# LED lamp loses intensity very fast

I have designed a UV lamp using UV LEDs (VAOL-5EUV8T4-ND), and it loses intensity very fast. I attach a figure with a picture of the lamp and the circuit. I am using a LM317 to generate a constant current for the LEDs. With the resistor values shown in the figure, the LM317 should generate a constant current of 0.160 A, which is split among the eight columns of LEDs. The LEDs data sheet reports a forward current of 20 mA, which what the circuit provides. The voltage drop across one LED is 3.2 V (measured with voltmeter). A voltmeter reader of the drop of potential across R2 and R3 gives 1.273 V as expected. I use the resistor R1 to reduce the voltage drop across the LM317. Not shown in the circuit, I have also placed two fans to avoid heating and all the circuitry is very cold as a result. The capacitor is a 0.1 microF tantalum capacitor. The 6.8 Ohm and 1 Ohm resistors are rated at 1 W and the 27 Ohm resistor is rated at 3 W. The AC/DC power supply provides 48 V. The lamp consists of eight columns with 12 LED each.

I have measured the UV intensity with the UV sensor ML8511 connected to the Arduino. The inset shows the intensity decreasing over time. I use the internal voltage reference of the Arduino to get a precise measurement of the UV light intensity. The lamp is in a temperature-controlled room in a shelf covered with black curtains.

The intensity drop is even visible by eye, so I don't think that the light intensity measurement is the issue, but rather something in the lamp. The issue is permanent, meaning that if I switch off the lamp and switch it on after some time, the UV intensity is equal to the last intensity measured. Do you have any idea of what could be the problem? Thank you very much.

• There are a lot of assumptions and little measurements. It should be constant current. Have you done the measurement? Also, how is the current distributed among the LEDs? Without load sharing resistors, that chain with the lowest Vf will get the most current and eventually get damaged. What power source are you using, is it able to provide enough voltage? Is the regulator within its SOA all the time? Check some things, don't assume. – PlasmaHH Oct 18 '16 at 11:29
• "The intensity drop is even visible by eye" - UV isn't eye-visible. It isn't entirely eye-safe, either ... – pjc50 Oct 18 '16 at 11:38
• Paralleling multiple strings without a series resistance maybe the cause. It's always safer to use a small series resistor (say 5 Ohm) per parallel string even driving with constant current. – Rohat Kılıç Oct 18 '16 at 11:42
• Measure. The. Current. – winny Oct 18 '16 at 12:24
• Thank you, I will measure the current and be back. As you can see in the figure, the emission spectrum has a tail in the visible. – Andrea Oct 18 '16 at 13:24

This is probably to be expected, assuming these are 20mA LEDs.

Without a separate current sharing resistor in each string, each string will conduct at a slightly different voltage, due to natural variations between devices.

The one that starts conducting earliest will take most of the current, heat up, reduce its conduction voltage, and take more current, until it is operating at about 8x its rated power, and ultimately destroyed.

Repeat for each string in turn.

See ANY description of how to drive multiple strings of LEDs for the solution.

I would keep the regulator, and split the 27 ohm resistor : short out the current one, and put a separate (27*8 = 216) or 220 ohm resistor in series with each string.

This must be a duplicate of dozens of questions here.

• Thank you for your message. I have done what you suggested and the LED lamp is still loosing intensity linearly. Specifically, I have removed the resistance R1 (now the capacitor is in the correct place, thanks to Andy aka for pointing this out). I have placed 220 Ohm resistors in series with each string of LEDs (8 resistors in total). I have measured the currents in each string by measuring the voltage drop across the 220 Ohm resistors and there is only a 3% difference between the smallest and the largest current in the eight strings. The mean current is 20 mA as expected. – Andrea Oct 18 '16 at 20:58
• After measuring the light intensity continuously for one night, I can say that the rate of linear decrease in light intensity is lower than before (i.e., without the resistors in each string), but the intensity is still dropping linearly quite fast (20% in 24h). – Andrea Oct 19 '16 at 8:12
• Furthermore, after one night the voltage across the current-limiting resistors has not changed (within 1% from the value measured yesterday). – Andrea Oct 19 '16 at 8:24
• That suggests 2 possibilities. (1) The LEDs may have been damaged by overcurrent to the point where further decay is inevitable. You'll need an expert on LED chemistry to confirm this. (2) The LEDs were cheap from an off-brand manufacturer and this is just what they do. If you bought them from Digikey/Mouser/Farnell etc I would not expect this (might be worth contacting customer service) : if they were from a site called FortyThieves.com or something, it's a possibility. In both cases, the answer is probably a new set of LEDs from a reputable supplier. – Brian Drummond Oct 19 '16 at 9:40
• Thank you very much for your help and for your reply. I have bought them from Digikey (the LEDs are these digikey.com/product-detail/en/visual-communications-company-vcc/…), the manufacturer is VCC. I am now trying to keep them at a current below 20 mA and see if the intensity decreases in time. In the meantime I will build a brand new lamp. One thing that I cannot understand is that if I run one single LED at 20 mA (regulated with a LM317) it does not loose intensity at all. This might also suggest that the ones I'm using are damaged irreparably. – Andrea Oct 19 '16 at 9:47

You have the 27 ohms (R1) in the wrong place - it should be before the input decoupling capacitor - that capacitor is needed right at the input to the LM317 to avoid instabilities. Next, split that resistor into several resistors each in series with each string of LEDs. This change attempts to equalize the current in each parallel string. Use a 220 ohm resistor is my advice.

Putting LEDs in parallel is asking for trouble as Brian has pointed out.

Or get rid of the LM317 and just connect each string (3.2 volts x 12 = 38.4 volts) in series with a 480 ohm resistor (possibly 470 ohm might produce too much current) directly across the 48 volt supply.

(48 - 38.4) volts/ 480 ohms = 20 mA.

• Thank you, I have done what you suggested in the first part of your reply, as also suggested by Brian Drummond. Unfortunately, the lamp is still loosing intensity linearly in time, even though each string of LEDs has a current of 20 mA going through. I will now try to connect all the LEDs in series with a 480 Ohm resistor as you suggest and measure the light intensity versus time. – Andrea Oct 18 '16 at 21:09
• I had not understood properly, of course I cannot connect all of them in series. I had tried as you suggest (with the 480 Ohm resistor) in the earlier version of the lamp, but the light intensity was dropping in those conditions too. – Andrea Oct 19 '16 at 8:28
• Maybe the 48V supply is gradually lowering in voltage or starting to fail. Try measuring this voltage. – Andy aka Oct 19 '16 at 9:46