Author Topic: Is it safe to connect inputs to VCC directly  (Read 9934 times)

0 Members and 1 Guest are viewing this topic.

Online iMo

  • Super Contributor
  • ***
  • Posts: 5570
  • Country: va
Re: Is it safe to connect inputs to VCC directly
« Reply #25 on: February 05, 2020, 12:45:20 am »
Inputs are not designed to pass much current, so there is "probability"  latching-up, and it's easier to install a resistor than estimate 12-20 inputs/outputs interference to each other .
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.
Have you ever seen how a CMOS input looks like?
Readers discretion is advised..
 

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Is it safe to connect inputs to VCC directly
« Reply #26 on: February 05, 2020, 12:58:41 am »
Inputs are not designed to pass much current, so there is "probability"  latching-up, and it's easier to install a resistor than estimate 12-20 inputs/outputs interference to each other .
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.
Have you ever seen how a CMOS input looks like?
Yes I did and it's a non answer.
 

Offline MasterT

  • Frequent Contributor
  • **
  • !
  • Posts: 851
  • Country: ca
Re: Is it safe to connect inputs to VCC directly
« Reply #27 on: February 05, 2020, 01:08:56 am »
As I said, I'm not interested in catastrophic failures. If there is over-voltage or power pin connection loss, then device is broken. It does not really matter what happens after that.

I don't buy this "good designer" stuff.

I see this matter for aerospace or military applications, but there you have to think about much more than just a simple pull-up.
Failures happen all the time. It's saving on resistors what makes them catastrophic.  Cellular towers and mobile phones all around us, no need to be in aerospace.   
 

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Is it safe to connect inputs to VCC directly
« Reply #28 on: February 05, 2020, 01:13:55 am »
It's saving on resistors what makes them catastrophic.
It's like saying saving on fur coat could make your life catastrophic. But I doubt it will make your life safer if you live in tropical forest. What I mean is that you should use stuff for a reason and I don't see how dumbly sprinkling resistors over unused inputs is supposed to make anything better.
Quote
Cellular towers and mobile phones all around us, no need to be in aerospace.   
And in that case you better tie things hard with low impedance paths instead of using resistors which cannot quench RF interference.
« Last Edit: February 05, 2020, 01:19:46 am by wraper »
 

Offline MasterT

  • Frequent Contributor
  • **
  • !
  • Posts: 851
  • Country: ca
Re: Is it safe to connect inputs to VCC directly
« Reply #29 on: February 05, 2020, 01:21:26 am »
Inputs are not designed to pass much current, so there is "probability"  latching-up, and it's easier to install a resistor than estimate 12-20 inputs/outputs interference to each other .
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.
I 'd refer to TI - White Paper
SCAA124 – April 2015
Latch-Up
Marty Johnson, Roger Cline, Scott Ward, Joe Schichl
Quote
During the design of signal ESD protection structures, there can be
intentional SCRs employed that encourage selected parasitic PNPN diodes to trigger under an ESD event.
While ESD is an unpowered event, this is not a concern and actually desirable. However, during normal
operation since an intentional SCR could exist between a signal pin to ground and the signal pin could be
tied or driven to a “hi” state, care must be taken by using design rules on the product to limit effects of
Signal Latch-Up, thereby, mitigating the effect of excursion that could trigger the signal ESD SCR and
effectively short the signal to ground. If a signal pin is tied or driven to a “lo” state, then the signal voltage
is below the ESD SCR holding voltage and signal Latch-Up will not occur.
In short, there is certain "design rules".
 

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Is it safe to connect inputs to VCC directly
« Reply #30 on: February 05, 2020, 01:32:40 am »
Inputs are not designed to pass much current, so there is "probability"  latching-up, and it's easier to install a resistor than estimate 12-20 inputs/outputs interference to each other .
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.
I 'd refer to TI - White Paper
SCAA124 – April 2015
Latch-Up
Marty Johnson, Roger Cline, Scott Ward, Joe Schichl
Quote
During the design of signal ESD protection structures, there can be
intentional SCRs employed that encourage selected parasitic PNPN diodes to trigger under an ESD event.
While ESD is an unpowered event, this is not a concern and actually desirable. However, during normal
operation since an intentional SCR could exist between a signal pin to ground and the signal pin could be
tied or driven to a “hi” state, care must be taken by using design rules on the product to limit effects of
Signal Latch-Up, thereby, mitigating the effect of excursion that could trigger the signal ESD SCR and
effectively short the signal to ground. If a signal pin is tied or driven to a “lo” state, then the signal voltage
is below the ESD SCR holding voltage and signal Latch-Up will not occur.
In short, there is certain "design rules".
You forgot to mention that in only happens when there is intentionally made SCR structure on inputs for ESD protection.
 

