Driving 140x 3W LEDs

[Audioguru]

Without seeing the datasheet for the UV LEDs, I wonder if the 700mA is the absolute maximum allowed current if they have perfect cooling? How will you cool them? With a spray of Liquid Nitrogen? Then the thermal extremes will kill them soon.

I agree that the You Tube video is completely wrong. The hFE of a transistor that you buy can be 50 or 300 or anything in between and it changes with temperature and current changes. The circuit would probably burn out at least one of the series LEDs then they all would not produce any light.

 

[Xenobius]

Without seeing the datasheet for the UV LEDs, I wonder if the 700mA is the absolute maximum allowed current if they have perfect cooling? How will you cool them? With a spray of Liquid Nitrogen? Then the thermal extremes will kill them soon.

Yes I get 'excited' too when I think about cooling but really my plan is based on another similar project of someone who already did it using 1W LEds. In 'his' project he used 12 LEDs for a smaller size PCB (unlike mine which is A4) and for the full exposure to occur it took around 30 seconds. Based on this I am assuming I will need less, probably MUCH less then 30 seconds so heat-wise I am not concerned. However I will be doing the following:

I bought star shaped aluminum PCBs which you can solder the LED onto, and then you can solder wires to the aluminum in specific places where solder is able to stick. I will have long strips of aluminum heatsink which I will mount the star shaped boards using screws. In between the led and the aluminum star I will have thermal paste, and again I will have thermal paste between the star and the large aluminum heatsink. Considering a maximum ON time of 30 seconds I think this is more then enough.

Alternatively I can do without the screws and instead I can use a 2-part thermal paste which is specifically for the purpose. It hardens and sticks the materials together.

 

[c_mitra]

I bought star shaped aluminum PCBs which you can solder the LED onto...

I believe the LEDs have already some form of heat sink- there is a central pad on the reverse which is expected to be soldered to a board (in this case the PCB). With 70 LEDs dissipating 3W each (200W) on a A4 size PCB, you can just get away with a regular fan. You should distribute the LEDs for maximum uniformity as a short distance (more of less equally spaced or use a hexagonal arrangement). On the reverse side, keep a large pad to dissipate the heat.

I doubt you will be able to solder to Al plate (perhaps they were anodized and that will make soldering impossible). But you will be able to glue them surely.

But using the mounted LED will not solve your problem unless you mount them directly on a heatsink.

If you use 1W LEDs, you can get away using the same number and the light will be about 1/2 but the heat produced too will be reduced. The LEDs will run cooler and at higher efficiency. And you can run them easily ar 300mA without problem.

 

[Xenobius]

Ok so we are talking 20 of these circuits should do the trick. Assuming the 30VDC are clean, I should have 1.25v across R1, I should have any voltage between 3.2 and 3.6 on the leds and a constant current of slightly less than 700mA as set by the 1.8ohm. Any additional voltage left 'unused' will be dissipated by the LM317, ie: 30v from PSU - (7 leds x 3.2v lower range) = 7.6V - whatever the LM317 uses but lets not take that into consideration for the purposes of leaving some headspace. So on the LM317 there will be a maximum of 7.6v x .7A = 5.32W while on R1 there will be 1.25v x .7 = 0.9W


As for the power supply, 20 circuits @ 700mA = 14A so **broken link removed**

 

[Audioguru]

I assume that 700mA is the maximum allowed current for your LEDs.
Then what about the range of reference voltage of the LM317 (1.3V max) and the tolerance of the 1.8 ohm resistor (minus 5%= 1.71 ohms)?
Then the current in the LEDs will be 760mA.
You are planning too close to the limits.

 

[Xenobius]

I assume that 700mA is the maximum allowed current for your LEDs.
Then what about the range of reference voltage of the LM317 (1.3V max) and the tolerance of the 1.8 ohm resistor (minus 5%= 1.71 ohms)?
Then the current in the LEDs will be 760mA.
You are planning too close to the limits.

Yes 700mA is the max of the LED so I could do a 2R and that will take the current down to 0.625A with a 5% tolerance the range will be from 0.595A up to 0.657A - so thanks for the tip

 

[c_mitra]

Yes 700mA is the max of the LED so I could do a 2R and that will take the current down to 0.625A with a 5% tolerance the range will be from 0.595A up to 0.657A...

Whereas this is correct in principle, you should focus on the final objective. The LEDs should get the same current and provide uniform intensity. Therefore the individual LM317s should have experimentally determined values of resistors such that each section gets about the same current. In addition you should determine the currents and the resistor values at the operating condition (say at 60C rather than 25C). That can be a tedious process but for can be done manually for a once-off production.

 

[Xenobius]

Thanks for your reply. According to the LM317 the resistor determines the current given. So with this in mind, I'm thinking... if all the resistors are identical, shouldn't the LM317s (all 20 of them) output the same current? Also you bring a good point about the working conditions. Whatever the tolerance of the resistors, they should all be in sync (assuming all will be at the same temperature)

 

[Audioguru]

The datasheets for all ICs show their range of values. The "1.25V" reference of the LM317 is from 1.20V to 1.30V. Then a resistor that is exactly 2.0 ohms produces 600mA to 650mA. Plus a small change in temperature plus the tolerance of the resistor value.

 

[c_mitra]

if all the resistors are identical, shouldn't the LM317s (all 20 of them) output the same current...

Sadly, no.

Like Audioguru has pointed out the reference voltage can be off by +/-5% (although most will be pretty close). In the same way, resistor values are often off by +/-20% (although you can get 1% tolerance). Common resistors are often +/-10% tolerance.

The important thing to remember is that the LEDs become more vulnerable at higher temperature; the 700ma current rating is (usually; but I can be wrong) for the ambient temp of 25C and should be derated for higher temp (like most semiconductors).

It is a good habit to be conservative - particularly with devices working close to the absolute max ratings.