
Old Germanium sure did; they started leaking.
Pure theoretically, Silicium are there for many, many decades without deterioration. At least this is what I learned. Operating them near SOA might perhaps give some parameter changes over the years.
/Hugo
Pure theoretically, Silicium are there for many, many decades without deterioration. At least this is what I learned. Operating them near SOA might perhaps give some parameter changes over the years.
/Hugo
serengetiplains said:
Tom,
Transistors do deteriorate but I think the rate is so slow that 30 years are unnoticable. It may be different if you take your stuff through the Allen belt on a trip to Mars😉 .
Anyway, I really don't have a clue on the rate of deterioration, but if the equipment is designed well it will be insensitive to parameter shifts in the components. For the same reason, 'modern' parts shouldn't make a lot of difference. You should be OK for another 30 years, at least.
Jan Didden
p-type semiconductors run out of holes over time and must be refreshed. There are quite a few online vendors for hole replacement kits.
FWIW, my two transistor amps are 22 and 30 years old, have had heavy use, and spec fine. No deterioration in performance that I can measure or hear (though I've replaced all the electrolytics and a few of the film caps as preventive maintenance).
FWIW, my two transistor amps are 22 and 30 years old, have had heavy use, and spec fine. No deterioration in performance that I can measure or hear (though I've replaced all the electrolytics and a few of the film caps as preventive maintenance).
SY said:
I should add that n-types need periodic demagnetizing. (Roughly ten years) This is best done by taking them out of circuit as the p-types suffer heavily from the nearby magnetic field that is obviously opposite to the general flow of electrons through the given device.
/Hugo
HI all
of course trannys can deteriorate!
classical problem was thermal cycling. After a (large) number of thermal cycles, the silicon chip may crack. Good old TO-3's used to do 100,000 cycles no problems, but plastic used to be worse. I have not seen any thermal cycling data for recent plastic devices. Thinner chips might help.
But as a general rule, the cooler the chip runs the more cycles it will do.
another classical problem was ionic contamination. Once again TO-3 metal cans prevented e.g. sodium getting into the chip. Plastic may be more porous, but the manufacturers should have used a better passivation on their chips too to stop that, so may be recent plastic devices are OK. If sodium gets in the most likely problem is that the leakage currents start increasing.
Another problem is electrical overstress. If any device is runs in a circuit which overstresses it, perhaps an inverter without a snubber, the junctions may be suffering. If any point in the device gets very hot, it may cause a change in characteristics. This is up to the designer to make sure his circuit isn't a destroyer.
Otherwise, the transistor might only show a gain change over a period of maybe a millenium or two if all else is equal. 30 years is nothing!
So if you want your trannys to last a long time, design within spec., and keep them as cool as possible!. Most diy-ers on these pages seem to like big sinks. For good reason. Cray, the supercomputer people, didn't want any junction to run over 55 C. That's not a bad guideline, actually. But difficult to achieve in a hot amp!
Another less well publicised effect but already hinted at is deterioration from space borne-radiation. Don't take your hifi on too many airplane trips or the silicon may deteriorate from cosmic ray events. But at maybe one or two atoms knocked out of place per trip, it will still work for many years.
cheers
John
of course trannys can deteriorate!
classical problem was thermal cycling. After a (large) number of thermal cycles, the silicon chip may crack. Good old TO-3's used to do 100,000 cycles no problems, but plastic used to be worse. I have not seen any thermal cycling data for recent plastic devices. Thinner chips might help.
But as a general rule, the cooler the chip runs the more cycles it will do.
another classical problem was ionic contamination. Once again TO-3 metal cans prevented e.g. sodium getting into the chip. Plastic may be more porous, but the manufacturers should have used a better passivation on their chips too to stop that, so may be recent plastic devices are OK. If sodium gets in the most likely problem is that the leakage currents start increasing.
Another problem is electrical overstress. If any device is runs in a circuit which overstresses it, perhaps an inverter without a snubber, the junctions may be suffering. If any point in the device gets very hot, it may cause a change in characteristics. This is up to the designer to make sure his circuit isn't a destroyer.
Otherwise, the transistor might only show a gain change over a period of maybe a millenium or two if all else is equal. 30 years is nothing!
So if you want your trannys to last a long time, design within spec., and keep them as cool as possible!. Most diy-ers on these pages seem to like big sinks. For good reason. Cray, the supercomputer people, didn't want any junction to run over 55 C. That's not a bad guideline, actually. But difficult to achieve in a hot amp!
Another less well publicised effect but already hinted at is deterioration from space borne-radiation. Don't take your hifi on too many airplane trips or the silicon may deteriorate from cosmic ray events. But at maybe one or two atoms knocked out of place per trip, it will still work for many years.
cheers
John
I always use flyback diodes keep the outputs from reverse biasing, even if it is for tiny amount of time, do to reactive loading. This is not a good situation for a transistor at any time.
Re: Re: Do transistors deteriorate?