Offline MasterT

  • Frequent Contributor
  • **
  • !
  • Posts: 851
  • Country: ca
Re: Is it safe to connect inputs to VCC directly
« Reply #31 on: February 05, 2020, 01:51:27 am »
Quote
Cellular towers and mobile phones all around us, no need to be in aerospace.   
And in that case you better tie things hard with low impedance paths instead of using resistors which cannot quench RF interference.
I'd not be so sure about 5 GHz, that connecting pin - essentially small piece of antenna, to ground is more safe. Another end may have 1/4 length to  be hardly energized.
 I can't comment what kind of SCR is in use nowadays, last time I seen CMOS in X-ray in 90-th. But my understanding, since human body doesn't change much since than, has same charge capacity and lethal for CMOS energy, the same time size of components (MOSFET, shottky diodes ) shrink substantially, means that simple diode-R-diode chain may not be enough to satisfy same ESD standard. So, manufacturers have to incorporate much more sophisticate protection, SCR or whatever.
 

Offline james_s

  • Super Contributor
  • ***
  • Posts: 21611
  • Country: us
Re: Is it safe to connect inputs to VCC directly
« Reply #32 on: February 05, 2020, 02:25:25 am »
More like stupid practice how to increase manufacturing costs and problems due to higher component count. Also you reduce noise immunity or robustness against board contamination or tin whiskers. Say CMOS output (connected to input) has much lower impedance than a few k resistor you would use as pull-up.

How does tying unused inputs to Vcc or ground increase manufacturing cost and increase component count?
 

Offline Miti

  • Super Contributor
  • ***
  • Posts: 1405
  • Country: ca
Re: Is it safe to connect inputs to VCC directly
« Reply #33 on: February 05, 2020, 02:33:21 am »
Connecting inactive inputs to ground or Vcc through appropriate resistors allows testing the IC in-circuit by driving the input from a “stiff” test voltage on a test jig.  This is part of design-for-test process.
This argument is just strange. Why would you ever want to override an input, which value is fixed at the design stage? What are you testing here exactly?

Why don't you decouple the inputs that are driven by other outputs? So you could override those as well for testing?

What is so special about pins that must have fixed value?

To kill a clock so the MCU doesn't start running the code. To keep an MCU in reset or otherwise it would interfere with your ICT or boundary scan. To pull up a "test enable" pin to put a JTAG compatible device in... well... test mode.
This is a strong requirement in our DFT guidelines.
Fear does not stop death, it stops life.
 

Online ataradovTopic starter

  • Super Contributor
  • ***
  • Posts: 11905
  • Country: us
    • Personal site
Re: Is it safe to connect inputs to VCC directly
« Reply #34 on: February 05, 2020, 02:37:45 am »
To kill a clock so the MCU doesn't start running the code. To keep an MCU in reset or otherwise it would interfere with your ICT or boundary scan. To pull up a "test enable" pin to put a JTAG compatible device in... well... test mode.
Those are provisions for dedicated pins. I'm not arguing with that.

I'm asking specifically about single 74-series gates. Let's say I use 4-NAND and I only need 3 inputs. All I need is to force one other input to be always high. Any testing like that should happen through other 3 pins. This extra pin carries no function in the design. Unless you assign it a function to be a test input.
Alex
 

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Is it safe to connect inputs to VCC directly
« Reply #35 on: February 05, 2020, 02:40:03 am »
More like stupid practice how to increase manufacturing costs and problems due to higher component count. Also you reduce noise immunity or robustness against board contamination or tin whiskers. Say CMOS output (connected to input) has much lower impedance than a few k resistor you would use as pull-up.

How does tying unused inputs to Vcc or ground increase manufacturing cost and increase component count?
I thought you meant adding resistors as good practice.
 

Offline tkamiya

  • Super Contributor
  • ***
  • Posts: 2178
  • Country: us
Re: Is it safe to connect inputs to VCC directly
« Reply #36 on: February 05, 2020, 02:55:19 am »
Quote
A lot of modern "TTL" stuff is actually CMOS.
Haven't seen anything claiming to be TTL. There might be written something like TTL voltage level.

Now I am genuinely interested in finding more.  Taken literally, you are saying 74LSxxxx are actually CMOS with TTL interface.  I am not finding anything on this.  Will you please give me an example device so I can further research this?  I have not seen anything like this.
 

