7414 Failure

September 18, 2010 10:59:44 PM CDT

This is a failed prototype for a utility board.

I'm a big an of incremental testing during circuit assembly: solder in a component, run a quick test to make sure everything works, then move on to the next part.

First of all, it's satisfying. You get to see the circuit do something every couple of minutes. Second, it's much easier to find and fix problems as you add each component than it is to build an entire circuit, then try to figure out why it doesn't work.

One of the easiest tests for a circuit (especially the digital ones) is to light an LED. Unfortunately, an LED wants at least a couple milliamps, and some parts of a circuit don't have that much current to spare. You're better off connecting something with high impedance -- like the gate of a mosfet -- to the node you want to test, then driving the LED from that.

That's what this board tries (unsuccessfully) to do.

The chip is a 7414 hex Schmitt inverter, and the LEDs are wired to the positive rail. When the input of any given inverter goes high, its output goes low. That gives the LED a path to ground, and there is light.

The board works perfectly in that respect.

It just doesn't do what I want it to.

The problem, as I'm sure some of you spotted immediately, is the floating inputs. The board shown doesn't provide any default input, so the inverters use whatever input they can find.. parasitic currents within the chip, skin contact across a couple of connections when you touch the board, stray radio waves treating the traces like tiny antennas, etc.

That means the default state for any LED not connected to a circuit is "random, changing at random intervals."

Not what I wanted.

Fortunately, the problem is easy to fix.