Ah, indeed. Thanks, I will investigate it more
Good idea.
Clicking on the texts for the pins in the ERC window will show you which items are being flagged by KiCad. (These are not always the exact same as where the actual error is).
I do not understand what is the reason that connecting via resistor is suggested and not directly shorting to GND or VCC. I could understand if the connection could happen to be long wire so in created by it circle someone could expect may be inducting some currents from external fields. But if you connect to just next pin being GND what can be the idea that R is better.
It’s an old-established practice for a good reason.
On today’s CMOS parts, there’s no problem in connecting an input directly to VCC/VDD or GND.
The problem occurs with tristate or bidirectional pins (and KiCAD ERC messages reflect this).
In stable operation, your circuit is fine.
But what happens during power-up/down or reset/initialization or a malfunction? It’s unknown, or at least extremely hard to analyze.
Pull-up/down resistors solve that problem.
BTW: 1 kohm is a ridiculously low value for CMOS, dunno where TI got that from. Ballpark is 100 kohms, but almost anything goes ( = what you have in the drawer).
10k is my traditional pull down value. Back in the days of TTL logic, we used 1k pull ups, as the TTL input was actually quite high current in “Low”
With input being driven by 100k I would be not sure what this input sees when 10V/m (or higher) high frequency field is present.
While I can understand the potential problem with bi-directional pins i don’t understand while input pins can’t be directly shorted to VCC or GND.
It is like saying that transistor base (or gate) can’t be directly connected to GND because something unexpected can happen when its collector (drain) voltage is on its way from 0 to VCC.
The use of a resistor here is due the the pin being tristate. If the direction pin got a dry joint and floated, you don’t end up shorting the power supply and melting the buffer. Committing resistors on inputs also made cut and strap modification easier back in DIP package times.
1 Like
I will add 10K resistors there when will refactor the project
I never said inputs couldn’t. Read again.
Check your RPi Pico symbol. Something’s wrong with pin 38. It should be a power input, but is apparently defined as bidirectional. Heaven knows why.
I didn’t questioned anything you said.
I just, by association to this discussion, wanted to get some feedback about what you know of requirement to not connect regular inputs directly to VCC/GND.
I hope in each my sentence I clearly written that I am speaking about inputs and that I don’t know why they should be connected by resistors. I didn’t mentioned that KiCad reports it as a bug, but only said that such requirement raises my doubts.
I have seen from time to time such information in past but never get a credible argument for it and just supposed that may be someone here have such.
First sentence I quoted in this thread:
made me to think that someone in this thread said that inputs should be connected by resistors.
In one our device I connected directly to GND most of microcontroller (so bidirectional) pins.
It was pen-drive like USB device with case made from two halves so ESD can go inside and hit microcontroller pins. I assumed that it is the only way to protect microcontroller against damage by ESD. Device passed ESD tests even spark were going inside it.
Now we have a sealed, airtight housing so no sparks reach inside, but I left connecting all not used pins to GND as they are few mm from external world and ESD tests are also against sparks nearby your device.
I think most microcontroller outputs withstand short shorting to GND so even before reset takes place the bidirectional pins happens to be set as output it shouldn’t damage them, I think.
1 Like
Yep, indeed, pin 38 in RPi Pico symbol is defined as “Bidirectional”
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.