"Bob Eld" <nsmontassoc@[EMAIL PROTECTED]
> wrote in message
news:D7wbk.3604$vn7.1702@[EMAIL PROTECTED]
>
> "Bob Monsen" <rcmonsen@[EMAIL PROTECTED]
> wrote in message
> news:NCubk.6440$LG4.3212@[EMAIL PROTECTED]
>> "Bob Eld" <nsmontassoc@[EMAIL PROTECTED]
> wrote in message
>> news:0zabk.16122$Ri.4973@[EMAIL PROTECTED]
>> >
>> > "Eeyore" <rabbitsfriendsandrelations@[EMAIL PROTECTED]
> wrote in message
>> > news:486C4867.F6DAF534@[EMAIL PROTECTED]
>> >>
>> >>
>> >> Bob Eld wrote:
>> >>
>> >> > "Eeyore" wrote
>> >> > > Bob Eld wrote:
>> >> > >
>> >> > > > Insure that the max peak of the audio signal is within
>> >> > > > the ADC range.
>> >> > >
>> >> > > Diode clamp it !
>> >> > >
>> >> > > Graham
>> >> >
>> >> > Yeah you can clamp it but then you loose that and higher values as
>> > peaks.
>> >> > The whole idea is to convert the peaks to numbers not to clip them
> off.
>> >>
>> >> You can actively diode clamp. Just don't want to hurt the poor ADC
> 'just
>> > in
>> >> case'. I doubt that for flickering lights, losing anything below say
>> >> 0x240
>> > will
>> >> hurt any
>> >>
>> >>
>> >> > There's really no need to clamp it. The processor input will have
>> > internal
>> >> > "diodes" that will keep the voltage from exceeding the rail
voltage
>> > plus a
>> >> > little.
>> >>
>> >> A discussion we've had many a time before here !
>> >>
>> >>
>> >> > Likewise the input voltage won't go below ground.
>> >>
>> >> It WON'T ? In what ideal world is this ?
>> >>
>> >>
>> >> > A series resistance should be added to limit the input current and
a
>> > shunt
>> >> > resistance
>> >> > added to keep the DC level from ****fting up due to rectification
by
>> > these
>> >> > clamps.
>> >>
>> >> Uh ?
>> >>
>> >> Graham
>> >
>> > Well, if there is an internal diode from the input to the negative
>> > rail,
>> > it
>> > will keep the voltage on the input from going below ground, ignoring
>> > the
>> > diode drop, of course.
>> >
>> > Now, if you capacitively couple to the input, this diode will rectify
> the
>> > negative going ****tion of the audio signal and charge the capacitor,
>> > positive towards the input. This will offset the zero base line in
the
>> > positive direction and screw up the peak detection. A judicious
> resistance
>> > to ground with an appropriate series resistance will limit the
current
> and
>> > drain off this offsetting charge. It ain't rocket surgery!
>> >
>> >
>>
>>
>> One thing to remember is that cmos inputs are usually prone to latch-up
>> disease. So, if the output goes higher than the positive voltage input
of
>> the IC, you can end up with a triggered SCR between Vcc and ground. Not
>> good.
>>
>> So, having a way to prevent this is a good thing, and generally
consists
> of
>> a diode from the input to the positive rail. For digital inputs, I
>> believe
>> most microprocessors use an internal diode. However, I'm not sure about
>> analog inputs, and would not want to risk it, particularly when a cap
>> from
>> an external input can easily cause this situation, and the solution is
so
>> simple.
>>
>> See this link for more information:
>>
>> http://ww1.microchip.com/downloads/en/AppNotes/00763b.pdf
>>
>> Regards,
>> Bob Monsen
>
> The PIC stuff has internal protection diodes on the inputs. As far as I
> know
> all of their processors are so protected on the IO ****ts whether
> configured
> analog or digital. For a typical A/D input, see the PIC16F676 Data Sheet
> page 47. There should be no trouble driving a few milliamps into these
> diodes clamping off a signal without any external diodes.
>
> I've never yet seen one latch up doing this. That's probably why the
> diodes
> are there.
>
>
That is a microchip appnote, so somebody there appears to believe it can
happen, at least in some cases. Their summary suggests that it'll be a
problem if the input in question is driven higher than Vcc at startup,
which
is a real possibility with an audio input which is driven externally.
On the other hand, I just spent 20 minutes trying to get a PIC12F675 to
latch up, and was not able to make it happen. So, maybe it isn't such a
big
deal. The appnote I posted pointed to a couple of cir***stances, like
inductive loads, driving power mosfets. Perhaps the spike has to be at
exactly the right time.
Regards,
Bob Monsen
|
