Is it safe to connect inputs to VCC directly

[Siwastaja]

Again, please explain how MOSFET gates tied to Vdd/Vss are supposed to pass any current. Latch-up does not result in current passing through the gates.

It's the case of the inputs being connected to a different Vdd, with wrong power-up sequencing.

Some logic IC families specify in the datasheet that they are safe against inputting voltages on unpowered chip. That's an actual marketing argument.

Needless to say, this does not apply to most use cases, including the OP's.

[wraper]

It's the case of the inputs being connected to a different Vdd

It's a completely different case from what was asked, doing so is simply stupid and those who do this should be burned on stake.

[Siwastaja]

Most of the "best practice" bullshit is based on mental copy-paste of something that usually is not related to the actual problem, often not even close, and given some engineering discipline, the disconnection is trivial to see.

It's a colossally bad idea to add random components because someone somewhere thinks that they are necessary in a completely different circuit, and to call that "good practice" is disgrace. Engineering just isn't that simple. Understanding the circuit is paramount.

[tkamiya]

imo,

That's a bunch of obvious but good points.  I think most folks (me too) do this without thinking.  Pick the closest low impedance (ie. power) point.  But taken literally, nothing stops folks from just daisy chaining empty input pins and just ground or pull up at one point.  That would be problematic.

[MasterT]

If I asked, is it safe to apply 5.5V? Everyone would say, yes, DS specifies acceptable voltage level. But reliability numbers would fall from 10^-7 to 10^-6 or like that. So, from reliability point of view, increasing voltage (or using safety resistors) should be considered based on Acceptable failure rates. And this strongly depends on a market, cost of device, and things that difficult to estimate - brand name ratings. Nobody would care if led-blinking toy breaks down, OTOH, if a scope failed because someone save on single resistor - would you buy another one from this maker?

[Siwastaja]

If I asked, is it safe to apply 5.5V? Everyone would say, yes, DS specifies acceptable voltage level. But reliability numbers would fall from 10^-7 to 10^-6 or like that. So, from reliability point of view, increasing voltage (or using safety resistors) should be considered based on Acceptable failure rates. And this strongly depends on a market, cost of device, and things that difficult to estimate - brand name ratings. Nobody would care if led-blinking toy breaks down, OTOH, if a scope failed because someone save on single resistor - would you buy another one from this maker?

What does this comment has to do with this topic, though?

And what the heck is a "safety resistor"?

Do note that sprinkling resistors randomly without a reason won't increase reliability, and if they are in series with actual nets (instead of, say, on the side of the board unconnected), each resistor slightly decreases the reliability (not by much, of course). I would expect the scope manufacturer to only design in resistors that are needed or do good for the circuit, and don't put in any excess.

[TimFox]

The specific reason to add (small) resistors at inactive input pins is to allow testing the device after it is soldered in the circuit.  It is a reasonable design rule.  By inactive, I mean inputs (e.g., enables) that are not exercised in operation.

[Wimberleytech]

Ok, I did some searching and I can't find a good definitive answer.

Is it safe to connect to inputs of the 74-series CMOS logic directly to VCC instead of using pull-ups? What are realistic pros and cons?

I'm not interested in what may happen if device fails catastrophically and the input becomes an output and stuff like that.  I'm only interested in normal operation.

I'm also not interested in TTL logic. The devices will never be substituted on accident. Seriously, some people still make this argument.

I guess the question may be rephrased as: are there realistic use cases where it is not safe to connect inputs directly to VCC?

My answer is: Other than for reasons of testing,  there is no realistic use case for using a resistor pullup (versus a direct connection) to VCC.
I read this thread before answering.  The testing idea was not something I thought of, but it is an interesting idea.  For production, you would not want to waste lots of resistors for testing features used only during qualification and/or debug.

[thinkfat]

Ok, I did some searching and I can't find a good definitive answer.

Is it safe to connect to inputs of the 74-series CMOS logic directly to VCC instead of using pull-ups? What are realistic pros and cons?

I'm not interested in what may happen if device fails catastrophically and the input becomes an output and stuff like that.  I'm only interested in normal operation.

I'm also not interested in TTL logic. The devices will never be substituted on accident. Seriously, some people still make this argument.

I guess the question may be rephrased as: are there realistic use cases where it is not safe to connect inputs directly to VCC?