The 150W Sanyo Predriver Transistors (two on the input stage) in a TO-220 package have been operating in my GFA-565 for 20 years at temperature at and above 70 degrees Celsius and they are still strong and running fine. I have considered replacing them though...
Meanwhile the Darlington transistors on the output stage (averaging 40 degrees Celsius operation under load for +1 hours) are metal-encased Toshiba units which have slight oxidation on the casings...I too have considered the plausibility of replacing or even upgrading them to a newer (if possible, better) design. What are your guys' thoughts on this? Sometimes it is not smart to modify an amplifier if the circuitry is particularly complicated....
From my understanding though, when replacing transistors, they must be matched with very strict tolerances to have a stably functioning amplifier. Can anyone confirm this with me?
QSerraTico_Tico said:
The 150W Sanyo Predriver Transistors (two on the input stage) in a TO-220 package have been operating in my GFA-565 for 20 years at temperature at and above 70 degrees Celsius and they are still strong and running fine. I have considered replacing them though...
Meanwhile the Darlington transistors on the output stage (averaging 40 degrees Celsius operation under load for +1 hours) are metal-encased Toshiba units which have slight oxidation on the casings...I too have considered the plausibility of replacing or even upgrading them to a newer (if possible, better) design. What are your guys' thoughts on this? Sometimes it is not smart to modify an amplifier if the circuitry is particularly complicated....
From my understanding though, when replacing transistors, they must be matched with very strict tolerances to have a stably functioning amplifier. Can anyone confirm this with me?
john_ellis said:
During fabrication semiconductors can be subjected to temperature exceeding 400C. The chips don't deteriorate due to temperature and would not normally crack.
What is affected is the intermetallics that are formed when bonding the gold wire to the aluminum bondpads on the chip during assembly. As part of the bonding process part of the aluminum bond pad is consumed by the gold ball bond, creating the intermetallics that form the welded connection. this consuming of the pad continues even after the part is packaged and put into service. This intermetallic formation (pad consumption) is accelerated by elevated temperatures and eventually the aluminum is completely consumed and the connection to the die is broken.
This was a common heat related failure in older devices. For more recent devices the mechanism is well understood and the wirebonding processes and equipment are far superior to their counterparts of even 10 years ago.
It's facinating technology that blew me away the first time I was exposed to it; still does.
Regards, Mike.
One thing to look out for is base-emitter junction reverse breakdown. Because the emitter is highly doped, the base emitter junction will exhibit reverse breakdown at a relatively low voltage (about 6V). If the transistor is operated with reverse current throught the base-emitter junction for a period of time, the value of beta will decrease. Hence, one ofter sees diodes between the base and emitter that will conduct in the reverse direction and limit the voltage to a value less than the base-emitter reverse breakdown voltage (see LM34).
The reverse breakdown operation damages the crystal structure of the silicon, which leads to more recombination in the base region, which lowers the value of beta. This damage can be mitigated by anealing. A demonstration of this process can make a good party trick, if you happen to attend parties with your transistor tester, a power supply, and a source of open flame. 🙂
The damage described above can also lead to an increased 1/f noise.
Note that the LM394 matched pair has the protection diodes built in. It would be a problem if one transistor suffered reverse breakdown and the other did not; they would no longer be "super matched"!
Tom
The reverse breakdown operation damages the crystal structure of the silicon, which leads to more recombination in the base region, which lowers the value of beta. This damage can be mitigated by anealing. A demonstration of this process can make a good party trick, if you happen to attend parties with your transistor tester, a power supply, and a source of open flame. 🙂
The damage described above can also lead to an increased 1/f noise.
Note that the LM394 matched pair has the protection diodes built in. It would be a problem if one transistor suffered reverse breakdown and the other did not; they would no longer be "super matched"!
Tom
Hi Mike
Well ...actually chips do deteriorate, but VERY slowly.
You are right in that temperatures over 400 C are used to make them.
But diffusion - the underlying physical process - works at all temperatures. It is just that this is exponentially dependent, so the diffusion process really slows down below 400. But in principle, it does not stop. This means that a transistor made today MIGHT have a gain change if you could measure it in, say, 10,000 years time, if someone hasn't zapped it before then.
I don't think we need worry too much about that!
Regarding chip cracking, it is possible for a thermal stress to crack them. Perhaps more likely is that the solder used to attach the die to the header gives up first, and the chip just dislodges from its header.
Either way, keeping temperatures down helps to prolong life. 😎
cheers
John
Well ...actually chips do deteriorate, but VERY slowly.
You are right in that temperatures over 400 C are used to make them.
But diffusion - the underlying physical process - works at all temperatures. It is just that this is exponentially dependent, so the diffusion process really slows down below 400. But in principle, it does not stop. This means that a transistor made today MIGHT have a gain change if you could measure it in, say, 10,000 years time, if someone hasn't zapped it before then.
I don't think we need worry too much about that!
