Using LEDs on Arduino (digital I/O) without resistor- the results

Actually I tried today (by the way I always use 2.2k or 4.7k for indication LEDs).

Here are the results. I tapped each LED carefully for a short time.

-old russian LED (green): a little more than 70 mA
-high efficiency 5mm yellow LED: nearly 80 mA
-white LED, 5mm: about 40 mA
-Blue LED, 5mm: about 50 mA
-red LED, 3mm, high efficiency: 78 mA
-red LED, 3mm, std brightness: 61 mA (this one actually showed abnormal low brightness)

All these currents are too high for LEDs.
They did not take damage neither the Arduino.
5 Volts supply was used.

It's not adviseable at all to use LEDs without resistors (Arduino digital I/O).

However, for multiplexed displays such currents are normal.

Also at lower voltage, you get much lower currents, especially for blue or white LEDs.

Indication LEDs however can work using a few mA and leaving out the resistor, will waste a lot of energy.

I can think of problems that arise when you use more than one LED at once without resistor (no multiplex).

It's actually not true neither the LED nor the Arduino will "fry" or "pop up in smoke" instantly.

However most people who use batteries are interested to extend the battery life as much as possible.

Someone maybe could try and measure the currents for a small multiplexed LED matrix, at lower voltage than 5 volts. Both for individual LEDs, and the current flowing into the Arduino.

All these currents are too high for LEDs.

I don't know what does the data sheet for the LED say.

They did not take damage

How do you know this. Did you measure the brightness before and after, and did you test the lifetime of the LED that had been subjected to these currents. The truth is that you have no idea if the LEDs were damaged or not.
In any event the currents quoted will damage an arduino if you connect it to them.

However, for multiplexed displays such currents are normal.

No they are not normal they are designed by idiots, there are a lot of idiots about so there might be a lot of them but they are not normal.

It's actually not true neither the LED nor the Arduino will "fry" or "pop up in smoke" instantly.

Nobody ever said this would be instant, it will be sen over the long term.

Someone maybe could try and measure the currents for a small multiplexed LED matrix,

I have done such tests at 5V.
http://www.thebox.myzen.co.uk/Tutorial/LEDs.html

Hello Grumpy_Mike.

I agree to most what you write. I only tapped the LEDs for a short moment. The brightness did not change. Most of these common 3mm/5mm at max. are designed for currents of 30mA.

I was thinking it would be useful to have some concrete numbers what will happen if you actually do connect LEDs directly.

Your research is also interesting I wonder about how 300 mA can develope. It's certainly worth reading.

I'd suggest you extend this web page, including some designs where actually LEDs are used without resistors. I have one small 5x5 matrix which runs weeks from 2x AA batteries so the currents can't be that high. It's a special case because it's at the margin of the LEDs threshold, but not neccessarily "bad design".

It's really lucky these 74HC series chips exist otherwise people would have to use expensive LED drive chips exclusively.

By the way I have spent years with various LED matrix circuits, at the beginning, I also did use resistors, and no multiplex, but this was way too much effort to wire.

What few people really are considering is the simple Ohm's law, and the Kirchhoff distribution rules.
If a LED is connected to a digital I/O, not the full voltage will drop accross the MCU internal MOSFET.
So the total power that is converted inside the chip is lower!

The datasheet absolute limits are for full swing over Vcc (I guess).

That especially becomes interesting, if the voltage is lower than 5 volts.
Let say, blue LEDs, and 3 volts. Only 0.5 volts will drop accross the chip internal structures!
It's a total different situation.

I agree it's wrong to connect a single LED at 5 volts directly to the digital I/O.

Also these specifications are valid over the full allowed range of temperature.
I'd say as long as the device keeps at room temperature, there also is some margin.

One rule that I follow is to monitor abnormal heat developement. If chips heat up just a little, that's normal. And of course, long-time testing, some months, if possible one year or more.

What few people really are considering is the simple Ohm's law, and the Kirchhoff distribution rules. If a LED is connected to a digital I/O, not the full voltage will drop accross the MCU internal MOSFET. So the total power that is converted inside the chip is lower!

No! Under normal conditions (with a current limiting resistor), the chip's output operates as a switch. This means there is either voltage across the internal circuit (with essentially zero current) or or current through the internal circuit (with essentially zero voltage). In a theoretically perfect switch, there is never current and voltage at the same time and no power is dissipated. This is why the light switch on your wall doesn't heat up.

When you hook-up an LED without a current limiting resistor, you get about 2V across the LED with the remaining 3V dropped inside the chip, at the same time that current is flowing. That generates heat INSIDE the chip.

