TDA2002 and other amps, do these get 'worn'?

Thread Starter

Nicholas

Joined Mar 24, 2005
139
Hey guys

Just todays stupid question. I'm fixing a board that has bad sockets
and other stuff, and I look at the TDA2002 amp that has been on the
board since 1980. So I wonder: is this a work-or-not unit, or do they
get worse over time? Should I pop in a new one or one of the TDA2003's
I have lying around?

I want to fix the board for the future too, sort of preemptive strike :)

Thanks!!

Nicholas
 

MrChips

Joined Oct 2, 2009
30,708
In general, semiconductor electronics do not wear out.
On a rare occasion they can be damaged and might need replacing.
 

takao21203

Joined Apr 28, 2012
3,702
Semiconductors can degrade and latch up partically.

After a while it could be that signal/noise ratio or output levels do no longer meet specification, without the whole chip broken as such.

I have experienced it a few times.
 

bountyhunter

Joined Sep 7, 2009
2,512
Hey guys

Just todays stupid question. I'm fixing a board that has bad sockets
and other stuff, and I look at the TDA2002 amp that has been on the
board since 1980. So I wonder: is this a work-or-not unit, or do they
get worse over time? Should I pop in a new one or one of the TDA2003's
I have lying around?

I want to fix the board for the future too, sort of preemptive strike :)

Thanks!!

Nicholas
Any well made semiconductor device should not "wear"..... but an interesting side note: in 1979 - 1981 there was a lot of bad silicon starting material in the industry that was contaminated with various things and the ICs we made in those years were notorious for working fine (at first) then blowing up after a random amount of time. The Dolby decoder IC in the high end cassete recorder I bought in 1980 blew up just out of warranty.... BTW, the IC was made by national Semi.

So the answer is, the ICs should be good.... but I worked at fairchild semi from 1979 - 1981 an we made tens of thousands of TDA2002 and TDA2003 devices and a lot of them were bad.
 

#12

Joined Nov 30, 2010
18,224
Ignoring obvious failures,
In the most minuscule analysis, heat makes molecules move. In fact, heat is the second ingredient in diffusing dopants into the silicon. Tens of thousands of cycles of heating and cooling can degrade the gain or signal to noise ratio of some transistors.

Opinion: If you have no evidence of the IC gradually moving, "out of spec", ignore it.
 

Thread Starter

Nicholas

Joined Mar 24, 2005
139
Thanks for your answers!

I'm just thinking, that since I have the board out anyway, and I have all the
spares on hand, caps and TDA2003's etc why not do it all in one go?

Does the 'if it aint broken..' cliché really apply? The board will get fresh sockets,
fresh Rubycon/panasonic caps and new 10W TDA2003's

Thanks again!
 

#12

Joined Nov 30, 2010
18,224
Replacing components causes changes in the circuit board traces, bending wires, and sometimes, even, mistakes.:eek:

Mess with it enough times, and the solder pads come off the fiberglass.

My stereo amplifier has been working since 1977 with no repairs except spraying cleaner into the volume control. I'm not going to replace the power transistors just because I have a drawer full and the amp needs a new, "on/off" LED. They might last another 37 years if I don't break anything off or ruin a solder pad.
 

bountyhunter

Joined Sep 7, 2009
2,512
Thanks for your answers!

I'm just thinking, that since I have the board out anyway, and I have all the
spares on hand, caps and TDA2003's etc why not do it all in one go?

Does the 'if it aint broken..' cliché really apply? The board will get fresh sockets,
fresh Rubycon/panasonic caps and new 10W TDA2003's

Thanks again!
In electronics my rule is: if electrolytics are more than a few years old, they go straight to the trash can. Not sure about the IC's.
 

Thread Starter

Nicholas

Joined Mar 24, 2005
139
Replacing components causes changes in the circuit board traces, bending wires, and sometimes, even, mistakes.:eek:

Mess with it enough times, and the solder pads come off the fiberglass.