My answer is: Other than for reasons of testing,  there is no realistic use case for using a resistor pullup (versus a direct connection) to VCC.
I read this thread before answering.  The testing idea was not something I thought of, but it is an interesting idea.  For production, you would not want to waste lots of resistors for testing features used only during qualification and/or debug.

Unless you're really space-constrained, is it really an issue to have a couple of additional resistors sprinkled onto the board? Especially if they're not introducing another unique BOM item.

[741]

Here is a similar thread
https://www.eevblog.com/forum/projects/is-it-ok-to-directly-tie-logic-inputs-to-vdd/

[MasterT]

If I asked, is it safe to apply 5.5V? Everyone would say, yes, DS specifies acceptable voltage level. But reliability numbers would fall from 10^-7 to 10^-6 or like that. So, from reliability point of view, increasing voltage (or using safety resistors) should be considered based on Acceptable failure rates. And this strongly depends on a market, cost of device, and things that difficult to estimate - brand name ratings. Nobody would care if led-blinking toy breaks down, OTOH, if a scope failed because someone save on single resistor - would you buy another one from this maker?

What does this comment has to do with this topic, though?

And what the heck is a "safety resistor"?

Do note that sprinkling resistors randomly without a reason won't increase reliability, and if they are in series with actual nets (instead of, say, on the side of the board unconnected), each resistor slightly decreases the reliability (not by much, of course). I would expect the scope manufacturer to only design in resistors that are needed or do good for the circuit, and don't put in any excess.

May be won't. Since you don't provide specific numbers to estimate reliability, and me nether, what I'm trying to say is that question has no sense w/o strictly confidential  manufacturer's reliability report for specific part number. How much failure rate goes up if not follow to recommended design rules.
 OP says, that doesn't need a highest military/ aerospace reliability marks. But consumer product is also demands reliability numbers not equal to zero.

[TimFox]

This question probably arises because of a real feature of original 7400-series TTL, which was actual TTL, using a multiple emitter input transistor instead of DTL diodes to do logic.  The breakdown voltage of the input structure was less than the smoke level of the rest of the circuit.  Therefore, with a direct pull-up to Vic, there was an unnecessary risk of damage if Vcc overshot on power-on.  The situation is different with later bipolar and CMOS families.

[Wimberleytech]

Ok, I did some searching and I can't find a good definitive answer.

Is it safe to connect to inputs of the 74-series CMOS logic directly to VCC instead of using pull-ups? What are realistic pros and cons?

I'm not interested in what may happen if device fails catastrophically and the input becomes an output and stuff like that.  I'm only interested in normal operation.

I'm also not interested in TTL logic. The devices will never be substituted on accident. Seriously, some people still make this argument.

I guess the question may be rephrased as: are there realistic use cases where it is not safe to connect inputs directly to VCC?

My answer is: Other than for reasons of testing,  there is no realistic use case for using a resistor pullup (versus a direct connection) to VCC.
I read this thread before answering.  The testing idea was not something I thought of, but it is an interesting idea.  For production, you would not want to waste lots of resistors for testing features used only during qualification and/or debug.

Unless you're really space-constrained, is it really an issue to have a couple of additional resistors sprinkled onto the board? Especially if they're not introducing another unique BOM item.

A couple? No big deal. Ten? Dunno.  This boils down to designer's choices really.  There is no "correct" answer I suppose.

[wraper]

Unless you're really space-constrained, is it really an issue to have a couple of additional resistors sprinkled onto the board? Especially if they're not introducing another unique BOM item.

Adding parts "just because" without any understanding or measurable evidence of any benefits is not a good design practice IMHO. Over-engineer here and there and result is simply worse, as minimum because of larger size and manufacturing price.

[wraper]

There is no "correct" answer I suppose.

IMHO the correct answer is what isn't needed shouldn't be there. You'd better add some real ESD protection where it might help instead of sprinkling useless resistors.

[Siwastaja]

Oh, there certainly is a correct answer. Except for some very specific cases for testing (then the designed would know what they need!), no one has provided any pointer for what the resistors are supposed to do in the circuit. If no such mechanism cannot be shown at all; even speculated; the correct answer is clear as day.

[graybeard]

But what does SCR latch-up has to do with connecting inputs to the VCC directly?

nothing

[graybeard]

It is fine for CMOS to connect logic inputs to the chip power supply VCC.