leaving pin floating

[A. Lemon]

Hi!

Is it so bad to leave an input pin floating (from a PIC)?

Is the increased current consumption from input noise at threshold point the
only drawback?

Thanks!
Al

 

[Joerg]

Hi!

Is it so bad to leave an input pin floating (from a PIC)?

Not familiar with PICs but, generally, it is bad. Can't you at least
pull up?

Is the increased current consumption from input noise at threshold point the
only drawback?

On most uC it's the usual CMOS dilemma: Around the VCC/2 level the first
stage will draw cross current. Not a lot but possibly enough to toast
your whole power budget. This cross current is usually not specified so
while it might seem ok on a prototype in production you could be in for
a rude awakening. If the input doesn't have a Schmitt function the next
stage will also cross conduct and that can be a lot more current.

IOW I suggest not to do that.

 

[qrk]

Not familiar with PICs but, generally, it is bad. Can't you at least
pull up?



On most uC it's the usual CMOS dilemma: Around the VCC/2 level the first
stage will draw cross current. Not a lot but possibly enough to toast
your whole power budget. This cross current is usually not specified so
while it might seem ok on a prototype in production you could be in for
a rude awakening. If the input doesn't have a Schmitt function the next
stage will also cross conduct and that can be a lot more current.

IOW I suggest not to do that.

Some uCs have internal pull-ups. If so, then it's OK to leave the
input unterminated. If not, then pull it high or low. If you terminate
it with a resistor, you can use this pin as a spare input or output.
Handy for debug or an oopsie. You can also program the pin as an
output if it's dual function.

 

[Joerg]

Some uCs have internal pull-ups. If so, then it's OK to leave the
input unterminated. If not, then pull it high or low. If you terminate
it with a resistor, you can use this pin as a spare input or output.
Handy for debug or an oopsie. You can also program the pin as an
output if it's dual function.

Yes, you can do it internally if the uC has that capability. What I
wanted to point out is that Al must actually turn that feature on by
writing to a port register. If it's just configured as a plain input
it'll be floating and that's not a good thing.

 

[Tim Wescott]

Yes, you can do it internally if the uC has that capability. What I
wanted to point out is that Al must actually turn that feature on by
writing to a port register. If it's just configured as a plain input
it'll be floating and that's not a good thing.

The TI 28xx DSP chip has weak pullups coming out of reset. It's very nice.

I just took out the pepper shaker and sprinkled a bunch of pulldowns all
over a design, not because the chip doesn't have pullups (or downs), but
because I wanted to make sure that each processor pin had a lead that I
could attach green-wire mods to.

It's just that phase of the project...

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google? See http://cfaj.freeshell.org/google/

"Applied Control Theory for Embedded Systems" came out in April.
See details at http://www.wescottdesign.com/actfes/actfes.html

 

[Joerg]

The TI 28xx DSP chip has weak pullups coming out of reset. It's very nice.

I just took out the pepper shaker and sprinkled a bunch of pulldowns all
over a design, not because the chip doesn't have pullups (or downs), but
because I wanted to make sure that each processor pin had a lead that I
could attach green-wire mods to.

It's just that phase of the project...

Sometimes I use extra vias large enough to hook a wire into, either for
soldering or for a scope probe. With sacrificial wire, of course. I hate
it when folks stick the bare Tektronix probe tip in there and then it's
all bent.

On my last project I am really glad we built it on Veroboard and not
onto a prototype layout. It's that dreaded PID thingie that'll stare me
in the face again tomorrow. Looks like I'll have to wire in a few more
opamp stages and some poor-man's logic. Drat! But I should be happy,
with a layout I'd be sitting there with the Dremel.

 

[Luhan]

Hi!

Is it so bad to leave an input pin floating (from a PIC)?

Is the increased current consumption from input noise at threshold point the
only drawback?

What is you reason for floating the pin???

Floating pins can cause a lot of noise spikes and you just never know
which one will cause a problem some time.

Luhan

 

[Luhan]

What is you reason for floating the pin???

Floating pins can cause a lot of noise spikes and you just never know
which one will cause a problem some time.

Luhan

On second thought, I leave pins 'floating' all the time while
prototyping PIC designs! I just program them all as outputs if they
are not otherwise used.

Luhan

 

[James Beck]

On second thought, I leave pins 'floating' all the time while
prototyping PIC designs! I just program them all as outputs if they
are not otherwise used.

Luhan

As it had been pointed out, there is a finite time that the pin is in
its' default condition (and in PICs it is usually a high impedance
input) before you make it an output and drive it to some state. Never a
good idea. Resistors are cheap.

Jim

 

[Arlet]

As it had been pointed out, there is a finite time that the pin is in
its' default condition (and in PICs it is usually a high impedance
input) before you make it an output and drive it to some state. Never a
good idea. Resistors are cheap.

Jim

If you insist on doing that, you should also put resistors on all
output pins that have a high impedance load.

 

[Luhan]

As it had been pointed out, there is a finite time that the pin is in
its' default condition (and in PICs it is usually a high impedance
input) before you make it an output and drive it to some state. Never a
good idea. Resistors are cheap.

Resistors are far too much trouble for prototyping. Of questionable
value for one-off devices.
And appropriate for most commercial designs.

"Theoretically" there is the time after reset when the floating inputs
'could' cause some extra current drain - in practice, I have never seen
a problem.

Luhan

 

[Rich Grise]

If you insist on doing that, you should also put resistors on all output
pins that have a high impedance load.

Why?

Thanks,
Rich

 

[Luhan]

Resistors are far too much trouble for prototyping. Of questionable
value for one-off devices.
And appropriate for most commercial designs.

"Theoretically" there is the time after reset when the floating inputs
'could' cause some extra current drain - in practice, I have never seen
a problem.

Ok, I powered up a PIC16F819 with one input adjusted to the point at
which maximum noise comes out another pin (the software reads one pin
and outputs to the other).

The worst case increase in VCC current is about 2ma. Less than I had
expected.

I suspect that unlike standard CMOS circuits, there is no need to make
the inputs 'really, really fast' when the processing speed is the
limitation. Also, I believe it is the output circuit on most CMOS
chips where the midline voltage causes the current to spike.

Luhan

 

[Arlet]

Why?

For the same reason. If a device starts up with pins defined as inputs,
there will be a brief delay before the firmware can reprogram them as
outputs, during which they can float as inputs. Of course, this is only
important if the load connected to that pin has high impedance. For
output pins connected to, say, a LED, this is not a concern, as the LED
Vf is typically below the CMOS threshold.

Sometimes this default input state can be maintained for a long time,
for instance, if the reset pin is kept asserted.

 

[James Beck]

Resistors are far too much trouble for prototyping. Of questionable
value for one-off devices.
And appropriate for most commercial designs.

Well there is that way and then there is the "right" way and that is to
tie all unused inputs to power or ground through a resistor.
The worse possible time to break that rule is when you are doing a one
off design. Would you rather work on the design or chase intermittent
strangeness because of floating inputs? What happens when the one off
design gets hit with a little static and there are floating inputs?

"Theoretically" there is the time after reset when the floating inputs
'could' cause some extra current drain - in practice, I have never seen
a problem.

That you know of. See above.

Jim

 

[James Beck]

Why?

Thanks,
Rich

Yeah, why?
If the pin was an input to start with and then gets set to an output I
can see making sure there is something there other than the cap on the
gate of a FET, but if the output is a dedicated out then there are
probably low impedance drivers holding a state.
Maybe I'm missing something. Wouldn't be the first time.

Jim

 

[Luhan]

Well there is that way and then there is the "right" way and that is to
tie all unused inputs to power or ground through a resistor.
The worse possible time to break that rule is when you are doing a one
off design. Would you rather work on the design or chase intermittent
strangeness because of floating inputs? What happens when the one off
design gets hit with a little static and there are floating inputs?


That you know of. See above.

Take a look at Microchip's website for reference designs. I see no
resistors used on output lines. Nor have I noticed any on the many PIC
sites I've seen.

Luhan

 

[James Beck]

Take a look at Microchip's website for reference designs. I see no
resistors used on output lines. Nor have I noticed any on the many PIC
sites I've seen.

Luhan

Outputs?
I thought we were talking about inputs..........

Jim

 

[Luhan]

Outputs?
I thought we were talking about inputs..........

Follow the thread back up, its about whether programming unused PIC
pins as outputs is a problem because of the time between power up and
when the pins actually get set to output.

I would never recommend leaving inputs floating.

Luhan

 

[Arlet]

Yeah, why?
If the pin was an input to start with and then gets set to an output I
can see making sure there is something there other than the cap on the
gate of a FET, but if the output is a dedicated out then there are
probably low impedance drivers holding a state.
Maybe I'm missing something. Wouldn't be the first time.

Take for instance pins used as an address/data bus for an external
Flash/RAM memory. If the microcontroller starts with all pins defined
as generic inputs, and relies on firmware to define these pins as
outputs, there will be a finite time in which these pins can float.

For dedicated output-only pins, there is always a well defined level,
of course, but many microcontrollers have multiple functions per pin,
configurable by software, and they may default to input.