Well all arguments that can be found about this topic are valid to some degree. Few of the arguments are totally wrong.

Let's talk about cars, most of them after a few years will need repairs due to normal wear-and-tear, and after some mileage, the whole thing will move towards a state "beyond repair".

There are many electric goods (gadgets) sold commercially which are rather bad design. They get too hot inside, and the buttons will fail after only a few years, as well the displays will fade.

So what are your problems? If it works a few years, it's OK.

And I'd say, lowering the voltages as we see it is happening for good reasons.

Some years ago I was "paranoid" too (a bit) that LEDs would burn out.
My experience is most circuits intend for low voltage have capabilities to withstand 5 volts for short time (when the flash memory is updated during developement).

I guess the 74HC internal structures also don't behave like an Ohms resistor, but I don't have the equipment to research that.

If you really see people using single LEDs without resistor, tell them 70mA or even 300mA will flow through them, and that is no good for the LED. They will experience LED detoriation after a very short time as well and see for themselves.

There are countless examples including some I saw myself for commercial designs, which can be "questionable". Except Bang&Olufsen "component graves" designing/manufacturing consumer goods seems to be some kind of "black art".

The brightness did not change.

How do you know? If you only looked at it then that is not good enough to make that statement. The brightness needs to be measured in order to estimate the life shortening that has occurred from over current.

There are countless examples including some I saw myself for commercial designs, which can be "questionable". Except Bang&Olufsen "component graves" designing/manufacturing consumer goods seems to be some kind of "black art".

Yes there are many designs that are engineered to a price rather than reliability but there are many many more that are not. A manufacturer is under the obligation to support a device for five years after he has ceased making it. Field returns are expensive epically late ones. I don't believe that nowadays any one designs a circuit to fail after a certain time.

Grumpy_Mike:

The brightness did not change.

How do you know? If you only looked at it then that is not good enough to make that statement. The brightness needs to be measured in order to estimate the life shortening that has occurred from over current.

  1. Because I know how it looks like, at first the color will detoriate, and then eventually the LED even will make "pop".
  2. Because 40mA to 70mA for 1/2 a second in my opinion won't really do much damage. These currents only start to do damage, when the LED has heated up already.

Grumpy_Mike:

There are countless examples including some I saw myself for commercial designs, which can be "questionable". Except Bang&Olufsen "component graves" designing/manufacturing consumer goods seems to be some kind of "black art".

Yes there are many designs that are engineered to a price rather than reliability but there are many many more that are not. A manufacturer is under the obligation to support a device for five years after he has ceased making it. Field returns are expensive epically late ones. I don't believe that nowadays any one designs a circuit to fail after a certain time.

Typically most PCBs using electrolytic capacitors will experience massive problems after about 20 years. Especially when the chips timing originally already is at the margin. The capacitors detoriate, and at some point of time, voltage levels won't be good anymore. Spurious errors will occur and soon, they will increase, at some point of time the whole PCB will fail. I did not make this up, I had one game console PCB, (which was new never used, but about 20 years old), it started up for a few minutes, but then errors occured, and after some hours, it did not start up at all anymore. And (cheap) electrolytics all over the PCB.

One Mitsubishi made VCR from 1986 actually had very good capacitors, still work like new.

Definitively there are special kinds of pushbuttons, which are specified for only 10,000 "push" operations. Where normally it's at least 100,000 times.

The designs are often bad, not neccessarily on purpose, but having totally superfluous components, as well too much heat developement. I don't know if they simply don't care, or if this is normal for some kind of goods. They all work fine but after some years, you can see first signs of problems developing.

LED circuits aren't so popular now for consumer goods except some smaller displays, but LED calculators were common some 2 decades ago. I never really saw resistor networks inside them, so they must have adapted the controller chip, somehow.

Maybe, Grumpy_Mike, you don't understand my point of talk.
It's that eventually, not many people really will use LEDs without resistors.
It is of course, bad. But I tried to explain, there are many examples of questionable design, which can be seen inside commercial goods. Not only the cheapest ones.

Bang&Olufsen for instance is a brand where you would expect, even after 20 years, that it still can be used, without problems, and that repairs are possible. You'd see it in the price tag...

My thought is that all this bad design really can't be stopped. Some hobby users actually never will design anything that is sold, but it seems to be normal (some kind of black art) to rely on questionable design.

manufacturer is under the obligation to support a device for five years after he has ceased making it

At the customers' expense. I don't think you mean warranty. That's much shorter.