My stereo amplifier has been working since 1977 with no repairs except spraying cleaner into the volume control. I'm not going to replace the power transistors just because I have a drawer full and the amp needs a new, "on/off" LED. They might last another 37 years if I don't break anything off or ruin a solder pad.
That's a lot of years for the caps, yes? Isn't that well beyond their
life cycle?
 

takao21203

Joined Apr 28, 2012
3,702
Best would be to crank up the amp and see if it still works reasonably and test for some hours.

Maybe put a VGA cooler on the hybrid IC (guess it is such an IC?)

Simply attach with some "no more nails", it has good heat stability and bonds to most surfaces (but not blank aluminium).
 

#12

Joined Nov 30, 2010
18,224
That's a lot of years for the caps, yes? Isn't that well beyond their life cycle?
You can be sure no manufacturer said they would last 30 or 40 years, but I'm not going to replace them until they quit working properly, and they work every day. I also have a Ford with 140,000 miles on it and some light bulbs that I installed in 1984, but I'm not going to replace them until they do something besides work properly every day.
 

ronv

Joined Nov 12, 2008
3,770
Components have a "bathtub" failure curve, but the right hand side of the bathtub is way out there. You are probably more likely at this point to have a failure with a new IC than with your old one because the new one will need to live thru the left hand side of the bathtub. But in either case the probability of failure of a single IC is probably less than 1 or 2 in a million in a lifetime.
 

GopherT

Joined Nov 23, 2012
8,009
The caps are normally more problematic if they are large, if the unit has been sitting idle for long periods of time, and they are low quality.

Look on digikey for the range in prices for a 35 v 4700 uF capacitor. The range is broad and there is a reason some people pick the higher priced caps.
 

#12

Joined Nov 30, 2010
18,224
The left side of the bathtub is called, "infant mortality". Last seen by me as 3% in the first year for commercial grade chips.

If you can figure out when the right side of the bathtub gets to 3% per year, you will know when it's a 50-50 chance that replacing a known good IC will be have a probability of being an improvement.
 

bountyhunter

Joined Sep 7, 2009
2,512
Well, maybe, but as I said, we (the semi industry) made a few thousand tons of defective semis in the time frame of 1979 - 1981. It was so bad that we couldn't even get production runs through the fab. The units that passed final test died an indeterminate time later.

Ironically, I bought a very expensive Dual cassette recorder (state of the art then) in 1981 and the audio processing IC in one channel died in a few months. Sounds like karma to me.

As for infant mortalities:

In 1981, I worked at Qume and one of the things they did was send all of the incoming IC shipments out for a 100 hour burn in at 125C ambient temp.... to weed out the infant mortalities. Some of the lots showed a few percent failures, some showed as high as 50% failures. Anything higher than a few percent and the lot was rejected entirely.

One thing we found out early on: the stuff we got from japan at the time (like samsung) showed zero failures which meant they were burning them in before shipping.

Anyway, the infamous bathtub curve is always referred to but nobody has ever proved what hours correspond to what points on the curve. Well made semis generally outlive every other part of the system and badly made semis die at any time.
 

ronv

Joined Nov 12, 2008
3,770
Bad parts are bad parts, but most are worse early on - hence burn in.
We built disk drives - upwards of 30 million a year. I never saw a semiconductor failure rate get worse with age only mechanical parts like connectors from the vibration. Here is a little write up.
http://www.weibull.com/hotwire/issue21/hottopics21.htm
We weibulled every product looking for warrantee exposure.
 

takao21203

Joined Apr 28, 2012
3,702
semis can degrade from repeated ESD events and overloading. Or lets say they don't always die from it, sometimes recover again at least to the point something blinks.

After a while anyway it could be one IO bit is latching so it is only producing a fraction of the voltage swing.

Most time power Schottkys short and that's the end. Transistors can be more weird- even if the circuit has failed, sometimes they still produce a transistor function (amplifying) with a suitable tester (regulating the base current and with a LED so you can see it turning on).

It means the component went out of spec (degraded), but still somehow functions (not a total short or open contact).

Most of the time, ICs which run somer power current just pop, and that's the end irreversibly.

Such degrading events are rare, uncommon, can be caused by unusual handling, applying wrong voltage, ESD, desoldering/soldering again.

It would be totally rare just on their own to degrade for no reason within normal operation, except maybe (see above) badly manufactured ones.
 
Top