I have an adjustable 24VDC power supply. I have 1W LEDs that I was planning on stringing 7 or 8 in series to get approximately 24V across all of them together. They are rated 3.3-3.8V each. Do I need a resistor in series to maintain current? I was thinking so since a lit LED is basically a short? I can't find an answer to what it is really needed for, just that most people say to put a resistor in series to 'limit current'... if so, it would be a small resistor, around 2 or 3 ohm?
-
\$\begingroup\$ What range is your supply? Amps and Vmin:max Where is LED datasheet link? \$\endgroup\$– D.A.S.Commented Feb 22, 2019 at 18:07
-
\$\begingroup\$ 18-29.5V / 5A don't have a real datasheet on them, they are cheap knockoffs here are specs: 1W, 3.2-3.6V, 350mA, 100-110Lumen \$\endgroup\$– ChrisMan183Commented Feb 22, 2019 at 18:46
-
\$\begingroup\$ Ok then you can run 8S LEDs and try for 350mA if your heatsink is good enough or < 30'C/W \$\endgroup\$– D.A.S.Commented Feb 22, 2019 at 18:47
3 Answers
This range and your heatsink 'C/W controls the k= dV/dT for NTC coefficient on LEDs Vf and T ['C] so choose a low side R around 0.1 to 0.5 Ohm depending on heat sink k to avoid thermal runaway. Expect Vf to drop no more than 0.5V /24V at constant current from temp rise due to NTC effects.
A power change without a series R for a 0.2V change in V+ is ΔP=ΔV²/Rs where ΔV =(3.4{=Vf.nom}-2.8{=Vt})/If= 0.6V/0.3A = 2 Ohms
(24² -23.8²/ (7 * 2 ohms) = 0.71W rise in 7 LED stringfrom 1W to 1.1W using 24V. If LED heatsink is 50'C/W this raises junction 5'C and reduces Vf ~ -4mV/deg C or -20mV so this assumption is thermally stable.
I suggest you use a 0.1V current sense shunt on low side or 0.1V/0.3A= 1/3 Ω and adjust ΔV=0.1V for 300 mA and adjust while hot. You can now measure your thermal resistance by the change in ΔV from the datasheet thermal coefficient ( -3.4 to -4 mv/'C to ? ) as the string reaches steady state above ambient temp. A good design is case temp below 45'C meaning never too hot to burn finger. But using this method you can measure actual voltage change in Vf. This Vf is the actual change in threshold "knee" voltage Vt not the Rs bulk resistance.
0.3A* 0.1V = 30 mW so you can afford to use 8 LEDs at 3.4V = 27.2V with a 0.5W drop current sense at 0.35A from 29.5V max or R=Pd/I²= 0.5W/0.35²= 4 Ohms 1W with 0.35A*4= 1.4V drop results in 27.2V nom + 1.4= 28.6V
So if your LED tolerance is on the high side , you have to bypass 1 LED down to 7 and still regulate V+ for current.
I have an empirical formula to determine the heatsink required for CV control to prevent any string from thermal runaway due to NTC Vt effects from temp rise. ( but not included here) But a long string is less prone to thermal runaway since internal bulk resistance stabilizes the Vf voltage as mainly the threshold internal Vt voltage is affected by temp, not the bulk electrode resitance determined by LED power rating and size..
LEDs have a negative temperature coefficient, meaning that as they heat up, their resistance decreases, causing more current to flow and more heating.
Either a series resistor or current controlled driver is necessary unless the thermal dissipation of the LEDs dwarfs their potential output. Using a series resistor is less efficient than current control as you must use a resistor sufficient that the positive temperature coefficient(PTC) of the resistor counteracts the negative temperature coefficient of the LEDs, preventing thermal runaway. If you're using 7 or 8 LEDs to approximate 24V, you have roughly 3.2V 350mA LEDs(edit your question and add the specs/datasheet), and 350mA is at the top end of what you'd want to run through a current limiting resistor. If you do, you'll want to keep the voltage drop over the resistor small(while still having the temperature coefficients cancel) and use a fairly high wattage one (you can actually use some types of resistor as a fuse as well). You'll likely waste a lot of power regardless.
Instead I'd recommend using current controllers for LEDs of this wattage at full brightness.
On the other hand, if you underpower the LEDs, using a resistor becomes less of an issue and you'll increase the efficiency and lifespan of the LEDs, although in this case you'd still be better with current control.
-
\$\begingroup\$ Also add a PTC has to operate > 85'C and consume some power to regulate. \$\endgroup\$– D.A.S.Commented Feb 22, 2019 at 18:08
A pre-fabricated LED string may already has resistors integrated into it already. If you need a resistor don't guess. Calculate it out, though you will need to know the number of LEDs, the minimum forward voltage drop, and desired operating current of the LEDs.
I'm sure the answer is somewhere else on this site already: led resistor calculation, and resistor wattage?
But basically, the LEDs have a pretty fixed voltage drop across them. The balance of the voltage from the supply will appear across the resistor and then you just use V=IR to determine the current flowing through this resistor, and since everything is in series, this will also be the current flowing through the LED.
\$ I = (V_{supply} - (V_{LED} \times N_{LED})) / R \$
I recommend you use the minimum LED forward voltage since this scenario will cause the most voltage to be dropped across the resistor, and therefore result in the greatest possible current for the resistor value chosen. Doing it any other way may result in an actual operating current higher than what you calculated due to LED component differences.