No voltage can be applied to any pin when the board is not powered.
This is a *very* difficult condition to satisfy in any sort of real
world application. If you're using the Beaglebone to measure and/or
monitor something it's almost certain that the 'something' will be
there and turned on whether or not the Beaglebone is turned on.
OK, there are ways of doing it, but it's not easy. Even if inputs are
clamped in some way, some voltage will almost certainly appear on the
BBB inputs.
So, what does 'No voltage' mean? If it really means zero volts then
it really is impossible. Is there not be some (low) voltage that the
inputs will tolerate even when the Beaglebone isn't powered up? Most
chip specs tend to specify about 0.5 volts above and below the supply
voltage so this would mean + or - 0.5 volts from zero when the
Beaglebone isn't powered.
Not really. The idea of powering a chip via an I/O pin will always cause damage. It means voltage as specified by the datasheet of the component.
It is your choice, but, it will cause damage to the part.
If you power a system form one power source, it is not hard to do. Using two power sources that are not synchronized with each other, that is where the issue comes into play.
Gerald
[-- text/plain, encoding quoted-printable, charset: ISO-8859-1, 56 lines --]
Not really. The idea of powering a chip via an I/O pin will
always cause damage. It means voltage as specified by the datasheet of
the component.
I don't aim to 'power' it via the I/O pin! Maybe that's your way of
saying it but it's a very odd way. The likelihood is that there will
be a biggish resistor in series with the input to limit current and
there will probably also be some clamping diodes or maybe a buffer
amplifier but whatever you do there *cannot* be 'no voltage'.
What I'm asking really is what will be tolerated with no problems,
every chip spec I have ever seen specifies some sort of minimum, not
zero.
So what is the "voltage as specified by the datasheet of
the component."? That's what I'm asking really, I'll go and look at
the processor spec sheet.
It is your choice, but, it will cause damage to the part.
If you power a system form one power source, it is not hard to do. Using
two power sources that are not synchronized with each other, that is where
the issue comes into play.
Real systems don't come like that! Are you going to turn off every
single thing in you car/boat/house etc. just so you can power down a
Beaglebone monitor?
If you apply power to an I/O pin without power on the power rails of the processor, that is basically what you are doing!
Read the datasheet for the processor and look at the power sequencing required by the designers of the processor.
http://www.ti.com/product/am3358
Gerald
Given this restriction can anyone recommend a small board that can regulate 12v to 5v and 3.3v.
I will be driving motors and steppers at 12 v and want to use one source for that and the BBB.
Thanks
Eric
cl@isbd.net wrote:
> [-- text/plain, encoding quoted-printable, charset: ISO-8859-1, 56 lines --]
>
> Not really. The idea of powering a chip via an I/O pin will
> always cause damage. It means voltage as specified by the datasheet of
> the component.
>
I don't aim to 'power' it via the I/O pin! Maybe that's your way of
saying it but it's a very odd way. The likelihood is that there will
be a biggish resistor in series with the input to limit current and
there will probably also be some clamping diodes or maybe a buffer
amplifier but whatever you do there *cannot* be 'no voltage'.
What I'm asking really is what will be tolerated with no problems,
every chip spec I have ever seen specifies some sort of minimum, not
zero.
... and the processor spec *does* tell me!
The limits are specifically stated (as I expected) as follows:-
"Steady state max. voltage at all I/O pins"
"-0.5 volts to IO supply voltage +0.3 volts"
So, even with power off, some voltage *is* allowed and in fact it
should be fairly easy to keep the voltage within these limits using
Schottky diodes for clamping.
*This* is what I've been asking for.
BUT, there is no I/O supply if the I/O supply is not turned on. The I/O supply has to be there.
Look, go ahead and ignore me if you like. But I talk to the designers of this device daily and support hundreds of customers daily. And I see boards come into the RMA department all the time with blown processors due to this issue. The power sequencing diagram is there for a reason due to the multiple voltage rails inside this device. Violate the power sequencing and you will have issues.
Gerald
> [-- text/plain, encoding quoted-printable, charset: ISO-8859-1, 56
lines --]
>
> Not really. The idea of powering a chip via an I/O pin will
> always cause damage. It means voltage as specified by the datasheet of
> the component.
