Failed resistors

[shortbus=]

I guess you mean questions of the form:

No. I mean an "electronic engineer" that has to come here and ask questions about how to do design work. Work for a job and company he is already working for. Any more and we take it to PM.

 

[DrG]

Any more and we take it to PM.

You're going to take it to the Prime Minister? Sheesh, it is just a couple of resistors....it's not like it was some kind of a perfect call.

 

[gophert]

it's not like it was some kind of a perfect call.

Every time I hear, "a perfect call.", I think of Disney's, the goofy movie, and Goofy's "perfect cast" while teaching his sone how to go fly fishing. (I don't know why there is a John Cena commercial at the end of the video.

 

[Mickster]

No. I mean an "electronic engineer" that has to come here and ask questions about how to do design work. Work for a job and company he is already working for. Any more and we take it to PM.

The one who always chooses air travel, but returns to the origin.

 

[gophert]

I'm not seeing what the electrical design flaw might be. Just saying it is a common design error doesn't communicate enough information to make it true on its own. What is the error and what makes it a better design?

I just don't get it (the design issue) as currently explained.

 

[Nigel Goodwin]

I'm not seeing what the electrical design flaw might be. Just saying it is a common design error doesn't communicate enough information to make it true on its own. What is the error and what makes it a better design?

I just don't get it (the design issue) as currently explained.

It's a VERY, VERY common failure - and has been for decades - fitting a proper sized single resistor cures the issue - I would have thought the 'design error ' is obvious - reliable vs. unreliable?.

 

[alec_t]

I would agree that it's a design error to use a technique which is known to be unreliable in practice, but I think what everyone wants to know is 'WHY does it prove unreliable?'. My guess is that multiple resistors don't always share the load equally or are differently stressed. Tolerances, or adjacent stray capacitance or inductance might be a cause of that. Differences in thermal conditions might be a factor.

 

[Nigel Goodwin]

I would agree that it's a design error to use a technique which is known to be unreliable in practice, but I think what everyone wants to know is 'WHY does it prove unreliable?'. My guess is that multiple resistors don't always share the load equally or are differently stressed. Tolerances, or adjacent stray capacitance or inductance might be a cause of that. Differences in thermal conditions might be a factor.

Lot's of theories over the years - but nothing ever proved. It just 'is'

 

[Visitor]

Tolerance of the resistors wouldn't matter, as long as they are well within their specs. Even if they are of widely different specs, the current through each resistor is the same. One can't somehow "hog the load".

 

[Beau Schwabe]

Is the problem also persistent with through hole resistors or is this strictly a chip resistor problem?

And what if you use something like four 10k resistors in a classic text book series parallel configuration to achieve the same 10k resistance to dissipate the power load?

Why learn series parallel if it can't ever be applied?

- Just playing devil’s advocate here

 

[gophert]

This well-known problem has virtually no documentation on Internet forums, manufacturers Application Notes, or manufacturers' datasheets for chip resistors (I've checked several). Can anyone point to any documentation on the subject?
Is it an electric power issue, a thermal/mechanical stress issue, or witchcraft issue?

edit: I used the word "virtually" in the first sentance because I did find this thread as the only source of info on series resistor failures smd. To be honest, I found nothing.

 

[Beau Schwabe]

"This well-known problem has virtually no documentation on Internet forums... " - Honestly, just think how ridiculous this sounds.

Take for example a simple voltage divider with two resistors .... IN SERIES . Your telling me, that if I happen to NOT connect the center tap to anything in my circuit that one or both of the resistors in the divider will fail because they are connected in series? - That's essentially what everyone is claiming ... that's hog wash !!!

 

[ronsimpson]

I commonly use series resistors with high voltage. The larger sizes of resistors are much harder to get.

From my experience; Vishay said 150V on a 0805 resistor. When the boards come back from China the resistors are not from Vishay. I can't find out who made the resistors. They change color after use and that is not good.
China will say "yes Vishay" but no then are not. Then China will say "very good resistor just like Vishay" but they are not.

 

[rjenkinsgb]

China will say "yes Vishay" but no then are not. Then China will say "very good resistor just like Vishay" but they are not.

At a guess, I'd say lower cost manufacturers just assign a value to every resistor out the plant, rather than each being made to a tolerance and being rejected if it's not right.
Or they are buying the rejects from larger manufacturers and applying their own value markings.

That could end up with higher value resistors being lower value designs with flaws in the resistive track or layer, so most dissipation is in one weak spot rather than evenly distributed over the length of the resistor.

Imagine filing a notch in a 10K conventional carbon resistor to increase its value to 100K - it could read right but it would not take the same power, all dissipation would be in the "neck". A flaw in the resistive material applied to a mass produced part would have the same effect..

 

[Diver300]

"This well-known problem has virtually no documentation on Internet forums... " - Honestly, just think how ridiculous this sounds.

Take for example a simple voltage divider with two resistors .... IN SERIES . Your telling me, that if I happen to NOT connect the center tap to anything in my circuit that one or both of the resistors in the divider will fail because they are connected in series? - That's essentially what everyone is claiming ... that's hog wash !!!

The point here is not that two resistors are in series. It is that two resistors are in series where the voltage or power is too much for one resistor.

If one resistor could be used, the circuit would have been designed with just one resistor.

On a potential divider, two resistors are needed so that connection can be made to the mid point, but that doesn't mean that the resistors are being run anywhere near their maximum ratings in a potential divider.