Online oPossum

  • Super Contributor
  • ***
  • Posts: 1472
  • Country: us
  • Very dangerous - may attack at any time
Re: Is it safe to connect inputs to VCC directly
« Reply #37 on: February 05, 2020, 03:04:12 am »
74HCT and 74AHCT are CMOS with TTL thresholds
 

Offline tkamiya

  • Super Contributor
  • ***
  • Posts: 2178
  • Country: us
Re: Is it safe to connect inputs to VCC directly
« Reply #38 on: February 05, 2020, 03:10:04 am »
Oh, OK.  That I knew.  I thought you guys were talking about 74LSxxxx series being redesigned as quasi TTL.
 

Offline Miti

  • Super Contributor
  • ***
  • Posts: 1405
  • Country: ca
Re: Is it safe to connect inputs to VCC directly
« Reply #39 on: February 05, 2020, 03:19:41 am »
To kill a clock so the MCU doesn't start running the code. To keep an MCU in reset or otherwise it would interfere with your ICT or boundary scan. To pull up a "test enable" pin to put a JTAG compatible device in... well... test mode.
Those are provisions for dedicated pins. I'm not arguing with that.

I'm asking specifically about single 74-series gates. Let's say I use 4-NAND and I only need 3 inputs. All I need is to force one other input to be always high. Any testing like that should happen through other 3 pins. This extra pin carries no function in the design. Unless you assign it a function to be a test input.

Yes, our customers with good design experience do this all the time, pulling input pins directly to GND or Vcc, even dedicated pins. Many that didn't design-for-test before, have an eureka moment when we ask them to add pullup/dn resistors in our DFT reports.
Fear does not stop death, it stops life.
 

Offline tkamiya

  • Super Contributor
  • ***
  • Posts: 2178
  • Country: us
Re: Is it safe to connect inputs to VCC directly
« Reply #40 on: February 05, 2020, 03:40:18 am »
We've been talking about resisters.  Like what values?  Normal pull up like 470 ohm to 10K ohm range?  And the "test" that has been referred to, they are extreme tests like overlaying KV pulse on power line, etc, or zap the input with huge noise.  Right?  We are not talking about normal usage in office.
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17428
  • Country: us
  • DavidH
Re: Is it safe to connect inputs to VCC directly
« Reply #41 on: February 05, 2020, 03:49:26 am »
But what does SCR latch-up has to do with connecting inputs to the VCC directly?

If you had more than one Vcc supply, which used to be more common, and the input was connected to the other one, then power supply sequencing could cause SCR latch-up.  Now this would be more likely to happen at the interface between different voltage domains.

And.....  Why do input of TTL logic gate have to be pulled up or down?  It is completely safe to just leave it unconnected.  At least I've never done it and never seen any one else do it that way.  I thought the practice was for ?mos type devices only.

Some parts in the later bipolar TTL families, I think the ones built on complementary bipolar processes so some AS and FAST parts at least, had buffered inputs and present a much higher impedance to the input making them susceptible to capacitive coupling into an unconnected input.  I would not put it past having the same problem with advanced low-power schottky or even low-power schottky parts.
« Last Edit: February 05, 2020, 03:57:00 am by David Hess »
 

Offline tkamiya

  • Super Contributor
  • ***
  • Posts: 2178
  • Country: us
Re: Is it safe to connect inputs to VCC directly
« Reply #42 on: February 05, 2020, 03:57:40 am »
Thank you.  Learned something new today.  It was a simpler time when chip was made by Texas Instruments and started with 74, it was always TTL, and if the chip was made by Motorola and started with 4, it was always CMOS.  Yes, I'm that old....
 

Offline aix

  • Regular Contributor
  • *
  • Posts: 170
  • Country: gb
Re: Is it safe to connect inputs to VCC directly
« Reply #43 on: February 05, 2020, 06:25:54 am »
Let's say I use 4-NAND and I only need 3 inputs.

I do realise that your question is broader than this.  However, this particular example got me wondering: to get 3-NAND from 4-NAND, what are the pros and cons of tying one input high vs tying two of the four inputs together.
 

Online ataradovTopic starter

  • Super Contributor
  • ***
  • Posts: 11905
  • Country: us
    • Personal site
Re: Is it safe to connect inputs to VCC directly
« Reply #44 on: February 05, 2020, 06:56:57 am »
I do realise that your question is broader than this.  However, this particular example got me wondering: to get 3-NAND from 4-NAND, what are the pros and cons of tying one input high vs tying two of the four inputs together.
I have no idea. I don't think there is a significant difference. Intuition tells me that having one input fixed is better, it avoids one more thing changing, but I don't know for sure.

