On 2020-02-29, The Natural Philosopher  wrote:
> On 28/02/2020 22:01, Alister wrote:
>> On Fri, 28 Feb 2020 12:27:48 -0500, Dennis Lee Bieber wrote:
>>
>>> On Fri, 28 Feb 2020 11:39:05 GMT, Jan Panteltje
>>> 
>>> declaimed the following:
>>>
>>>
>>>> The GPIO as _digital_ input sees a 'logic zero' below some voltage and a
>>>> 'logic one' above  some voltage.
>>>> Th exact voltage can vary,  also  depends on temperature and production
>>>> spread,
>>>> but is somewhere between 0 and 3.3 V
>>>>
>>>>
>>>  Presuming common CMOS thresholds of 30 and 70%: <0.99V is LOW,
>>> 2.31V
>>> is HIGH. What the circuit does between those thresholds is indeterminate
>>> (I'd hope it holds the last valid state until the far threshold is
>>> crossed).
>>
>> that is exactly what does NOT happen.
>> the logic sate will flip at some in-determinant point between the two
>> thresholds. The thresholds are simply points at which the state is
>> guaranteed.
>>
>>
> Not even that. it will in fact hover between states drawing lots of
> current if its normal CMOS. You can make a crude analogue amplifier out
> of a CMOS inverter.


On the last statement: I can confirm that a CMOS inverter does
make a pretty decent analog amplifier for some purposes.  In the
latter 1970s, while a freshman studying EE, I used CMOS part
number 4449 inverters (balanced current drive) as analog gain
elements for an integrated stereo amplifier.  I think it was an
article in Popular Electronics that gave me the idea.  IIRC, the
article showed that unity gain was around 1 MHz with a 15V power
supply.  For audio purposes, the inverters worked very well.

The basic idea is to use DC negative feedback to bias the
inverter at the crossover point where the input and output
voltages are the same.  For audio purposes, it works best to use
split power supplies, +5V and -5V up to +5.7V and -7.5V.  You
want the power supplies to be pretty clean, because power supply
noise rejection is not spectacular.  You use capacitive coupling
between stages to keep stages isolated from each other in DC
terms.  If you want controlled gain, use a series resistor on the
input side to make the inverter's input a virtual ground.  For
audio frequencies with +7.5 and -7.5 power supplies, 10X voltage
gain (20dB) works well, with the feedback resistor being 10X the
value of the series input resistor.

If you need more gain than you can get with one inverter, do
_NOT_ chain three inverters end-to-end and put your feedback loop
around the chain.  I found out later that configuration is called
a ring oscillator.  What I had intended to be a phono input stage
with RIAA equalization turned into a rather strong oscillator,
probably roughly square wave at somewhere in the neighborhood of
200kHz.

The CMOS inverters served well for a few years until I learned
about and could afford BiFET op amps.  I redesigned the whole
thing with BiFET op amps, and that unit is playing music in the
living room as I type this.


--
Robert Riches
spamtrap42@jacob21819.net
(Yes, that is one of my email addresses.)

--- SoupGate-Win32 v1.05