Supposing I want to build a circuit where a logic gate will light an LED. Naturally I'm interested in how much current I can get out of the gate to light the LED.

Looking at the datasheet for the CD4017:
http://www.ti.com/lit/ds/snos357a/snos357a.pdf

On page 2 it specifies "High level output current" as -3.5mA typical, under the conditions:
Vdd = 15V, Vo = 13.5V, 25 degrees temp.

Here's what I am wondering:

• Why is this figure negative? Surely a positive output voltage should give a positive output current?
• Is this the maximum permissible limit for the chip, or is this simply what it would deliver into a short circuit? In other words, could I run the LED with no limiting resistor, relying only on the IC's own current limitations?

 

The specs are describing how far the chip will miss the supply voltage under such and such a load, and this is a pretty feeble chip. You better plan on using a transistor to boost the output current.

 

Logic chip typically follow a convention that all currents are defined as positive when flowing into the device.

By this convention, a positive voltage on an output is sourcing a negative current.

It's just a sign convention.

 

Supposing I want to build a circuit where a logic gate will light an LED. Naturally I'm interested in how much current I can get out of the gate to light the LED.

Looking at the datasheet for the CD4017:
http://www.ti.com/lit/ds/snos357a/snos357a.pdf

On page 2 it specifies "High level output current" as -3.5mA typical, under the conditions:
Vdd = 15V, Vo = 13.5V, 25 degrees temp.

Here's what I am wondering:

• Why is this figure negative? Surely a positive output voltage should give a positive output current?
• Is this the maximum permissible limit for the chip, or is this simply what it would deliver into a short circuit? In other words, could I run the LED with no limiting resistor, relying only on the IC's own current limitations?

As ErnieM already noted, ICs are traditionally defined such that current into the pin is positive. This is so that simulation tools have a consistent framework to work with, especially since sometime the same pin might be an output and other times it might be an input. Personally, I would have opted for the other way around, but I can see where they were coming from when considering it from a black box perspective and noting that current will come into the chip from the power supply, hence let's define that current to be positive. It also let's you calculate total power dissipation in a simple and intuitive way.

Notice that your question about it being the max current that can be delivered into a short circuit has a problem in that the conditions you cite specify that the output voltage is 13.5V, not 0V as it would have to be if the output were shorted to ground.

Remember that the purpose of a logic chip is to pass along information abouit logic states, which for most families is represented by a voltage. To do this, they have to sink and source some current. What we are concerned about is how much current can they source are sink while maintaining a voltage level on the output that will be recognized correctly by other logic chips.

Most logic chips are capable of sinking more current than they can source (and in some cases it is considerably more). So you may be able to drive an LED directly with a logic LO output but have more problems with a logic HI output.

Also, be aware that in addition to per-output current limits, there are per-chip current limits as well. If you are driving multiple LEDs from the same chip, this can be a factor.

 

Thanks for the answers. Fortunately I only need about 1.5mA for these LEDs, and everything seems to be working OK.

Logically, anything labelled "output current" should be positive when coming out! Hence my curiosity.

 

Thanks for the answers. Fortunately I only need about 1.5mA for these LEDs, and everything seems to be working OK.

Logically, anything labelled "output current" should be positive when coming out! Hence my curiosity.

The choice is basically a consequence of using the passive power sign convention. But it is definitely counterintuitive because we tend to think of the inputs and outputs as isolated entities instead of thinking of them as part of a larger abstract block in a yet larger abstract system.

 

I only need about 1.5mA for these LEDs.

That kind of makes my answer wrong.
Sorry about that.

 

Nah you were right it's a "feeble" chip.

Modern spec CMOS chips are better like HC4017 or AC4017 and some other types (he needs to check a CMOS IC reference book, or individual datasheets) but 20mA sink and 20mA source is commonly available.