>
I don't aim to 'power' it via the I/O pin! Maybe that's your way of
saying it but it's a very odd way. The likelihood is that there will
be a biggish resistor in series with the input to limit current and
there will probably also be some clamping diodes or maybe a buffer
amplifier but whatever you do there *cannot* be 'no voltage'.
What I'm asking really is what will be tolerated with no problems,
every chip spec I have ever seen specifies some sort of minimum, not
zero.
... and the processor spec *does* tell me!
The limits are specifically stated (as I expected) as follows:-
"Steady state max. voltage at all I/O pins"
"-0.5 volts to IO supply voltage +0.3 volts"
So, even with power off, some voltage *is* allowed and in fact it
should be fairly easy to keep the voltage within these limits using
Schottky diodes for clamping.
Schottky diodes aren¹t going to clamp the voltage to this range. Simply
use the 3V3 output from the BBB to enable the supply to your board.
Regards,
John
???
How would using the 3.3v IO output from the BBB be any different from
clamping the input voltages to the 3.3v IO output from the BBB? It's
the same thing surely?
One Schottky diode prevents the voltage going below 0.3 volts (it will
conduct such that the voltage doesn't go below 0.2 volts), another can
clamp the input to prevent it going above the 1.8 volt ADC supply
voltage.
It MAY, and I stress the MAY, be safe to allow voltages between -0.5 and + 0.3 Volts to appear at the I/O pins of the AM355x processors when they are not powered up, but using schottky diodes to keep them to within +0.3 volts of an external 3.3 supply rail WILL fry them.
The suggestion made above was to use the presence of the 3.3 V supply on the BBB to Enable / Switch the external supply On, or conversely the absence of the BBB 3.3 Volt supply removing or disabling the external 3.3 Bolt supply.
Has this made it clearer ? I do hope so, I hate it when the Magic Blue Smoke escapes from a chip 
>"Steady state max. voltage at all I/O pins"
>"-0.5 volts to IO supply voltage +0.3 volts"
>
>So, even with power off, some voltage *is* allowed and in fact it
>should be fairly easy to keep the voltage within these limits using
>Schottky diodes for clamping.
Schottky diodes aren¹t going to clamp the voltage to this range. Simply
use the 3V3 output from the BBB to enable the supply to your board.
???
How would using the 3.3v IO output from the BBB be any different from
clamping the input voltages to the 3.3v IO output from the BBB? It's
the same thing surely?
I didn’t say 3V3 IO, I said 3V3 output or 3V3 supply output (P9-3 or
P9-4). If you have a regulator on your 12V side, you probably have an
enable pin. Connect P9.3 to the enable pin.
One Schottky diode prevents the voltage going below 0.3 volts (it will
conduct such that the voltage doesn't go below 0.2 volts), another can
clamp the input to prevent it going above the 1.8 volt ADC supply
voltage.
I’m not sure you understand how a schottky diode works. Voltage drop at
forward biases of around 1 mA is in the range 0.15 V to 0.46 V, but if you
conduct more that 1mA, your volt drop will be higher. Also, the reverse
voltage will not clamp at all.
Regards,
John
[-- text/plain, encoding quoted-printable, charset: ISO-8859-1, 73 lines --]
BUT, there is no I/O supply if the I/O supply is not turned on. The I/O
supply has to be there.
If it's not turned on then it's 0 volts isn't it? Power supplies
don't go open circuit when they're turned off so the 3.3 volt line
will be sitting at close to 0 volts and you'd need to pump a lot of
current into it to drag it away from there. There will be 10k or
more of series resistance in the inputs so we're only talking about
less than 1mA of current going through the clamping diodes, that's not
going to move the supply line voltage by anything much.
Look, go ahead and ignore me if you like. But I talk to the designers of
this device daily and support hundreds of customers daily. And I
see boards come into the RMA department all the time with blown processors
due to this issue. The power sequencing diagram is there for a reason due
to the multiple voltage rails inside this device. Violate the
power sequencing and you will have issues.
I'm not violating the power sequencing, I'm obeying it carefully by
ensuring that there's no voltage on the inputs until the 3.3 volt I/O
supply appears.
cl@isbd.net wrote:
> >"Steady state max. voltage at all I/O pins"
> >"-0.5 volts to IO supply voltage +0.3 volts"
> >
> >So, even with power off, some voltage *is* allowed and in fact it
> >should be fairly easy to keep the voltage within these limits using
> >Schottky diodes for clamping.
