Pop goes my headlight

[Stupot]

Before everyone starts ranting about another " Speleotechnics Post "
i would just like to know if this is a common problem.

Emergered from a sump this evening and whilst taking off my mask etc notised a strong burning smell :shock: looked down at my helmet and saw that 1 set of 7 LEDs had gone out on my 14 LED Headlite, upon further inspection one of the resistors had fried itself.

Upon speaking to Jrat in the Pub apparently his had done the same only last week, i really hope this is not going to become a common occurence or iam going to revert back to a trusty candle.

Stu.

 

[potholer]

A burning smell *outside* the headset? How big was the resistor that burnt out?
Are there any signs/possibilities of a short circuit inside the headset - screw loose, bare wires that could touch, etc?

 

[potholer]

It seems unlikely a resistor would be underspecified in terms of power handling. I assume you were using a regular Headlite pack of some form?

I wonder if it's possible for 5mm LEDs to fail short-circuit (I'd have thought an open-circuit failure was much more likely). Can you check the LEDs to see if any are short-circuit?

 

[Dave H]

I've never come across one failing short-circuit. Water combined with the warm/hot resistor is the most likely suspect, in my opinion.

 

[potholer]

You'd think with the low impedeance of the circuitry that water would have little effect in conductance terms, and unless something was very hot already, one wouldn't expect water to cause acute physically damage. Possibly at worst a hot component cracking, but unless a short was somehow caused, the burning smell would seem to indicate something overheating, not just being quenched from a normal high running temperature and and breaking.

How hot do the resistors normally run, does anyone know roughly what value they are?

A back-of-the-envelope calculation would suggest that unless the diodes are run at rather more than the standard current of 20-30mA each, the difference between the usual (or maximum) voltage from a nominal 3.6V pack and the typical forward voltage 'consumed' by the diodes wouldn't be enough to overheat even a regular-sized ~0.5W resistor

 

[SamT]

:hi:

 

[Dave H]

If people really want to get to the bottom of this, then I would suggest the possibility of getting a broken unit to me (or some high quality photo's) and I could get our materials specialist to have a quick look at it.
Our PCB's have to work in a cold tank in Siberia(-50C), in a vibrating(5g upto 2MHz) helicopter over Iraq(+85C), and survive in high humidity conditions(95% with 5%salt) - all without active cooling. So if it is an environmental problem our guys would be able to make a few suggestions.

 

[Stupot]

Here's a picture of the culprit component although the burn marks are on the other side.

Stu.

 

[Dave H]

Stu,

A piccy of the burn would (apparently) be most enlightening (pun intended)

Dave H

 

[Stupot]

Hey Dave,

I can't really get a good piccy of it as it's on the revearse, but it looks a dark brown colour and the varnish type coating around the component has come away leaving the metal underneath.

I haven't managed too drop it off yet for repair and noticed last night that on the blue wire there were a few "nicks" which had slightly exposed the bare wire underneath, now i am pretty sure these are caused when the unit is opened up and then srewed back down after drying??

This may have contributed towards the problem.

Stu.

 

[Dave H]

Stu,

Wires - probably not - you would at least see a little melting around the nicks if that was the problem.

I don't have a lamp set like that so I don't know exactly what the layout is.

Fatigue cracking of SMT resistors does sometimes happen (unlikely), or you could have accidentally crushed the resistor slightly when the caplamp was put together. A cracked resistor will not necessarily fail instantly, but could generate heat along the fracture line, and could, if the heat build up was sufficient, cause a fire in a varnish type conformal coating. This could be many hours of use after the initial damage occured.

Some components absorb minute amounts of water and this can cause early failure, but it is unlikely in this situation.

Without seeing the actual burn patterns, our materials guys' best guess is that the resistor has become physically damaged at some point and the failure is due to that.

 

[Mine Explorer]

I'm slightly puzzled by the photograph and subsequent comment.

First off, looking at the photograph I'd say the source of your problems 'aint with the resistor. Have you actually had a meter across it and confirmed that it has failed?

The reason for my statement comes from two observations.

1) The conventional, through-hole, resistor is sitting on side 'A' of the printed circuit board, with no signs of burning beneath it. The opposite side of the circuit board (side 'B') has extensive burn marks.

Question: If the resistor got hot enough to burn the board to a crisp on side 'B', why hasn't it marked side 'A' sitting right next to the (very hot) component??


2) When a resistor gets hot enough to burn the circuit board it is getting VERY hot. 'FR4' pcb material starts to burn at around 130 deg. C. I don't know what material speleotechnics use, but it gives you an idea. When a resistor has been operating at these temperatures for extended periods of time (ie. long enough to cause extensive burning of the pcb) it usually becomes a bit discoloured itself. On dark resistors this might not be too noticeable, but on beige or pale blue ones it's often fairly clear. The photograph shows a perfect looking pale blue resistor.


This still leaves the question as to what caused the problem. The comment:

the varnish type coating around the component has come away leaving the metal underneath.

The photograph clearly shows the component and clearly shows no damage on the component side. The speleotecnics 7 & 14 LED headsets I've seen don't use surface mount resistors (in fact the first ones I saw didn't have any form of current limiting device!), so the other side of the board just features the tracks.

For the board material to burn there must have been something hot in close contact with it, this could either be in the back of the headset, OR it could be the pcb track itself. I also note that the photograph shows a dollop of hot-melt glue. If this stuff is in the area then it will stink to high heaven if it gets burnt (- trust me, I nearly evacuated our whole lab when a diode gave up the will to live under a coating of hot-melt!). Not to mention the pcb material itself, this smells awful as it burns! In comparison an overly hot resistor isn't too obnoxious.


