I recently soldered a number of white LEDs to a PCB using a Weller "standard" soldering station. I noticed some LEDs were glowing slightly during soldering. The PCB was placed on an ESD mat, but I doubt that this had an influence. Is that a thermal or electrical effect?

I suspect the former, because I am quite sure the metal iron tip is grounded.

  • \$\begingroup\$ how is the circuit to which you were soldering it? if i had to guess i would say there is a thermo electric effect over there, maybe a thermocouple formed somewhere or you had an arrangement of semiconductors that worked like a peltier cell, however more info is needed for a proper assesment \$\endgroup\$ – diegogmx Jul 8 at 12:53
  • 3
    \$\begingroup\$ @diegogmx No conceivable thermocouple effect will give you the >1V required to light a visible LED at the available temperature differences. It's clearly mains leakage. \$\endgroup\$ – Spehro Pefhany Jul 8 at 14:20

Although LEDs ought to include a reverse diode for ESD protection, the heater coil induces a magnetic field into the high impedance LED with enough mutual coupling to generate xxx µA of dim current.


I sold over 1 million 5 mm LEDs, and it wasn’t until I flew down to this one client in New Zealand and saw their poor handling and soldering practices that I changed the design of the LED to include ESD protection and a dozen recommendations to improve their volume manufacturing process. Then their 0.5% failure rate dropped to near zero.

Although the current may be low, the reverse voltage may exceed -5 V and partially wound the junction.

  • 26
    \$\begingroup\$ any particular reason to link a wikipedia article about new zealand???? \$\endgroup\$ – ΦXocę 웃 Пepeúpa ツ Jul 9 at 14:33
  • 9
    \$\begingroup\$ Peter added it in his fine edit. next time I go to Slovakia, we can discuss over a beer \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Jul 9 at 14:34

It is probably a small electric current from the tip of the soldering iron. Some irons have a grounded tip, others may not. In both cases, it is possible to have some voltage at the tip.

You can try to test for this with a multimeter set to AC volts from the metal body of the soldering iron to the LED, but a meter might not have enough sensitivity at about 10kΩ impedance. A better tool would be an oscilloscope with 10MΩ impedance. The iron may need to be "on" for accurate results.

For some reason, the electrical outlet in my old kitchen had about 69vAC on the ground, but only at a very tiny current, perhaps 20µA, which a meter would barely detect. This was tiny, but it was enough to "light" LED's for no apparent reason when touched with the iron. Note this made it unsafe to solder more sensitive components, such as digital IC's.

I had a suspicion this was related to an old refrigerator. It had a propensity for shocking you slightly if bumped with a wet arm. Unplugged it and sure enough, the leaked voltage at the nearby outlet disappeared.

  • 1
    \$\begingroup\$ It should be nedless to say but you might get away with soldering LEDs with a potential difference and small current but if you are soldering MOSFETs it might destroy them \$\endgroup\$ – Dirk Bruere Jul 8 at 13:42
  • 4
    \$\begingroup\$ How old is your meter? Isn't the input impedance of modern meters also in the range o 1 Meg Ohm? \$\endgroup\$ – jusaca Jul 8 at 16:15
  • 1
    \$\begingroup\$ Well I have an old Simpson 260 multimeter at a whopping 10kΩ/v impedance. Don't use it much anymore, but if I have one, somebody else is bound to also. :) \$\endgroup\$ – rdtsc Jul 8 at 18:40
  • 1
    \$\begingroup\$ Was the ground wire in your old kitchen disconnected from the actual ground? \$\endgroup\$ – user253751 Jul 8 at 22:50
  • 3
    \$\begingroup\$ FWIW, this electical setup behaviour usually means that there's a severe problem that should've been fixed by a dedicated electrician - most likely, if you'd have a ground fault interrupter, your breaker would go wild most of the time. As @immibis stated, it can be caused by a ground wire broken somewhere, or by an invalid chassis earthing scheme + a faulty device which has some current leaking through the chassis. In both cases, IMVHO, it's against electrical safety protocol to operate mains voltage devices under the conditions you described. \$\endgroup\$ – vaxquis Jul 9 at 10:09

This is probably from stray current in the tip. Another effect that is possible could be from the solder junction itself from thermal EMF effects. This would not happen with normal solder as normal solders usually have 3 µV/°C - 4 µV/°C of thermal EMF. If the solder was contaminated with a different metal or oxide then it could go up well beyond the voltage needed to power an LED. At 300 °C that would amount to 1.2 mV, not enough to power an LED, however...

