component failure mode thread

[Chuck H]

While reading this:

http://music-electronics-forum.com/t27307/

I thought it would be good to have a thread that examines the failure mode of different components and how it applies to design and safety criteria. There are so many great threads here that people download for reference. This could be one.

So, different types of resistors, caps, etc. Do they tend to go open or short? Any considerations for choosing a particular type of component for a particular application?

[Steve Conner]

Well, it shouldn't take long to wrap this one up! Several books could be written about it. If they haven't, maybe they should be.

As a massive generalisation:

Resistors and switches tend to fail open, caps tend to fail short. (But electrolytic caps can fail high-ESR, which is a kind of half-hearted open.)

Transformers and inductors can fail in several ways: an open winding, a shorted turn, a short between two windings that are supposed to be isolated, a short between a winding and the core. Any of the above can be intermittent: they won't show up under testing with a meter, but will cause trouble when high voltage is applied.

Semiconductors tend to fail catastrophically short. If they have bond wires (most diodes don't, all 3+ legged silicooties do) then a power supply capable of high current can blow the bond wires, shattering the plastic package. (I once got hit in the forehead by some MOSFET shrapnel this way.)

Potentiometers fail in the same way as resistors, but also "go scratchy".

Any components may emit smoke, hence the tired old joke about the "magic smoke getting out". However, good quality modern components are not supposed to spew toxic fumes or light your equipment on fire. UL rates plastic parts for flammability amongst other things.

Any more contributors?

[Chuck H]

No, that didn't take long.

What about the nature of failures and how it might be beneficial to select a certain part because how/when it fails will be safer for the user or circuit? Is there any difference in the conditions or causes for failure between ceramic and film caps? Or do they fail differently in any way? What about self healing film caps? When called upon to "self heal" is there any change in value or rating? And so is there any circuits in which you would want to choose or avoid specifically ceramic, film or self healing film for a safety or reliability reason? Resistor value drifts (sort of a failure)? Some types do it worse than others but may have other properties that make them better for some circuits. This is the sort of discussion I'm hoping for. Though I'm probably overthinking it.

[Steve Conner]

Here are the main safety issues I'm aware of:

You're limited in what you can connect directly to the line. You should only choose parts festooned with the little safety approval logos: UL, VDE, Semko and so on. The more the better.

As far as connecting capacitors to the line: Between live and neutral lines, they have to be "Class X": a spec that sets out self-healing properties. If used between either of the lines and ground (like the "death cap") they have to meet the more stringent "Class Y" spec and there is a maximum allowed value.

Like everyone else in the electronics industry, I avoid the safety issues as much as possible by designing my equipment to use external power supplies that I buy in complete. If it all runs off 12V, the risk of it killing someone is that much less.

Unfortunately this doesn't apply to the average tube guitar amp. In Europe, musical instrument amps would have to meet the requirements of the Low Voltage directive and in the USA there is probably something similar, maybe UL. The details vary, but the basic idea is all sorts of tests to make sure it won't catch fire or electrocute anyone even under serious abuse.

You may not actually have to get your product tested, if you take the time to understand the requirements and ensure that your product would pass if it were tested, and you don't get caught.

I'm sure RG will accuse me of oversimplifying greatly, and I am. Maybe he can be bothered typing out a fuller explanation. But as you can see, safety tends to be considered at a system level, not at a component level. Any modern components you can buy from reputable distributors should meet UL flammability criteria, and that's about the best you can ask at the component level.

[Tom Phillips]

Chuck,
Some in the electronics inductry deal with this via Failure mode, effects and criticality analysis (FMECA)
The questions you are asking are burried within the FMECA process. I say "burried" because there is so much more covered within the process. I think you will find the explanation at Failure mode, effects, and criticality analysis - Wikipedia, the free encyclopedia an interesting read. Or just google FMECA and you will get lots of hits.
Regards,
Tom

[trobbins]

The input valve plate-grid capacitor is an interesting safety related topic. If this is still being used on present production equipment by major manufacturers, then it would have been assessed. It would be interesting to know if it was a Y rated cap - or deemed a limited current type circuit - or not required to be 'shorted' wrt compliance requirement.

Not a purveyor of new amp equipment, I'd also be interested in what has been done about output voltage safety for larger amps - whether amp manuals specifically make a warning; gone over to Speakons; ?? Whilst not specifically a failure mode comment, it is the other main interface.

[R.G.]

I'm sure RG will accuse me of oversimplifying greatly, and I am. Maybe he can be bothered typing out a fuller explanation.

Moi? Thinking that something is oversimplified? I'm shocked. Absolutely shocked that you'd think such a thing.

Actually I don't have much to add. Safety issues other than warnings are probably not a good thing for this forum. Many of the participants won't follow the whole procedure, and will use a few guidelines a a safety talisman.

On the components stuff, one item that people don't think about is that the weak point in pots is the wiper. This tends to fail high resistance or open. That's the source of much of the scratchiness, an intermittent contact.

I'd also mention that sometimes it's not the failing component that's got a problem. Electro caps, for instance, should always be given some breathing room (literally!) from hot components. They have enough problems with internal heating without being mounted close to hot transformers and power tubes. Part failures can sometimes be traced to mounting problems on the PCB, or too much flex under vibration. Failure to properly apply heat sink goo or to tighten heat sink mounting screws correctly (not too tight or too loose) can cause cooking failures in power parts.