From the photograph and comments I'd say the most likely cause is a short circuit somewhere that has drawn exessive amounts of current through the pcb track, causing it to heat up and burn the pcb. Unlike Oldhams, the speleotechnics system doesn't feature any form of over current protection to prevent damage.

Where the short circuit came from is the next puzzle. As already mentioned, I think you can discount any water. This may lead to electrolytic corrosion, but at 3.6V it's not going to cause huge currents to flow. From the comments the more likely cause could be the nicks in the cable. If these have been there for some time and have gone unnoticed then it may explain the problem. Having installed one of those reflectors on someone elses behalf I already know it's a tight squeeze and you can't guarantee where the wires are going to lie.

On their own the nicks are unlikely to cause a problem, BUT the switch mechanism inside the headset consists of bare strips of moving metal. I think the most likely cause of your problem is that one of the nicks in the cable has come into contact with possibly the switch (or other terminal) and has created a short circuit. I seem to remember the tracks on the LED units are weedy thin things so will quickly heat up with excessive current flow. In comparison, the nicked cable and the terminals are both substantially thicker and able to handle the current without signs of damage. As the track heated up it burnt the PCB and quite likely fused it, hence why half the LEDs have stopped working. I haven't checked, but I think there's only one resistor for all the LEDs, so if the resistor had blown ALL the LEDs would stop working, whereas the original comment said only 7 out of the total 14 had "gone west".



OK, it's only a theory, to prove it one would need to see a photo of the other side of the board and to see exactly which wires are nicked and how large the nick is!

Cue next suggestion....

Cheerio,
Mine X.


PS: Just looking at the photograph again, by the look of it, the resistor sits at either the 3 O'Clock or 9 O'Clock position when it's in the headset - ie. 50% chance it's directly over the switch mechanism...

 

[Dave H]

Mine X,

Your comments rather underline my request to see the headset. I had assumed that the arrow was pointing to a SMT resistor on the back of the board.

One could suggest that the switch has actually caused the nicks in the wire?

I'd better dip out of this topic until I've actually seen a headset with this modification!

 

[Mine Explorer]

Mine X,
One could suggest that the switch has actually caused the nicks in the wire?

Mmm, I wouldn't say it's impossible. But I would think it unlikely. Again, a further photo of the nicks might give some clues?

The most useful view would be a photo of the burnt bit as that should show if the cause of the heat was a hot track, would show if the track has fused and even show if a SMT resistor was sitting on the back.

If there was a surface mount beasty sitting there, it would want to be a '1206' at the very least. I think a freshly charged headlite battery will sit at around 4.5V ish for a short time. If one surface mount resistor is used to limit the current for 14 LEDs, each running at 20mA the resistor will dissipate 252mW until the terminal voltage has dropped. A '1206' resistor is rated for 250mW! Having said that, for UL certification (which doesn't apply to speleotechnics equipment) a 1206 resistor at 250mW will run too hot for the PCB material (assuming FR4, which we use), so a good guide is not to take the power above 125mW.

Of course, if a SMT resistor was on the other side, it would raise the question what the beafy conventional device is doing there?!

 

[Stupot]

Thanks for all the pointers guy's but the lamp went on it's merry way to Speleotechnics today, one thing i will say is that there was no burn marks on the board just the component and there was no signs of any burning around the "nicks" in the wires.

The theroy about the switch causing a problem is a good one and i can see how it could get in the way, i have asked Speleotechnics for a report on what had caused the failure, so just maybe they might tell me :roll:

Thanks again.

Stu.

 

[Dave H]

Do we really have no members of this forum who work for equipment manufacturers who can help us with such matters? Or are they just hiding?

 

[potholer]

Regarding the 'other side' comment, subsequent puzzlement, and the later comment about the back of the PCB being OK, am I right in thinking that the non-visible side of the resistor (ie the side facing away from the edge of the circuit board) was the bit showing damage, which would explain why it wasn't easy to photograph?

Following up that speculation with a little more, I guess it's quite possible that a nicked wire could end up shorting to a component lead stub poking out of the back of the PCB, effectively shorting the resistor across the battery.
That *might* give a fairly graceful failure, putting something up to maybe ~10x the regular amount of power (ie heat) into the resistor, and quite possibly only for a brief time, depending on the nature of the contact - maybe just enough heat to cause the resistor to fail without doing so particularly spectacularly, and (especially with an intermittent or imperfect contact) also maybe sub-critically heating the resistor for long enough to heat up a bit of hot-melt glue, if the hot-melt was indeed overheated.
*If* an accidental connectio was the cause of the problem, one which caused the resistor to be connected across the battery would likely not cause a sufficient current flow to melt any nicked wire that could be the culprit, whereas a direct short across the battery via tracks and wires only would at least have a fair chance of causing such visible signs of damage.

Given the initial comment about '... one of the resistors had fried itself ...', I'd assume each set of 7 LEDs had its own resistor, so a PCB track failure isn't *necessarily* indicated by the half-failure of the unit.

I do know of someone who shorted out and killed a S-T Lithium pack whilst fitting a 14-LED reflector with the battery still connected, which he reckoned was something to do with wires protruding from the PCB, so I guess the physical protection against short circuit isn't absolute, even if it is normally sufficient.
I'll ask him tomorrow, (and maybe grab a look at his reflector unit) when we meet up for what I have been confidently assured may be our final push in the dig in [censored for personal safety reasons].