In the circuit that I was repairing, it appeared that the resistors were in series because of the voltage rating required. There was around 320 V across three resistors. I had three units, all of which had failed in the same way, with one of the resistors going open circuit, so there clearly is an issue.

I could have been as simple as the wrong resistors being used, as the maximum voltage rating will be different for different types.

 

[alec_t]

the current through each resistor is the same.

That is what designers commonly assume. But transients and stray capacitance can result in current differences, as the sim below demonstrates. C1 represents stray capacitance affecting primarily one of two series resistors.
Whether or not the difference is sufficient to cause failure is inconclusive.
.

 

[Nigel Goodwin]

This well-known problem has virtually no documentation on Internet forums, manufacturers Application Notes, or manufacturers' datasheets for chip resistors (I've checked several). Can anyone point to any documentation on the subject?
Is it an electric power issue, a thermal/mechanical stress issue, or witchcraft issue?

edit: I used the word "virtually" in the first sentance because I did find this thread as the only source of info on series resistor failures smd. To be honest, I found nothing.

It's not a specific chip resistor problem, it was exactly the same in the through hole days as well.

As I suggested above, a 'possible' reason for the failures is that they tend to be in the 'unreliable values' range - as anyone with old service experience will tell you, certain values (or range of values) seem to be far more unreliable than others. For example, failure of the anode loads in triode amplifier stages, historically these have always failed - and still do on brand new equipment (guitar amps). While the wattage and voltage ratings are well within their specifications, they die - replacing them with a resistor twice the wattage means they will never fail again.

When you repair a valve guitar amp, first thing to do is run your meter along the anodes of the triodes (pins 1 & 6), you'll invariably find one, on most, of them will be low voltage - and all will be way out of spec, always going higher until they are effectively O/C.

 

[DrG]

This well-known problem has virtually no documentation on Internet forums, manufacturers Application Notes, or manufacturers' datasheets for chip resistors .

I found MANY such examples of circuit failures in series resistors. I concede, however, that I do not have an MA in Journaling like you do, so I probably just don’t know how to use Google.

 

[DrG]

"This well-known problem has virtually no documentation on Internet forums... " - Honestly, just think how ridiculous this sounds.

Take for example a simple voltage divider with two resistors .... IN SERIES . Your telling me, that if I happen to NOT connect the center tap to anything in my circuit that one or both of the resistors in the divider will fail because they are connected in series? - That's essentially what everyone is claiming ... that's hog wash !!!

Take a step back please and objectively look at the content of the thread.

“that is essentially what everybody is claiming” Well, no, in fact that is not what everybody is claiming.

I know this is a lengthy response that might exceed the attention span of some folks, but I hope that you (based on that quite cool discussion of reversing SCL and SDA to avoid addressing issues [after several head pounding hours, I can’t see why it would not work in at least some instances]) will take the time to read it all.

The OP is reporting experiences of failures in series resistors in a particular product.

Nigel is reporting that such failures (in series resistors) are a common based on his considerable experience repairing a variety of products. This is his observation. IOW he has seen that problem many times – there are several resistors in series and that is where the failure has occurred.

I have no reason to doubt the veracity of both claims. Neither would have any reason to make it up and both would have the experience, in one case specifically, and in the other more generally.

Particularly notable is the additional information that when the multiple series resistors were exchanged with a single and higher wattage resistor (relative to the wattage's of the individual series resistors), and the problem was fixed and did not recur.

One of several possible reasons that multiple series resistors could be used rather than a single resistor is so that one could use lower wattage parts. There is agreement on this, both in the thread and in many other places. Not to say that there are no other reasons.

If one assumes the same likelihood of a resistor being outside of the rated wattage is the same or similar in the relevant cases, than the likelihood of a failure using three series resistors because it is out of tolerance is 3 times that of using a single resistor. Since a product failure occurs in both instances (in general), that analysis alone can explain the phenomenon. No ethereal, faith-based, explanations need be invoked.

Rated wattage is only one characteristic of the resistor and rated voltage is also applicable as well as time and heat-based long-term changes. The difference in likelihood of these characteristics could also follow the multiple, with regard to probability, as described earlier.

The number of products containing multiple resistors in series vs. a single resistor is not given. No claim regarding the percentage of failures in the series case vs. the single case is made and need be speculated, because a very low percentage can easily be consistent with the observations.

Obviously, the operating characteristics of the circuit relative to the tolerances of the resistors are a critical factor.

The OP is puzzled as he does not see operating characteristics close to tolerance. But, it is a small sample and there has already been a failure and the deviations in tolerance of the resistors has not been determined (and cannot be determined in the case of the failed ones).

Nigel’s anecdotal reports also do not include a comprehensive discussion of each circuit’s performance in this regard. I would not expect them to. Even a novice like myself can understand that cases exist with particular products having particular flaws that are revealed post-production. The idea that “everyone” might know this is completely understandable if you qualify the “everyone” as being those individuals who do the repair work.

Your simple resistor divider example could, in fact, demonstrate precisely the same phenomena, but you would need to replicate the experiment many times and know something specific about the tolerances and their deviations.

Finally, I acknowledge that Nigel (in post #37) suggested, what I believe to be, the same idea I am presenting here, but in a much briefer form.

 

[gophert]

I found MANY such examples of circuit failures in series resistors. I concede, however, that I do not have an MA in Journaling like you do, so I probably just don’t know how to use Google.

If you really "found MANY such examples", the ungodly long rambling in your following post could have certainly been abbreviated. Do you care to share any of the results you found or do you have to keep them in your box of secrets?