9
\$\begingroup\$

I have two lights that I use for working on stuff in my shop, and it seems like the PCB heated up enough that solder between two leads actually melted and bridged itself.

The light was working great until today, and suddenly got very dim - so I cracked it open to check.

Is there any reason besides heat and extra solder that this would happen?

I'm not sure what to think otherwise, but it does seem like there should be a heatsink on that LM317 at the least. Curious if this is a common occurrence that can be avoided or if it's a fluke.

Here's what the bridge of the bad light looks like:

Bad Light Bottom

Here's the bottom of the good light PCB:

enter image description here

Here's the board itself. Its used to drive 90 SMD LED's in series with ~250VDC from 120VAC/58mA

enter image description here

Here is a better lit picture of the top of the board. This is a picture of the actual bad board. The first one was a top image of the good board - which is the same but different color and brand components for some reason. The leads going to the LED's are disconnected so I could move it around easier.

enter image description here

Here's a closer closeup by the LM317 where it bridged as requested.

enter image description here

Edit: For anyone curious, I did fix the bridge and it seems like the 317 is bad. Lights are still dim, and DC output is at ~325 where the working one is at 250.

\$\endgroup\$
4
  • 3
    \$\begingroup\$ It looks to me like there is a bad solder joint on the "good" board where the red-and-white lead is attached. If there was a similarly bad joint on the failed board, that could cause local heating and melt the solder. The bad solder joint needs to be re-worked. \$\endgroup\$
    – Andrew Morton
    May 8, 2021 at 14:26
  • \$\begingroup\$ @AndrewMorton yeah you're right. These boards are pretty shotty all around. These are outdoor string lights and easily replaced, so I think at this point putting these kinds of boards in is how they get return business - or just for lack of caring. I should probably just build my own boards since these are so minimal. \$\endgroup\$
    – Devin Carpenter
    May 8, 2021 at 14:39
  • 3
    \$\begingroup\$ @DevinCarpenter It's well known that LED bulb manufacturers intentionally drive too much current through the LEDs to make them fail sooner, so you'll buy a replacement sooner. (Some high-up in Dubai recently mandated that lamps sold there not be designed this way, so it's possible to get ones that last longer if you import them.) I wouldn't be surprised if this is another attempt to do that, but it could just be poor design. \$\endgroup\$
    – Hearth
    May 8, 2021 at 15:00
  • \$\begingroup\$ The LM317 is wired for constant current, so it won’t be regulating the output voltage. \$\endgroup\$
    – HandyHowie
    May 8, 2021 at 22:39

5 Answers 5

Reset to default
12
\$\begingroup\$

I would suggest that the solder bridge has always been there and that the LM317 was never doing its job as a constant current supply. The current sense resistor was shorted out by that solder bridge. Your LEDs will have been getting overdriven by excessive current all the time and have eventually failed.

It should be obvious from looking at the resistor connected to those pads if it has overheated, it will look burnt. I think you will find that it looks ok.

I think once you have removed the solder bridge the circuit will work again. However, your LEDs will be broken in some way.

Since the LM317 is configured as a current regulator in this circuit, either off load or connected to a higher resistance faulty load, it’s voltage output will be higher than expected. You could measure the current supplied to your good LEDs by the good board and check that your suspect board supplies the same current to the good LEDs.

\$\endgroup\$
5
  • \$\begingroup\$ Hmm. Well I've tested both resistors and all the LED's one by one and they are still in working order and correct resistance. Honestly to me the pad does look burnt; the solder isn't shiny and smooth and it looks a buildup of burnt flux very close to the joint due to excessive heat. Also the circuit doesn't work entirely - still operating at much higher voltages than it's counterpart and the LEDs light up but are dimmer (I don't understand how that works). \$\endgroup\$
    – Devin Carpenter
    May 9, 2021 at 2:33
  • 4
    \$\begingroup\$ @DevinCarpenter Often LEDs don't fail entirely, but just lose their efficiency. You could try exchanging the working PCB to the LEDs that were with the bad PCB, to test if the LEDs are still ok or not. \$\endgroup\$
    – jpa
    May 9, 2021 at 6:53
  • \$\begingroup\$ I agree with @jpa and have added a little to my answer. \$\endgroup\$
    – HandyHowie
    May 9, 2021 at 7:06
  • 1
    \$\begingroup\$ It appears that the circuit uses a typical "capacitor dropper" supply from AC. That will automatically limit the current to some value, which is probably why the LEDs have worked as long as they have. \$\endgroup\$
    – jpa
    May 9, 2021 at 7:37
  • \$\begingroup\$ @jpa & Howie these are great ideas! I definitely didn't think to do that. I'll give it a shot and test whether it's the board now or the LED's that are the issue. \$\endgroup\$
    – Devin Carpenter
    May 10, 2021 at 1:44
10
\$\begingroup\$

It's possible; I've seen it happen when I messed up a design. It's usually a sign of either a very poorly designed board, or a broken board.

I'm not sure this is what happened on this board, though. The burnt flux residue adjacent to the pin looks a little suspicious.

I haven't reverse engineered the board to see what would happen if those pins were bridged, but it's possible that they were like that from the factory, and the bridge just didn't cause problems until prolonged use caused something to fail. Can you get us some close up photos of the top of the board, near where the bridge is? The picture you gave has that spot right in the shadow of the 317, and it's hard to see what's happened there.