Regarding chip cracking, it is possible for a thermal stress to crack them. Perhaps more likely is that the solder used to attach the die to the header gives up first, and the chip just dislodges from its header.
Either way, keeping temperatures down helps to prolong life. 😎
cheers
John
From Toshiba transistor data sheet:
Note: Using continuously under heavy loads (e.g. the application of high
temperature/current/voltage and the significant change in
temperature, etc.) may cause this product to decrease in the
reliability significantly even if the operating conditions (i.e.
operating temperature/current/voltage, etc.) are within the absolute maximum ratings.
Please design the appropriate reliability upon reviewing the Toshiba Semiconductor Reliability Handbook
(“Handling Precautions”/Derating Concept and Methods) and individual reliability data (i.e. reliability test report
and estimated failure rate, etc).
Pete B.
Note: Using continuously under heavy loads (e.g. the application of high
temperature/current/voltage and the significant change in
temperature, etc.) may cause this product to decrease in the
reliability significantly even if the operating conditions (i.e.
operating temperature/current/voltage, etc.) are within the absolute maximum ratings.
Please design the appropriate reliability upon reviewing the Toshiba Semiconductor Reliability Handbook
(“Handling Precautions”/Derating Concept and Methods) and individual reliability data (i.e. reliability test report
and estimated failure rate, etc).
Pete B.
From Toshiba transistor data sheet:
Funny that this subject comes up now. I was discussing JFets with a former Telefunken EE a few days ago. He insisted that plastic package JFets (e.g. Toshibas) were known to have issues with long term parameter stability and recommended to use metal package transistors instead. Supposedly, the instability is caused by the plastic material gradually decomposing (at higher temeratures?), thereby changing the characteristics of the silicon chip.
I wonder if anybody has any evidence or information on the subject?
Funny that this subject comes up now. I was discussing JFets with a former Telefunken EE a few days ago. He insisted that plastic package JFets (e.g. Toshibas) were known to have issues with long term parameter stability and recommended to use metal package transistors instead. Supposedly, the instability is caused by the plastic material gradually decomposing (at higher temeratures?), thereby changing the characteristics of the silicon chip.
I wonder if anybody has any evidence or information on the subject?
john_ellis said:
Agreed, used within the manufacturer's guidelines this is not really an issue any more.
Regards, Mike.
...would B-Jfets and mentioning hexfreds and opamps a.o. not benefit from cooling them to (almost) the temperature of einsteins point of relaxation.. deep cryogenic treatment.
Relation between silicon and heat is evident and causing lot of repairs in practice..... going the opposite way might be very efficient.
Wish i could try some transistors and some of opamps first ...
Max
Relation between silicon and heat is evident and causing lot of repairs in practice..... going the opposite way might be very efficient.
Wish i could try some transistors and some of opamps first ...
Max
Actually, tehn relation of silicon and heat is quite well understood, as it was referenced WRT diffusion processes, used in the manufacture of semiconductors.
Eliminating high power related failure modes (local high temperature, material migration and similar), the majority of remaining failures are related to chemistry and mechanics of encasing the actual semiconductor into a standard case. These can also be made worse during the process of assembling a product, by introducing mechanical and therma stress and strain.
Also, another mechanism of failure is deterioration of the case components, which can lead to the semiconductor being exposed to external environment. I routinely deal with 30 year old plastic transistor cases, and for the most part, failures happen on the small parts due to corrosion of pins resulting in cracks in the plastic or opening up od the plastic to pin boundary as a fissure, resulting in mechanical failure or bond contamination/corrosion. I have even had a case of a plastic cased transistor that measured differently depending on it's orientation!
Eliminating high power related failure modes (local high temperature, material migration and similar), the majority of remaining failures are related to chemistry and mechanics of encasing the actual semiconductor into a standard case. These can also be made worse during the process of assembling a product, by introducing mechanical and therma stress and strain.
Also, another mechanism of failure is deterioration of the case components, which can lead to the semiconductor being exposed to external environment. I routinely deal with 30 year old plastic transistor cases, and for the most part, failures happen on the small parts due to corrosion of pins resulting in cracks in the plastic or opening up od the plastic to pin boundary as a fissure, resulting in mechanical failure or bond contamination/corrosion. I have even had a case of a plastic cased transistor that measured differently depending on it's orientation!
HUSH UP!
You are giving people ideas. Next thing you know your favorite high end magazine will be running ads for expensize "Tansistor Resoration Device" or "Transistor Enhancement Spray" or "Make Your Ordinary 100 Watt Transistors Produce Up To 15oW With Our Patented Treatment".
MOSFETs are next.
You are giving people ideas. Next thing you know your favorite high end magazine will be running ads for expensize "Tansistor Resoration Device" or "Transistor Enhancement Spray" or "Make Your Ordinary 100 Watt Transistors Produce Up To 15oW With Our Patented Treatment".
MOSFETs are next.
- Status
- Not open for further replies.