[sdiy] Op Amp AC Coupling/DC Biasing (was Blocking audio =?iso-8859-1?q?at=09power=09off?=)

[Florian Teply]

On Thursday 11 September 2008 19:09:59 Justin Owen wrote:
> >
> >Source - 20k (R18+R13) - 10k (Load) - GND,
> >so, the voltage measured on the load would be one third the voltage of the
> > source.
> >
> >Do we agree so far? ;-)
>
> Multimeter says Yes - so I'm in...
>
> Thanks - go on...
>
Okay, now as we agree to why you have feedthrough when the OpAmp is switched 
off (hope you didn't just measure it but got the idea, please ask otherwise), 
let's see how we could overcome this situation.

As Fabio already pointed out, a non-inverting configuration would be better in 
that case.

I put a very basic example to http://www.teply.info/noninverting.png in order 
to show what i mean.

Let's magine the same circumstances we had with our last thought experiment: 
source connected to one input, output connected to some resistive load, no 
power applied to the opamp.
Now as we put some voltage across one input, say at R1, there will be a 
voltage at the noninverting termianl of the opamp, but as it's unpowered, it 
again acts as if it wasn't there. so, the only possible way for currents 
would be through R3 to ground and back to the source. As the output consists 
of no voltage source (again, opamp is unpowered), there won't be currents 
flowijng through R4 and R5, so the output basically stays at ground. Even 
though we have a similar situation as with the inverting opamp (a resistive 
path from input to output), there won't be a voltage at the output as long as 
the opamp is not powered. The trick is: the circuit formed doesn't rely on a 
load to be closed. In the inverting configuration, current has no means to 
return to the source unless through some connected load, while in the 
noninverting configuration, there will always be a path back to the source.

Now, this was the ideal world, welcome home at reality ;-)

In the real world, opamps come comparably close to ideal behaviour, so most 
rules still apply here. When unpowered, they still behave like they're nearly 
not there, but with some academic approach, one could possibly measure some 
resistance between the input terminals, but that would be in the range of 
several gigohms, which is like an open circuit compared to a couple of 
kiloohms we put in as "real" resistors. Even more academically, one could 
measure some resistance between input and output terminals of an 
real-world-opamp, but that would probably be even higher than the input 
resistance, which again in effect is an open circuit. So, in those 
applications we talk about, most people won't be able to tell real and ideal 
opamps apart.
IF we were talking about some high-power high-frequency high-gain applications 
like the transmit power stages of your favourite radio station, things would 
be different, but, well, this is audio...

HTH,
Florian

-- 
This is my signature...