In this case question came up for XOR gates, so no way to tie things together.
Alex
 

Online tggzzz

  • Super Contributor
  • ***
  • Posts: 21227
  • Country: gb
  • Numbers, not adjectives
    • Having fun doing more, with less
Re: Is it safe to connect inputs to VCC directly
« Reply #45 on: February 05, 2020, 08:23:56 am »
To kill a clock so the MCU doesn't start running the code. To keep an MCU in reset or otherwise it would interfere with your ICT or boundary scan. To pull up a "test enable" pin to put a JTAG compatible device in... well... test mode.
Those are provisions for dedicated pins. I'm not arguing with that.

And that's an example of the problems caused by omitting context. You did write. (with my  emphasis)...

Connecting inactive inputs to ground or Vcc through appropriate resistors allows testing the IC in-circuit by driving the input from a “stiff” test voltage on a test jig.  This is part of design-for-test process.
This argument is just strange. Why would you ever want to override an input, which value is fixed at the design stage? What are you testing here exactly?
...
« Last Edit: February 05, 2020, 08:28:07 am by tggzzz »
There are lies, damned lies, statistics - and ADC/DAC specs.
Glider pilot's aphorism: "there is no substitute for span". Retort: "There is a substitute: skill+imagination. But you can buy span".
Having fun doing more, with less
 

Online ataradovTopic starter

  • Super Contributor
  • ***
  • Posts: 11905
  • Country: us
    • Personal site
Re: Is it safe to connect inputs to VCC directly
« Reply #46 on: February 05, 2020, 08:28:18 am »
Well, you did write...
There is a difference between the pin that must normally be held high when not in use, like programming and test pins,  and a pin that is absolutely never needs to be overridden, like that 4th unused input on a 4-NAND gate.

Nobody tries to override random parts of the circuit "for testing". That's just nonsense.
« Last Edit: February 05, 2020, 08:31:29 am by ataradov »
Alex
 

Online tggzzz

  • Super Contributor
  • ***
  • Posts: 21227
  • Country: gb
  • Numbers, not adjectives
    • Having fun doing more, with less
Re: Is it safe to connect inputs to VCC directly
« Reply #47 on: February 05, 2020, 08:30:31 am »
I do realise that your question is broader than this.  However, this particular example got me wondering: to get 3-NAND from 4-NAND, what are the pros and cons of tying one input high vs tying two of the four inputs together.
I have no idea. I don't think there is a significant difference. Intuition tells me that having one input fixed is better, it avoids one more thing changing, but I don't know for sure.

In some cases the extra load capacitance might be an issue.
There are lies, damned lies, statistics - and ADC/DAC specs.
Glider pilot's aphorism: "there is no substitute for span". Retort: "There is a substitute: skill+imagination. But you can buy span".
Having fun doing more, with less
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17428
  • Country: us
  • DavidH
Re: Is it safe to connect inputs to VCC directly
« Reply #48 on: February 05, 2020, 10:42:40 am »
Let's say I use 4-NAND and I only need 3 inputs.

I do realise that your question is broader than this.  However, this particular example got me wondering: to get 3-NAND from 4-NAND, what are the pros and cons of tying one input high vs tying two of the four inputs together.

Except for edge cases where doubling the capacitance and leakage matter, you can always do that and it is even the recommended method for TTL where it does not even increase the fan-in of a single gate.
 

Online iMo

  • Super Contributor
  • ***
  • Posts: 5570
  • Country: va
Re: Is it safe to connect inputs to VCC directly
« Reply #49 on: February 05, 2020, 11:29:02 am »
As indicated by above posters - connecting a CMOS input "directly to Vcc or Gnd" of the SAME chip is safe.

What could be an issue - and that is what the people try to indicate above - is when somebody "thinks the CMOS input is connected directly to the Vcc or Gnd" (mind the word "directly"), but in reality:

a) the wire is rather long - it pickups transients from its surrounding
b) the Vcc or Gnd potential (at the point of connection) is DC biased (ie a voltage drop on Vcc or Gnd rails) against the Vcc or Gnd at the CMOS chip
c) the Vcc voltage at the CMOS chip vs. the voltage at the point of connection differ during powerup
d) the Vcc voltage at the CMOS chip vs. the voltage at the point of connection is noisy..
etc.

In those cases there could be a peak current flowing through the first two (upper or lower) clamping diodes (the input of a CMOS logic gate consists typically of 2 diodes - 1 resistor - 2 diodes) and the excessive current may cause some issues in the silicon of that CMOS chip. As I wrote above in past the "connection points" where the unused inputs were connected to were sometimes "far away" of the actual CMOS chip, and/or were connected to different Vcc and Gnd power rails. Thus adding a resistor close to the CMOS input pin helped to limit those spurious current peaks.
« Last Edit: February 05, 2020, 01:54:33 pm by imo »
Readers discretion is advised..
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf