Re: BeagleBone Black and GPIO issues

[Mouse <mouse%Rodents-Montreal.ORG@localhost>]

>>> When I enable a pulldown the the voltage on the pin reads 1.6 V
>>> which doesn't make me super happy, but it doesn't seem to be
>>> floating.
>> Did you try strapping it to Vcc and GND with some 3-10 K of
>> resistance to see if that affects the voltage?

> OK, with a 6.7 kohm resistor connecting the GPIO pin to 3.3 V, I got
> a 0.54 V drop across the resistor and a 2.91 V drop from the pin to
> ground.  This is what I expect after inferring a 30 kohm pulldown
> from the datasheet.

What kind of meter are you using?  In particular, what is its input
impedance - how much does it load the circuit under test?  If you are
using an electromechanical meter (typically a moving needle driven by
parasitic current), it is likely to be drawing something in the
100uA-1mA range at full scale.  If you're using a digital meter, unless
it's _really_ cheap, to a first approximation we can treat it as having
infinite input impedance.

For ideal resistors of 30k to ground and 6.7k to an ideal 3.3V supply,
the junction point should be at 2.69755V.  I note 0.54+2.91=3.45; this
is within 5% of 3.3, but not by much - what do you measure the nominal
3.3V power rail at?

If I redo the calculation with a 3.45V supply, I get 2.82016V at the
junction point, still lower than your measurement.  Perhaps your 6.7k
resistor is lower resistance than nominal and/or the 30k pullup is
higher than nominal; the latter strikes me as particularly likely, but
see below.

> Also, I set my multimeter to ammeter mode and put it inline between
> the GPIO pin and 3.3 V and it read the nominal leakage current of
> 0.11 ųA.

That's not enough parasitic power to do much to most electromech meter
mechanisms, so I'm going to guess you do have a digital meter as
sketched above.

>> Depending on various factors, 1.6V could very well be what you get
>> for a floating input pin.
> The floating voltage appeared to be about 3.1 V, so I don't think
> it's actually floating, but something is going on that I don't
> understand.

Indeed!

When you say "[t]he floating voltage", you mean...what?  The measured
voltage with both pullup and pulldown disabled?  You could be measuring
the thermally-generated voltage across the protection diodes or
something equally meaningless.

If you disable both pullup and pulldown and use 6.7k to ground, what do
you get?  6.7k to 3.3V?

If I've understood right, you have

- Pulldown enabled, nothing but the meter connected: 1.6V
- Pulldown enabled, 6.7K to (nominal) 3.3V: 2.91V

This is consistent with many possibilities, most of them not plausible
here (because they involve voltage sources outside the 0-3.3V range),
but certainly enough of them plausible.  Could you switch that 6.7K to
pin-to-ground instead and see what you get?

In particular, assuming an ideal 30k pulldown and an ideal 3.3V supply,
I can explain the 1.6V measurement by postulating 31.875k to 3.3V.  If
I set the "3.3V" rail to 3.45V instead, I get 34.6875k instead.  These
are close enough to the nominal pulldown resistance to make me wonder
if you think you have the pullup disabled but actually have it enabled
somehow.

If I assume a 3.45V supply, 30k to ground, 31.875k to 3.45V, and also
6.7k to 3.45V, then, assuming ideal everything, I get 2.91252V at the
junction point.  This is provocatively close to your stated measurement
of 2.91, leading me to urge even more strongly that you look into the
possibility that the pullup is somehow enabled.

/~\ The ASCII				  Mouse
\ / Ribbon Campaign
 X  Against HTML		mouse%rodents-montreal.org@localhost
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B