Some metal combinations (like copper to copper oxide) have 1000 µV/°C of thermal EMF, which would provide enough voltage to light the LED (but who knows what the source impedance is). If you had the right metal to metal junction, this could explain this effect.

An interesting thing is pilot lights on gas devices use thermal EMF to generate enough power to keep a solenoid open. If the gas light goes out, then the temperature drops on the thermocouple, it can't provide enough current for the solenoid and the solenoid closes for safety reasons.

Either way, if you wanted to find out which it was, get a voltmeter and connect it between the tip of the soldering iron and ground. If there is voltage on the tip without soldering, then it could be stray current from the soldering iron that is lighting the LED.

  • \$\begingroup\$ You need to consider the peak reverse voltage not the DMM Vac voltage. -1 \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Jul 8 at 16:24
  • \$\begingroup\$ What do you mean? Usually whenever I test with a DMM for current or voltage, I test for DC and AC. The voltage needs to clear 2-3ish volts \$\endgroup\$ – Voltage Spike Jul 8 at 16:54
  • \$\begingroup\$ Since the diode impedance is asymmetric, so is the voltage. With a mutually conducted current. Thus Vr peak voltage is the criteria. I have seen LEDs wounded by this process and then it is evident by a rising junction capacitance in a Lissajeu AC test (V vs I )and is the early symptom to infant mortality.!!! \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Jul 8 at 16:58
  • \$\begingroup\$ I see your point, it would be better to put the meter to ground. \$\endgroup\$ – Voltage Spike Jul 8 at 17:02
  • 1
    \$\begingroup\$ @Catalyst en.wikipedia.org/wiki/Thermocouple#Gas_appliance_safety mentions thermocouples/thermopiles directly driving gas valves, for a start \$\endgroup\$ – nekomatic Jul 9 at 13:07

Don't fool around with this! Test it.

Seek out a grounded, GFCI-protected outlet (kitchen, bathroom) or RCD-protected outlet, preferably a 6 or 8 milliamp rated one.

Remove the cover plate from the outlet*, so you have access to the yoke (metal wings of the receptacle which take the mounting screws). This yoke should be tied to the Equipment Safety Ground,** which bypasses the GFCI. We need access to that.

Now, plug the soldering iron in, and touch it to the yoke repeatedly as it warms up. Don't hold it there, or you could melt the receptacle or plastic junction box.

When that's done, if it's possible to flip the plug over (looking at you Germany), flip the plug over and test again.

If it trips the GFCI, into the trash it goes

If the soldering iron is leaking current to the tip, it will trip the GFCI in about 10 milliseconds. That is an indication to throw it out, because it is internally leaking mains current onto the tip - which means it will leak mains current onto you. 10ma can kill. It can also stun you, and unconscious person + soldering iron is a bad combination.

If it is an expensive unit, you can maybe do a teardown to see if you can find dirt, splashed solder or frayed insulation that might be causing this.

What is the test testing? The GFCI detects by comparing current flow on hot and on neutral, to assure they are the same. If they are not, some current is taking a third path. So it detects all current flow that is not hot->neutral. We are using ground to create a third path for testing. We flip the plug over to exchange hot and neutral, because neutral->ground is harder to detect than hot->ground.

* As a low voltage DC electronics person, if you normally have a healthy fear of all things mains power, that is a good thing. And if you don't have that fear, obtain it. LVDC does nothing to prepare you for the hazards, rule-set and mind-set of mains power. But you can certainly do this.

** No relation to electronics GND/VSS/"common", that would be neutral. Equipment Safety Ground is a shield, but does not contact any mains conductor in any way (except for exactly one equipotential bond). As such, it entirely bypasses the GFCI - really, GFCIs don't use ground (except GFCI receptacles, to pass it through to appliances).


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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