> Schottky diodes aren¹t going to clamp the voltage to this range. Simply
> use the 3V3 output from the BBB to enable the supply to your board.
>
???
How would using the 3.3v IO output from the BBB be any different from
clamping the input voltages to the 3.3v IO output from the BBB? It's
the same thing surely?
One Schottky diode prevents the voltage going below 0.3 volts (it will
conduct such that the voltage doesn't go below 0.2 volts), another can
That should say -0.3 volts of course.
[-- text/plain, encoding quoted-printable, charset: ISO-8859-1, 68 lines --]
It MAY, and I stress the MAY, be safe to allow voltages between -0.5 and +
0.3 Volts to appear at the I/O pins of the AM355x processors when they are
not powered up, but using schottky diodes to keep them to within +0.3 volts
of an external 3.3 supply rail WILL fry them.
You've misunderstood me, I'm going to clamp to the BBB's 3.3 volt I/O
supply.
The suggestion made above was to use the presence of the 3.3 V supply on
the BBB to Enable / Switch the external supply On, or conversely the
absence of the BBB 3.3 Volt supply removing or disabling the external 3.3
Bolt supply.
Has this made it clearer ? I do hope so, I hate it when the Magic Blue
Smoke escapes from a chip 
That's a possible way of doing it but I don't like the extra
complexity, it's another way for things to go wrong. If that
'logical' connection between the BBB's 3.3v IO supply and my external
supply goes wrong then the inputs get fried.
I prefer simply using the BBB's 3.3 IO supply *directly* (with a diode
of course) to clamp the inputs. We're talking about very small
current here, I'm aiming to have 10k (or more maybe) series resistors
in the inputs so the current through the diode even with a 3.3 volt
difference between input signal and BBB input it's only 300 microamps
or so.
>> >"Steady state max. voltage at all I/O pins"
>> >"-0.5 volts to IO supply voltage +0.3 volts"
>> >
>> >So, even with power off, some voltage *is* allowed and in fact it
>> >should be fairly easy to keep the voltage within these limits using
>> >Schottky diodes for clamping.
>> Schottky diodes aren¹t going to clamp the voltage to this range. Simply
>> use the 3V3 output from the BBB to enable the supply to your board.
>>
>???
>
>How would using the 3.3v IO output from the BBB be any different from
>clamping the input voltages to the 3.3v IO output from the BBB? It's
>the same thing surely?
I didn't say 3V3 IO, I said 3V3 output or 3V3 supply output (P9-3 or
P9-4). If you have a regulator on your 12V side, you probably have an
enable pin. Connect P9.3 to the enable pin.
>
We're not talking about things with power supplies, I'm measuring
battery voltage. Strangely enough most batteries don't have an
'enable pin'. 
I'm after the simplest way I can think of to do this. A simple
resistor divider to drop the battery voltage I'm measuring down to
something in the range 0 to 1.8 volts and then some way of protecting
the BBB input from spikes and when it's turned off. Adding buffers
(op amps presumably) with their own power supplies etc. is (to my
mind) unnecessary complexity.
>One Schottky diode prevents the voltage going below 0.3 volts (it will
>conduct such that the voltage doesn't go below 0.2 volts), another can
>clamp the input to prevent it going above the 1.8 volt ADC supply
>voltage.
I'm not sure you understand how a schottky diode works. Voltage drop at
forward biases of around 1 mA is in the range 0.15 V to 0.46 V, but if you
conduct more that 1mA, your volt drop will be higher. Also, the reverse
voltage will not clamp at all.
The ones I am looking at have a forward voltage drop of around 0.2
volts at 1mA, since the series resistors on the input (after dividing
the voltage down to 1.8 volts) will be 10k or so the current won't
even get to 1mA - so less than 0.2 volts.
There is one diode clamping to the 0v rail, that prevents the input
voltage going below about -0.2 volts. There will be another diode
clamping to the ADC supply voltage which will prevent the input
voltage going above ADC supply voltage plus about 0.2 volts again.
I started my electronics career in the 1960s, I'm not entirely
inexperienced or unqualified!