As for the subsidiary questions:
- Never, ever intentionally put in a resistor that will dissipate 100% of its rating in normal-worst operation: that is, full output, hot but human survivable air temps, etc. Derate to about 50% of rated for maximum full-output operation. This has the added effect of reducing part temps, and that in general increases the life of the part, for any part. Also, resistors tend to reach temps of 150-200C (not F) when run at 100% of their rated power. In addition to being a human burn hazard, the can cook their solder joints and char PCBs.
- Ditto, never intentionally bias/set up a circuit element like tube, transistor, whatever, that's running at 100% of its power rating for normal/full output conditions. Same comments as on resistors.
- Never, ever, intentionally use a cap whose rating will be exceeded under conditions of 20% high line voltage. In the USA, that's about 144Vac on the incoming line, and for 240Vac countries, 288Vac. The AC power line can and will do this to you. If your cap is marginal there, you'll have a reasonable expectation of caps letting go. Maybe not on the first transient, but soon, and for the rest of your life... er, I mean, one of the transients will get it and kill it early. The results are most spectacular on power filters, but they can happen to coupling caps, cathode caps, etc. too.
- Always think about what happens when the junior-high-school dropout guitar wizard decides to leave the thing turned on with no tube in it. For a whole weekend. Or with just a rectifier tube in it. The idea of designing in at least partial immunity to forseeable misuse by the owner is a part of any serious design procedure. Humans WILL intuitively run the thing in the worst possible way.

In general, for choosing a part for an application, you have to know (1) the applied voltages, including min, max, and normal (2) the applied currents, min, nominal, and max, (3) the internal power dissipation max and (4) the external heating load. When you know those, you can pick a part. Otherwise, you're guessing.

[Chuck H]

I'd like to relate an early experience in amp modding... I used a ceramic cap in a location that got fairly warm/hot and the cap, while safely over rated for voltage, did change value noably as the amp changed temperature. The result was an amp that sounded different under different operating and environmental conditions. Of course I don't make this sort of mistake anymore. The point of this thread is to examine these things so that builders will know how to choose components and/or how to position them in the design. So far, so good. I remember the many ratings for ceramic caps being confusing.

It was brought up in another thread that for an OT primary shunt filter a carbon comp resistor might be a good choice because they are resistant to damage from voltage spikes. Of course CC resistors are known for their "mojo" but generally shunned because of their thermal and shot noise. This is also the sort of thing I'm hoping to define. But it seems this subject is hitting the ground without even a bounce.

[Steve Conner]

I'm struggling to come up with suitable nuggets of mojo for this thread. By and large, modern components are high quality and reliable, and if you follow RG's guidelines they will give a long and happy life. The same isn't necessarily true of whatever NOS garb... erm... treasures you might pick up on EBay. Speaking as an engineer, I can't understand why people use "vintage" capacitors.

Anyway. Often when you have a scratchy pot, the pot itself is fine. The problem is DC flowing in the wiper along with the desired audio signal. Many classic guitar amp circuits allow this by connecting the wiper straight to a tube grid, which can leak a surprising amount of current. The first tube can make your guitar's volume pot scratchy. Bipolar op-amps like the NE5532 do exactly the same thing. So really the DC should be kept off the wiper with a capacitor.

Relays and switches are unreliable beasts that deserve another book. For switching audio signals, you want a hermetically sealed unit with gold or mercury wetted contacts. But nobody ever uses these because of the expense.

[Enzo]

I have a bunch of relays from telephone company systems. Now THEM are fancy relays, as you describe.


I learned a lot about relays working on pinball machines in another industry. Fancy contacts are cool, but relays are designed to clean themselves too. WHen contacts close, they overmake - they continue past the point of contact a ways, causing the two contact surfaces to "scrub." This tends to rub the surface clean with each closure. Even in smoky taverns and other hostile environments, it was amazing how long relay contacts would function.

I still have a collection of relay tools next to me on the bench. contact blade formers, point files and burnishers. The springy contacts in those Switchcraft jacks on all the Fenders are not at all unlike relay contacts.


Here's one for you. Those little plastic transistors, the black TO92 ones? They look opaque, but they aren't, at least not at IR wavelengths. On something like a Fuzz Face, probably doesn't matter. But a sensitive circuit or a not overly stable one can be affected by bright lights. And that works both ways. Under the bright work lights of your bench a circuit might work great, but in the dark it acts different. Or vice versa. Or try as you might to ground away some hum problem, never realizing it is a transistor picking up the flourescents optically, not the sensitive gain circuit.

I'd certainly defer to Steve and RG on this, but it seems that if you design something with respect for the parts so things don;t WANT to fail, then we can just worry about the random parts failures.

If we are talking about the often suggest "gee, what if the preamp tube shorts from plate to grid, won;t my guitar have B+ all over it?" I don;t lose any sleep over that.

I recall many years ago working on a dollar bill validator. It worked OK sitting there, but as soon as I turned in on its face for some adjustment, it triggers a cycle. I went nuts for hours tring to find the loose connection, funny relay, whatEVER that was sensitive to position. I looked for loose hardware touching, tried flexing all the boards and subassemblies. Then one magic moment I discovered that it was my work light shining onto something that was doing it. Never would have been an issue buttoned up inside the equipment, but I didn;t know that. A piece of black tape over a transistor, and problem solved.