# Sizing LED Resistors for Multiplexed LED Displays

So let's say I have an LED matrix with a 1:8 duty cycle. The matrix consists of these LEDs. These are all the specs the supplier has given me:

Emitting color: Red
Diameter: 5mm
Lens color: Water Clear
Forward voltage(V): 2.0-2.2
Current(mA): 20
View angle: 25
Luminous intensity(MCD): 8,000 - 12,000

I run it at 5V. By these specs I am not even sure if that is allowed but it seems that all red LEDs like 5V.

So for a single LED the resistor value should be :

E = IR
(5-2)V = .02A * R
3 = .02A*R
3/.02 = R
150 Ohm

Right?

Now, if this is a multiplexed display, since each diode is getting current only 1/8th of the time, can I increase brightness by overdriving the LEDs with smaller value resistors? I've heard some discussion that this is allowable, but I don't know how to calculate it. Any ideas? These are cheap diodes and I don't mind "taking a chance" here. Might I try 100 Ohm to see if it helps with brightness? Or can I be even more aggressive? I can always put a few of these on a breadboard and let them burn in overnight to see how well they handle it with a 100% duty cycle.

Well, I tried a few values, 150, 100, and 47 so far to see by my eye how bright the LED looks. I can't really tell looking directly at them, they are all so bright so I use the desk magnifier to project them on the ceiling. The LED is noticeably brighter at 100 ohm (30ma) than 150 ohm (20ma), which is to be expected. Surprisingly, to me, it looks no brighter at 47 ohm (~60ma) than at 100 ohm. If anything, it isn't quite as bright. Is this because of overloading? I've let 150 and 100 run for about 90 minutes and neither have burned out. The 47 ohm experiment was terminated because it didn't appear to have any benefit and the quarter watt resistor was overheating (hot to the touch) anyway. The 100 ohm and 150 ohm quarter watt resistors seem to be having no overheating issues.

Typically you can find a spec like this for the LEDs:
Peak Forward Current (1/10th duty cycle, 0.1ms pulse width) IFM 50 mA

You should look up
Manufacturer Ligitek
MPN LED 5mm Red Water Clear
SKU A-706
and see what they say about yours.

Typically you can find a spec like this for the LEDs:
Peak Forward Current (1/10th duty cycle, 0.1ms pulse width) IFM 50 mA
5mm Red LED - 630 nm - T1 3/4 LED with 30 Degree Viewing Angle | Super Bright LEDs

You should look up
Manufacturer Ligitek
MPN LED 5mm Red Water Clear
SKU A-706
and see what they say about yours.

That's the big problem, that is the reseller ID and not the manufacturer code. Tayda are not nearly as good at providing information and a datasheet as Digikey. Hell, they aren't even as good as Futurlec.

When I look at Ligtek's site and check out the 38 pages of 5mm LEDs that they manufacture, I try to match those specs to the products on the list. First problem, there is no LED with a 2.0 o 2.2V forward voltage. So I have no idea which LED this is. Maybe they don't make it anymore and Tayda took the last container full for a song. Who knows.

Oh well. If this LED doesn't burn out by morning at 100% duty cycle I am using the 100 ohm resistor.

Life test - that's one way to go 8)

When conneting multiple leds to a single current limiting resistor there is the probabilty that the leds will have slightly different Vf. If that is the case the led with the lowest Vf will get a larger current. And led have a very sharp "knee" in their U/I curves.

Its better so ensure that each led has its own current limiting resistor. With multiplexed leds you can do that by writing your code so that only one led is lit at the time. Because each led will be lit only a very short time you should set the current to a high value to get enough intensity.

nilton61:
When conneting multiple leds to a single current limiting resistor there is the probabilty that the leds will have slightly different Vf. If that is the case the led with the lowest Vf will get a larger current. And led have a very sharp "knee" in their U/I curves.

Its better so ensure that each led has its own current limiting resistor. With multiplexed leds you can do that by writing your code so that only one led is lit at the time. Because each led will be lit only a very short time you should set the current to a high value to get enough intensity.

This is a multiplexed LED matrix and there are no LEDs in series or in parallel. Specifically, the way it is set up is 8 cathodes and 64 anodes, 512 LEDs in total. Only one cathode is grounded at a given time for a 1:8 duty cycle. Current limiting resistors will be on the anodes, once I decide on the value I want. So far, I intend to use 8 TIP120s with base resistors of around 2K for the cathodes and 8 UDN2981s (octal high-side driver) for the anodes.

You going for a cube eventually, hence all the drivers?
64 LEDs is 1.28A @ 20mA each.
TIP120 is a darlington arrangement part, has a high Vce. (up to 4V).
N-channel MOSFET would be better, low Rds & low heat dissipation.

You have a source for UDN2981? They seem to be getting scarce.
Probably have shift registers to drive them?Why not get shift registers that can source some decent current - CD74AC164

make sure its AC164, not HC

You going for a cube eventually, hence all the drivers?
64 LEDs is 1.28A @ 20mA each.
TIP120 is a darlington arrangement part, has a high Vce. (up to 4V).
N-channel MOSFET would be better, low Rds & low heat dissipation.

You have a source for UDN2981? They seem to be getting scarce.
Probably have shift registers to drive them?Why not get shift registers that can source some decent current - CD74AC164
http://www.digikey.com/product-detail/en/CD74AC164E/296-4235-5-ND/375630
make sure its AC164, not HC

Not a cube, the 2D matrix I made a while back. It is 16x32 but it is wired into a 8x64 configuration. I choose to use 64 drivers instead of shift registers because I want to be able to PWM this matrix. I am going to attempt to drive it with a CPLD and if the CPLD doesn't cut the mustard then maybe with an FPGA. The microcontroller will be there for the brains, the CPLD is the pre-driver to the real driver circuitry that I described previously. Anyway, that is the plan.

My UDN2981s were purchased about 6-9 months ago, in total I bought 25. I do see some decent prices on eBay if you don't mind throwing the dice with a Chinese seller. I have used G&C before for MAX7219s and got a really good price and they all work so far, probably have used 30 of them. They have smaller lots at slightly higher prices on the UDN2981s:

http://www.ebay.com/itm/50PCS-UDN2981A-UDN2981-IC-SOURCE-DRIVER-8CHAN-18-DIP-NEW-GOOD-QUALITY-D6-/300893204126

I posted this some time ago and you commented on it:

Finally got off my butt and did the wiring, what a pain: