Pulling more than the max current from an IC in bursts?

I apologize if this is a begginner question or easily found on google (I tried, I swear). I'm a bit of a newbie to electronics and am in over my head in a project that I started.

I'm making a 34x40 LED display and multiplexing the LED's in a line-scan type setup. I'm using shift registers on the anode side (to source the LEDs) and some DMUXs on the cathode side.

The thing is, my LED's use 20ma for optimal brightness, but when I scan through the lines (each line is on for ~500uS), the LED's are too dim for my application.

Would it be possible to resize the resistor to allow for a higher current in bursts (I've calculated it to be around 600mA assuming a duty cycle of ~3%, but my math may be VERY wrong) without damaging my IC (rated to source 30mA max).

I really appreciate any advice and any tolerance to a beginning hobbyist who got in over his head. If there's any problem with this post or if you need clarification on something, just post and I'll do my best.

Thanks!

The scale up probably isn't linear. Because of the eyes persistence, a less bright pulse may do the trick (the eye will stretch the duration of the pulse giving it the same effect as a wider pulse time).

Also, check the spec -- you might find that there is a higher current spec (usually around 100ma at 1/10th duty cycle, 0.1ms pulse width). The 30ma rating is usually the maximum _Continuous_ Forward Current. At 500us, you're well within spec.

But, limiting the current with a resistor is unreliable, unless the supply voltage is high enough to swamp out LED forward voltage variation (the forward voltage between LEDs can vary significantly -- and can change over the life of the device). You might be able to get away with this is this is a one-off and you cherry pick the lights, but that still doesn't account for variations over time or even variations due to temperature (but that is only an issue of this will be used outside of room temp).

Much better to use a current regulator to drive the LEDs especially if driven to maximum specifications.

Hi, Sadly you will destrot the device, this is not a reliable design / solution for the problem. Not the way we would approach the problem if posed by any of our clients as we are asked to provide solid reliable solutions. Can I ask if this is just some project you are playing with or have you been tasked with solving / providing this design?

For logic type devices, the output currents are specified as the maximum guranteed while providing some guaranteed voltage level. So for example an output (sourcing) current of a device may be rated at 20mA at 3.0V and an input (sinking) current of a device may be rated at 50mA at 0.8V. Exceed those currents and the output voltage will get lower and higher respectively. Also for logic type devices, they can typically sink more current than they can source - you may be able to restrcuture your circuit topology to take advantage of that. For higher power devices you get into issues with bond wire current capacity and thermal dissapation from the IC. You may be able to parallel IC connections as CMOS devices share current well.

This is a project I'm working on for a gift. The end goal is to have the board display a game of Tetris, as a birthday gift to a friend.

Look at the device data sheet. It will usually have a section headed Absolute Maximum Ratings - it may specify maximum output current. If you exceed them you should expect the device to fail.

A lot of logic devices specify output short circuit current and may have notes in the data sheet about aximum duration and number of shorted outputs.

The only workable solution to your problem is a buffer/driver that can safely source the required current - provided that you don't exceed the LED's maximum current too. If you can't get the required brightness without exceeding maximum currents your design is defective.

One last comment. The ratings on current are discussed as a function of thermal loading of the LED. As a result, as long as the overall current in time does not exceed the thermal loading of the part, time slicing at a higher current should work. This was a trick I learned in College in Microcontroller classes as a freshman. Many of the kits did not have drive resistors and controlled the load of the LED by minimizing the amount of time that the LED was turned on in a full on state.

Last, I do recommend that you cautiously up the current till you get the desired result, but time slicing on an LED should work as long as the average power per second does not exceed your maximum and the scanning rate is high enough, 10hz or higher(I would try to scan at `1Khz) to ensure that the LED is not left on too long.

Yeah, I've been doing this with great results (time tested, too). it seems to me, the limiting factors are:


1. Heat within the LED element itself.

2. Current carrying capacity of the small wires within the LED that connect the die to the leads.