What to do with unused inputs and outputs?

[sci4me]

I assume that unused inputs (of logic chips (I'm working with 74LS mainly right now)) can just be tied low (or high depending on the case...) but what about outputs?

[Yansi]

Pull the unused inputs up or tie ground (will draw more current, I think 2.4mA per input when grounded, if I remember that right). Outputs can be left unconnected.

[sci4me]

When using the term "pull", do you mean through a resistor? Or do you mean to just connect the input straight to Vcc?

[Yansi]

I think TTLs should be tied high only through resistor, but some time ago when I did a small and quick research about this, I haven't concluded anything definitive. If you want tie them high (spare some supply current), put ther eg. 4K7 to 1k resistor. It won't harm anything, just for sure.

I remember from a college being told that I mustn't connect the TTL device input directly do Vcc. The reason I was told was the input transistor is being revere biased that way. But if you look on the internal schematic of a TTL logic gate, I can't see any obvious reason why shouldn't I connect the input directly to V+. To breakdown the EB junction you need more than 5 to 7V or so, and that won't happen. So that much for a college 

Luckily these days we have CMOS gates with hig impedance inputs that can be tied directly to ground or V+ (sometimes even higher than V+ for some special types of logic like LVC gates). TTL is totaly OBSOLETE today, simply don't use it if you are not forced to do so.

[Hideki]

From a google groups post:

The definitive snobs of electronics (and I mean that in a good way!), Horowitz and Hill, authors of The Art Of Electronics, have this to say.  (I'll paraphrase).

If you want to force an unused TTL input low, it is perfectly okay to ground it.  If you want to force it high, though, it's slightly safer (for the well being of the device) to use a pull-up resistor.  The reason is that the power supply pins can tolerate more overvoltage than the input pins can, so just in case there's a glitch in the power supply and the voltage increases briefly (say to 5.5V), you're covered. They recommend that you tie all unused inputs that are to be forced high to each other, and then pull them all up with one resistor.

For CMOS, there's no such danger.  You can freely tie any input to either power supply rail.  And of course no one needs to be reminded that it's very bad to leave a CMOS input unconnected.

I also agree with Yansi. Stay away from TTL if you don't absolutely need to use it.

[sci4me]

This will sound dumb is this the right way to do it:

[0xdeadbeef]

There are typically three options for unused pins - which is preferred depends one use case.

1) Configure as input and pull to ground
2) Configure as output and switch to ground
3) Configure as input and activate internal pull device

At work, we tend to use #1 primarily for ADC inputs, #2 for most NC pins and #3 for pins which are be used in different placement options or as configuration pins.

[Hideki]

"you tie all unused inputs that are to be forced high to each other, and then pull them all up with one resistor"

There is nothing wrong with using one resistor per input if you have too many resistors and want to use them up.

0xdeadbeef: This question is not about microcontrollers

[0xdeadbeef]

Oops. Should have read the whole posting

[sci4me]

Oops. Should have read the whole posting

Heh no biggie. Thanks guys!

[slateraptor]

I also agree with Yansi. Stay away from TTL if you don't absolutely need to use it.

I don't agree...especially of the end-user is new to electronics.

CMOS devices:
1.) ...have appreciably faster rise times, which can be real problem in terms of signal integrity for new players.
2.) ...are a lot more susceptible to bricking from ESD, especially in dry climates and without proper tools.

This will sound dumb is this the right way to do it:

Yes.

[TimFox]

Modern design practice for automatic testing prefers that all unused inputs be connected to ground or high through a resistor, so that board probing can verify full function of all devices (including unused sections).  With TTL, the pull-down resistor to ground must be small (check VIL and IIL specs for logic family).

[tszaboo]

Just connect 13 to ground, and connect 11-12 together. Why would you waste board space and components for a such mundane part of the circuit?

[Zero999]

Although it's not best practise, it's perfectly fine to leave unused TTL inputs floating (disconnected) as they're already pulled high internally. The trouble is it makes it more susceptible to noise pick up but this isn't normally a problem, unless the input is connected to a long piece of wire/PCB trace and is near a source of noise.

[wraper]

CMOS devices:
1.) ...have appreciably faster rise times, which can be real problem in terms of signal integrity for new players.

Not true. TTL (S, AS, LS, ALS, F...) as well as CMOS come in different flavors. Just select the series with appropriate speed. CD4000B is much slower than 74LS, 74HC approximately the same as 74LS, 74VHC very fast.

[retrolefty]

  And of course no one needs to be reminded that it's very bad to leave a CMOS input unconnected.

 I learned this the hard way when I first used a CMOS device (4000 series I believe) back in the early 80s. I was only using two of the four gates of the device and when I touched the device with my thumb it caused a blister burn, from oscillation resulting from un-terminated inputs. Took a long time for me to trust those early CMOS devices as I had so little problems with TTL logic chips.

[vk6zgo]

  And of course no one needs to be reminded that it's very bad to leave a CMOS input unconnected.

 I learned this the hard way when I first used a CMOS device (4000 series I believe) back in the early 80s. I was only using two of the four gates of the device and when I touched the device with my thumb it caused a blister burn, from oscillation resulting from un-terminated inputs. Took a long time for me to trust those early CMOS devices as I had so little problems with TTL logic chips.

I had something similar with 4066 quad analog switches back around the same time period.
I left the unused inputs & outputs disconnected & the things got red hot.
After connecting all the unused inputs to zero volts they were a bit cooler but still hot.

A 6.8v negative supply was jury rigged--------long way from Electronics shops,& 10 PM ,so used whatever zeners were in the Station store.
I connected that to the unused inputs & the rotten things were at last turned off!

This was an official project,so I was "under the hammer" to get it finished.
It had "bugs" I still hadn't ironed out,but the application it was designed for became obsolete,so it was never used,in any case!.