Mistakes can definitely be made, also by big companies; Philips also used to be a big manufacturer of consumer equipment, and they used 300 mW-rated transistors that dissipate 450 mW in the amplifier section of their FR675 receiver:
Philips FR675 repair: replace 2SC2240 with ZTX458
Philips FR675 repair: replace 2SC2240 with ZTX458
My experience of amplifiers with cmos input switches is that even when the supplies are below limits, that they tend to go into latch up when connected to source with the power on. This is a result of RCA plugs mating centre contact first while floating at half mains supply in Class II equipment. I have learnt my lesson a few times, but still sometimes forget to switch off first.
That could be a reason to use a supply with a very tight current limit and not too much decoupling, for example resistor - Zener diode - 10 nF. When the short-circuit current is less than the holding current of the parasitic thyristor and the energy in the capacitor is small enough, it should recover from latch-up without damage.
How would you design a test to screen out the parts that can survive the higher supply voltage? Run all parts through at the required supply voltage (plus some margin) and sell the ones that make it hoping that the test won't have affected the reliability? That seems like a rather sketchy proposition.
Parameters such as noise, operating frequency, supply current, etc. that can be measured in a non-destructive way can be screened for. That say, screening is typically rather expensive. If there is sufficient market for a higher performance part, the manufacturers will much rather design a higher performance part - and can be willing to do so for a key customer willing to buy millions of them.
Absolute maximum voltage ratings tend to be dictated by the semiconductor process. The products are guaranteed to last for at least 10 years of continuous operation at the worst case operating conditions. As we can't delay product launch by ten years so see if the parts fail within that timeframe, we put the parts through accelerated lifetime tests. These test take place in autoclaves where the parts are operated under worst case conditions at 100% humidity and elevated temperature (150 ºC junction temperature). This test typically lasts 1-4 weeks. While the test does not guarantee that the parts will last 10 years in the customer's hands, it does at least give enough confidence that such a lifetime can be expected.
Tom
I've come across many instances of "designed-in" obsolesence in the shop. Un-heatsinked regulators, under-wattage resistors, you name it, I've seen it.
And this practice goes back decades....into the "tube era' as well. Milking a extra few watts from an amp to make advertising more impressive. And all with the hopes that once the thing breaks down, the owner might buy a "new one" to keep the cash flow going for the manufacturers and sellers.
Tom
It was standard practice for Crown amplifiers to use Motorola/On power transistors screened to run at higher voltage. These parts did have house numbers but started life as standard parts. I thought they screened them by testing at higher voltage with a current limited supply.
As Crown effectively no longer exists for repair use the stock parts are often just swapped in and run for a burn in to see if they will work. Not reasonable for manufacturing, but the seems to be best way to repair things when manufacturer parts are no longer available.
As the breakdown mechanism probably does not change with age, higher voltage current limited tested may just be adequate.
It was standard practice for Crown amplifiers to use Motorola/On power transistors screened to run at higher voltage. These parts did have house numbers but started life as standard parts. I thought they screened them by testing at higher voltage with a current limited supply.
As Crown effectively no longer exists for repair use the stock parts are often just swapped in and run for a burn in to see if they will work. Not reasonable for manufacturing, but the seems to be best way to repair things when manufacturer parts are no longer available.
As the breakdown mechanism probably does not change with age, higher voltage current limited tested may just be adequate.
oh most definitely, why do you see so many items of everything die a year after the warranty expires. They design things to eventually fail because they want you buying more.
Not all parts die at a certain service life. Even electrolytics have a service life that is very dependant on ambient equipment temp, even if the designer knew ripple situation and local heat transfer. So it is all statistics and how much of a population gets stressed (eg. from the likes of ambient temp, and mains voltage surges, and stupid things users may do).
Let’s not wordsmith... not designed to last = designed to fail.
I have an engineer friend who left the aircraft part design profession for this very reason.
Once a part was designed and tested, it was whittled away and retested, rinse and repeat over and over until it barely surpassed the required minimum failure metrics and mandatory replacement windows set forth by regulation. So, not the strongest or most reliable part, but the part that cost the least to meet the bare min standard.
I am sure this is a practice found in many industries, especially those that are not as tightly governed as aircraft parts.
It's been going on for ages. Earl "Madman" Muntz was one of them.... Clipping out parts from a TV set until it barely ran..... then sold them cheap.
Madman Muntz - Wikipedia
I've used garden variety silicon diodes between the V+ and/or V+ and V- (in the case of bipolar power rails) and the power pins many times, to reduce the voltage applied to chips when a small reduction is needed. Never had a problem. Of course, the diodes should be checked for leakage and for the value of their forward voltage drop to keep the V+ and V- at the power pins balanced with respect to each other.
Last edited:
If personal experience serves, (sorry, I bet it will be demeaned as "anecdote"
), I once needed "+/-22V Op Amps".I setup a test board with a DIP socket, grabbed a couple Op Amp "sticks" and started testing them, by driving them to clipping into an "easy" load, I think it was 10k if memory serves.
Most could handle the negative peak; but typically top peak clipped early, not flat topping but showing a "hook" shape.
Maybe I was lucky but about 25% worked well; the rest were NOT destroyed by any means and were used elsewhere from regular +/-15V rails.
Same with 2N3055: I used some 16000 to 20000 of them from 1969 to about 2004/2006 when faking became unbearable.
All of them tested in a self made rig, *most* stood from 75/80V up to mind boggling 105/110V .
Standard 100W Guitar/Bass amp used +/-42V rails ; transistors meeting "just" datasheet specs were used in 60W amps with +/-35V rails, all performed reliably.
Oldest amp I keep contact with is from 1972 ... still working ; regularly get 25 y.o. or more for routine maintenance, basically "mechanical" problems: dirty/worn pots, jacks, switches.
So at least this "anecdote" is backed by solid "experience" ; some 14000 amplifiers in 50 years.
Jim,
The original RCA 2N3055 was made on a hometaxial power transistor process. As this process was not economical by the 70's the product became the more modern epitaxial transistor. Not surprisingly the performance changed.
As the original specifications were by more modern standards quite low and the model number extremely popular, it became common to call overrun parts or even ones that did not meet higher standards a 2N3055.
I suspect you were using them well above the original specification of a beta of 20-70 and translated into a modern Ft 40,000 hertz!
As processes got better yields there were fewer reject parts marked as 2N3055 and what was being sold had lower spec.s closer to the originals!
The original RCA 2N3055 was made on a hometaxial power transistor process. As this process was not economical by the 70's the product became the more modern epitaxial transistor. Not surprisingly the performance changed.
As the original specifications were by more modern standards quite low and the model number extremely popular, it became common to call overrun parts or even ones that did not meet higher standards a 2N3055.
I suspect you were using them well above the original specification of a beta of 20-70 and translated into a modern Ft 40,000 hertz!
As processes got better yields there were fewer reject parts marked as 2N3055 and what was being sold had lower spec.s closer to the originals!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.