I'm replacing a load of blown backlight bulbs (incandescent) with LEDs.
Normally, I'd stick a resistor on each and every LED, but the board I'm fixing has loads of bulbs and no space for resistors, and they're already connected in parallel. The LEDs I'm using need 2V, but the only handy power is 5V.
I was going to use a linear voltage regulator to limit the voltage to 2V, but is there any reason I can't use a series of chunky diodes to drop the voltage to around 2V before it hits the mass of paralleled LEDs? (4 silicon diodes @ 0.7V drop each would leave 2.2 across the LEDs, well inside their max forward voltage of 2.8V)
Is there any reason normally why I can't use a diode to drop voltage before LEDs? Am I right in thinking Rs are chosen over diodes because a) they cost less and b) you can pick just the right R rather than having to daisychain diodes?
You need a Current source to drive a led, not Voltage.
Dynamic resistance of the diode or led is low, and small variation in its value would create a big variation in the current and brightness. This is why resistors in series with each individual led is common practices. Or you can use current source build around JFET.
Recognize that an LED is a diode. If you hook multiple LEDs up in series you will get a voltage drop.
If you have a 5VDC source and your LEDs work optimally at 2VDC, you can put two in series and handle the 1VDC drop with a resistor.
It works like this: LEDs drop a voltage across their terminals but at full voltage, have very little current limiting capacity. (They ARE Diodes) Therefor, if you do not limit the current on the diodes, they will allow as much current as they can to come through and will fry.
You need to know two primary things about your diodes
Nominal Voltage drop and Max current
I am going to assume you have LEDs with the following values for demo purposes.
2 vdc = Nominal Voltage Drop
25 mA = Max Current
We don't want to drive the LEDs at Max current so we will use 22 mA.
We have a 5 volt source. If we wire two LEDs in series at a drop of 2Vdc each, that gives us 1 volt.
Now in order to limit the current we will use the equation V=IR
V = 1
I = 22mA (Expected Current)
1/22mA = 45.5 Ohm resistor
Now, a 45.5 Ohm Resistor is hard to find. So lets redo the equation at Max current and then we know that we have to find a resistor slightly bigger than that value
1/25m = 40 Ohm
This is just one way to wire LEDs. As previously stated, you can drive them with a current source that is regulated OR you could wire them so that their voltage drop is forced into the linear range were they do limit current. (This is inefficient and they will not be bright)
You can also heat sync the lights and overdrive their current or use PWM control to force current limitations. These methods should be researched if you believe that you have a need for them.
The problem is that I'm retrofitting an existing PCB. The bulbs I'm replacing (12 of them) are already connected in parallel, and there's no room to get even the finest wirewrap / kynar wire in there to reroute. Not even room to squeeze in 0603 resistors. The most I can do is control the power that gets applied to the array.
The LEDs want around 12mA and drop 2v - so the whole lot'll take around 150mA. Current control is no good for paralleled LEDs though; if one dies, the rest will soon follow.
This is why I'm stuck with looking at voltage control rather than current control. I wonder whether an LED would survive 5V (no resistor) at 10-20% duty cycle PWM.
Voltage control wouldn't work for parallel led's due negative voltage-temperature curve.
It's known as temperature runaway, when one led take more current it would heat up, and pass even more current etc.
PWM also isn't a solution , led has max pulse current limits, even duty circle could be set to low value 10%, inside this window something should limit current anyway, so it wouldn't go to infinity.
The problem is that I'm retrofitting an existing PCB. The bulbs I'm replacing (12 of them) are already connected in parallel, and there's no room to get even the finest wirewrap / kynar wire in there to reroute.
If you hack at the PCB tracks you can break them and insert a resistor. If not then simply you can't do it. You can't fight the laws of physics and expect to win. Have you never seen Star Trek?
Thanks for all the help and advice, folks; looks like this is a no-go. Have to live without backlights on the panel, or look at a completely different approach like fibre-optics.
(It's a thin 4-layer PCB, no way to safely cut tracks without x-raying the board, no way to add resistors without machining some cavities into the lightguide that fits snugly onto the board surface)
Cheers for the swift responses, saved me from making a time-consuming mistake
Voltage control to multiple leds will never be stable and safe.