I think I know why but throw it out to the analogue experts.
I am powering a 3mm blue LED from an 24 V AC power supply with a drop resistor, circuit digram below. After a while, maybe 30 minutes, the LED blows.
According to datasheet, the max reverse bias is 5v, anything above reduces the lifetime of the LED.
When the LED is reverse biassed, the voltage is shared across the doide, resistor and LED with very little current flowing - to actually determine the voltage across the LED requires looking at the V-I graph for the diode and LED. I expect that when the current is determined, the voltage across the LED exceeds its maximum.
I am powering a 3mm blue LED from an 24 V AC power supply with a drop resistor, circuit digram below. After a while, maybe 30 minutes, the LED blows.
According to datasheet, the max reverse bias is 5v, anything above reduces the lifetime of the LED.
When the LED is reverse biassed, the voltage is shared across the doide, resistor and LED with very little current flowing - to actually determine the voltage across the LED requires looking at the V-I graph for the diode and LED. I expect that when the current is determined, the voltage across the LED exceeds its maximum.
Attachments
Blue Leds typically have a much lower saturation current than their silicon counterparts. This means that all of the leakage caused by the Si diode will appear as a reverse voltage across the LED.
Remedy is simple: instead of using the protection diode in series, place it in antiparallel. If you mind about power dissipation, keep the series diode, but add a an antiparallel diode (it can be a 1N4148).
You could replace it with a resistor, in the 10K range if you are after the last micro-cent, but I don't think it makes sense nowadays: a generic diode is not going to be more expensive than a resistor, and it even might be cheaper.
Don't underestimate blue and white LEds: they are cheap, produced by billions everyday, but they are very high-tech, and don't tolerate abuse, unlike grandpa's semi's
Remedy is simple: instead of using the protection diode in series, place it in antiparallel. If you mind about power dissipation, keep the series diode, but add a an antiparallel diode (it can be a 1N4148).
You could replace it with a resistor, in the 10K range if you are after the last micro-cent, but I don't think it makes sense nowadays: a generic diode is not going to be more expensive than a resistor, and it even might be cheaper.
Don't underestimate blue and white LEds: they are cheap, produced by billions everyday, but they are very high-tech, and don't tolerate abuse, unlike grandpa's semi's
Fit either a series diode or reverse biased diode to stop the reverse polarity from killing the led.
What is the maximum forward current of the LED?
Fascinating that it blows despite the series diode. I would expect that even if the LED goes in avalanche during the negative peaks, it would survive that for a very long time because the series diode very much limits the avalanching current. Apparently that's not true.
@Mark Tillotson seems to know more about device physics than anyone else around here, I wonder what his opinion is.
Fascinating that it blows despite the series diode. I would expect that even if the LED goes in avalanche during the negative peaks, it would survive that for a very long time because the series diode very much limits the avalanching current. Apparently that's not true.
@Mark Tillotson seems to know more about device physics than anyone else around here, I wonder what his opinion is.
Marcel, avalanche process in blue LEDs is destructive, even at minute current levels. It has to be completely ruled out for reliability.
Blue LEDs are very smart devices, but the exquisite delicacy of their structure (even the 1 cent types from China) means they are much more vulnerable to abuse than "ordinary" semi's as we know them.
This extends to other colors made from blue: mostly white, but also supergreen, etc.
Blue LEDs are very smart devices, but the exquisite delicacy of their structure (even the 1 cent types from China) means they are much more vulnerable to abuse than "ordinary" semi's as we know them.
This extends to other colors made from blue: mostly white, but also supergreen, etc.
Best & simplist fix for any LED powered on AC is the current limiting resister
& a common Si Diode in parallel with the LED to shunt the reverse half of the AC sine wave.
& a common Si Diode in parallel with the LED to shunt the reverse half of the AC sine wave.
a blue led glows bright at 0.4ma, so 24v/0.4mA is 62k resistor, parallel an 1n4007 anti phase diode for protection...
Are they very sensitive to ESD? A 2 kV human body model discharge could force 200 nC of charge through them in the wrong direction.
Then again, according to the OnSemi datasheet, the typical room temperature reverse leakage of a 1N4148 is about as high as 20 nA at 20 to 40 V, so that 200 nC is only equivalent to 10 seconds of leakage, 20 seconds of AC supply with a series 1N4148.
Last edited:
Indeed! I used 1M when running from 12V and it was still too bright in my application!
Yes they are: maybe not as sensitive as an unprotected MOSfet, because their capacitance is larger, but you can definitely damage them if you mishandle them (it happened to me more than once)
1N4007 on one leg, 47 to 150K on the other, in series should work fine.