\$\endgroup\$
4
  • 1
    \$\begingroup\$ Added the closeup and better lit images as requested. Yeah I agree though, that burnt flux really threw up some red flags. I tested the diodes, caps, fuse, resistors and varistor and they all seem to be working - but haven't tried the LM317. I'm about to take the solder off the bridge and see if that does the trick. Hopefully I don't have to order a new IC. \$\endgroup\$
    – Devin Carpenter
    May 8, 2021 at 13:45
  • 3
    \$\begingroup\$ @DevinCarpenter The 317 is cheap, should cost you about ten times more to have it shipped than the part itself. You can get them from ST for $0.56 each. I don't think this board is worth fixing, though. \$\endgroup\$
    – Hearth
    May 8, 2021 at 15:04
  • \$\begingroup\$ good to know. I'll test the 317 outside the circuit before I buy a new one. But these things would be good to order a few of at least right - considering the amount of applications they can be used for. Unless there's a better solution I suppose. \$\endgroup\$
    – Devin Carpenter
    May 9, 2021 at 2:37
  • \$\begingroup\$ @DevinCarpenter Sure, the LM317 is a handy part to keep around. I'd go for the TLV1117 over the '317, personally, since it's a lower-dropout part without being much more expensive. Lower current output, though. \$\endgroup\$
    – Hearth
    May 16, 2021 at 20:01
7
\$\begingroup\$

Yes, it is absolutely possible. Overheating SMDs will fall off the board when the solder melts, a kind of "fail safe" mechanism... but thru-hole parts won't.

It looks like the resistor was the source of the heat, that's believable, as resistors can get above solder melting temperature without showing too much visible damage. Plus it is placed with the resistor body just aiming down into the hole that melted, with less than 2mm wire, so that's great for heat transfer...

So you could replace the resistor with a higher power one, and when you solder it, don't cut the legs too short so it stands above the board and gets airflow...

do not check the temperature of the LM317 with you finger while this thing is plugged in

\$\endgroup\$
2
  • \$\begingroup\$ Hmm ok so the length of the legs going into the board can be enough to help with heat displacement? Wow I had no idea. By higher power you just mean a bigger one that can take more amps right? And thanks for the tip - I bet it's got, and I'm just worried enough by getting a shock from the AC to not touch it :D \$\endgroup\$
    – Devin Carpenter
    May 8, 2021 at 13:50
  • 1
    \$\begingroup\$ Yeah there is no transformer, so this thing is not isolated. \$\endgroup\$
    – bobflux
    May 8, 2021 at 14:28
6
\$\begingroup\$

Here's the bad board and the good board together - they are as different as chalk and cheese: -

enter image description here

Notice the green coating on the lower (good PCBA) - that's solder resist and would, if present on the bad board, largely prevent happening what you claim to have happened. It looks to me like you bought the bad device from a totally different source than the good device and somebody has re-engineered the board and applied significant cost-cutting. I've be very doubtful about the failed product.

\$\endgroup\$
7
  • 1
    \$\begingroup\$ Huh - you can tell the top board is a cheaper quality just by the color? I thought they would be made of the same silicon makeup? These boards were both in devices from the same manufacturer - and so I just figured with the different boards/parts they were just buying/manufacturing these in bulk and what they ordered depended on part availability. Strange that they are mixed quality devices unless some were made to break? They are lights meant to be replaced in a light string.. \$\endgroup\$
    – Devin Carpenter
    May 8, 2021 at 13:56
  • 2
    \$\begingroup\$ @DevinCarpenter there is no solder resist coating on the bad board. This is a significant difference and, who is to say that whoever sold you the lamps have not decided along the way to buy finished circuit boards from a crappy supplier. There are plenty of suppliers who don't have a significant quality control system and this can easily happen. It' looks like a whole supply chain quality problem to me. \$\endgroup\$
    – Andy aka
    May 8, 2021 at 14:32
  • \$\begingroup\$ I didn't know that that was something not standard all around. Thanks for the useful bit of info! \$\endgroup\$
    – Devin Carpenter
    May 8, 2021 at 14:41
  • \$\begingroup\$ @Andyaka Are you sure there's no solder resist on the bad board? I wouldn't expect it to still be green if so, especially not the traces. It might be lower quality solder mask though. \$\endgroup\$
    – Hearth
    May 8, 2021 at 14:57
  • 3
    \$\begingroup\$ Bottom looks like FR-4, top looks like Phenolic material. Also single vs double layer. \$\endgroup\$
    – Wesley Lee
    May 8, 2021 at 19:21
6
\$\begingroup\$

It's possible. But that solder bridge is pretty long and skinny; even in the absence of solder resist on the board it's unlikely that you'd get such a thin neck connected to another pad. It's much more likely that the bad board was assembled incorrectly, with the wire going into the pad right on the edge of the board stuck in far enough and folded over toward the other pad so that a bridge was formed.

If you were to remove the solder from the two pads that are bridged, I suspect you'll find that there's wire underneath, and that you're not looking at a true solder bridge at all -- rather, you're looking at a wire bridge that's backed up with solder.

\$\endgroup\$
2
  • 2
    \$\begingroup\$ Interesting theory. I did remove the bridge and there was no wire under it that I saw. The wire wasn't leaning from what I could tell either and seemed to be a pretty minimal amount of through-hole, but it still looked like a lot of solder on that joint. \$\endgroup\$
    – Devin Carpenter
    May 9, 2021 at 2:17
  • \$\begingroup\$ Well, so much for that hypothesis! \$\endgroup\$
    – TimWescott
    May 9, 2021 at 3:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.