1N4148 is not a good choice I think.
Check polarity of diode, reverse it if needed (crazy things happen).
1N4148 is not a good choice I think.
Check polarity of diode, reverse it if needed (crazy things happen).
Last edited:
Indeed, that is what surprised me, you live and learn, although to be fair, I have never used a LED in reverse bias.
I expect if I were to check the datasheets, then when both diodes are in reverse bias, the voltage across the LED would exceed it's allowed maximum.
Thanks for all replies - putting the diode in anti-parallel so when LED is reverse biased, the diode is forward thus keeping the voltage low across the LED👍. By coincidence I was looking at such a circuit albeit with a capacitor instead of resistor for running off mains; no need to look any further into that.
A bit last minute but placing a capacitor across your blue LED, perhaps just 1nF, in your original circuit would result in a, safer, net positive DC voltage across the diode at all times. Try one and scope it for reverse bias.
Fascinating issue. I would have thought just the 4k7 to be too low in value and the circuit text book standard and adequate. I would have added a 10 µF electrolytic cap in parallel to the LED to prevent "blinking" (also possibly preventing blow up of the LED at the same time). Thanks for new insights!
This while many use ordinary red, green and amber LEDs on AC without series diode....
This while many use ordinary red, green and amber LEDs on AC without series diode....
Last edited:
Older IR, red, amber, yellow and standard-green LEDs behave very much like any diode: they have an avalanche (or zener voltage?) and behave as such. They can also work as varicaps, and some were rather effective.
With newer generation, blue, super-green, white LEDs, the situation is different and ideally they should never see a reverse voltage, even if it causes a negligible reverse current.
Note (and I am sure you know that) that a capacitor directly across a LED is very ineffective at filtering out variations, because of the exponential I to V relationship and the attendant low dynamic resistance.
Just add a resistor between the cap and the LED (a few tens of ohm is sufficient and will only degrade slightly the efficiency) and you will have a much steadier light with a lower cap
With newer generation, blue, super-green, white LEDs, the situation is different and ideally they should never see a reverse voltage, even if it causes a negligible reverse current.
Note (and I am sure you know that) that a capacitor directly across a LED is very ineffective at filtering out variations, because of the exponential I to V relationship and the attendant low dynamic resistance.
Just add a resistor between the cap and the LED (a few tens of ohm is sufficient and will only degrade slightly the efficiency) and you will have a much steadier light with a lower cap
Yes I know but it reduces the flickering enough for my eyes. The cap and below 1 mA current apparently work as the blue ones I have used don't fail. However the purple ones I used all turned to emitting blue light after a while, same counts for pink ones. In fact I can not see any difference in color between those after a few weeks?!
BTW somewhere in time blue LEDs seem to have changed as the recent ones are extremely bright at very low currents. The blue ones emit ultra irritating light so best avoided at all 🙂 Never bothered to find out why but blue light does something with dimensional sight/view (on some eyes at least). Astonishing but some cheap Chinese devices manage to lighten a room with a single blue LED.
All things considered feeding blue, white, purple, pink and super-green LEDs DC instead of AC seems the best solution 😀 Or just follow the "machine code" and use simple green standard LEDs for power on indication.
BTW somewhere in time blue LEDs seem to have changed as the recent ones are extremely bright at very low currents. The blue ones emit ultra irritating light so best avoided at all 🙂 Never bothered to find out why but blue light does something with dimensional sight/view (on some eyes at least). Astonishing but some cheap Chinese devices manage to lighten a room with a single blue LED.
All things considered feeding blue, white, purple, pink and super-green LEDs DC instead of AC seems the best solution 😀 Or just follow the "machine code" and use simple green standard LEDs for power on indication.
Last edited:
I've been running cheap Chinese blue LEDs (https://datasheet.lcsc.com/lcsc/2106061735_MEIHUA-MHL5013UBTD_C2764902.pdf) on 6.3VAC using a 5k1 dropping resistor for years with no burnt out LEDs... They DO burn out using 100k on 120V though (Of course they do! Max reverse voltage is only 5V!) so a series diode sorts that.
From the datasheet: "If reverse voltage is continuously applied to the LEDs, it may cause migration resulting in LED damage."
I usually run them on DC though because I don't like the flicker.
From the datasheet: "If reverse voltage is continuously applied to the LEDs, it may cause migration resulting in LED damage."
I usually run them on DC though because I don't like the flicker.
Last edited:
Not sure but point taken. Place the capacitor in parallel to the dropping resistor/led?