Op-amp Biasing

[tonmann]

Here's a question that has got me stumped.

A standard, non-inverting op-amp stage with +9V, Vref (bias supply) and ground.

Input through C1 to non-inverting input, R1 from non-inverting input to Vref, R2 from output to inverting input, R3 from inverting input to......... Vref or ground ?
And why.

[Eltek]

Here's a question that has got me stumped.

A standard, non-inverting op-amp stage with +9V, Vref (bias supply) and ground.

Input through C1 to non-inverting input, R1 from non-inverting input to Vref, R2 from output to inverting input, R3 from inverting input to......... Vref or ground ?
And why.

This is because if you use a single power supply, the OP-AMP wouldn't be able to amplify negative going parts of the audio signal and would simply clip them at 0V. If you connect the thing as you described(R3 to Vref!), you will be able to amplify both halves of the signal since it will be ac coupled(C1 in series) and will get it's offset(by means of R1). Apart from this, you should put also a capacitor in series with the output and then put a resistor from the remaining cap pin to GND. Take the output signal from the junction of this last RC combination and GND.

Cheers,
Ivan

[R.G.]

A standard, non-inverting op-amp stage with +9V, Vref (bias supply) and ground.Input through C1 to non-inverting input, R1 from non-inverting input to Vref, R2 from output to inverting input, R3 from inverting input to......... Vref or ground ?
And why.

R3 to Vref.

Because the + input is held at Vref. If you attach the noninverting input through an input resistor to any voltage other than Vref, the difference in the voltages is seen as an input voltage and amplified.

Even if R2 = R3 giving a nominal gain from the inverting side of -1, if you take it to ground, you have V+ at Vref and V2 through R3 to what amounts -Vref (with respect to what is on the + input). That -Vref is then multiplied by the R2/R3 ratio and the output will try to sit at +2Vref (that being the +Vref where the noninverting input is being held and the -Vref times -1 going into the inverting input.)

This is why most of the time you see the inverting input taken through a resistor and then through a capacitor to either ground or Vref. The capacitor blocks DC, so the output bias point becomes simply whatever DC voltage is on the noninverting input pin.

[tonmann]

Firstly I must appologise for a very badly formulated question - I was tired, I speak English but think in German, my wife ran off with the drummer etc, etc are no excuses for wasting your time. Hopefully I can re-phrase the question with the help of a diagramme:



As far as I am aware all three circuits work, my question concerns A verses B. What is the difference, if any, and why do some circuits connect C2 to ground while others to Vref.

[R.G.]

As far as I am aware all three circuits work, my question concerns A verses B. What is the difference, if any, and why do some circuits connect C2 to ground while others to Vref.

As long as Vref has an internal impedance that is very low compared to the R-C impedance to the inverting input, over the frequency band of interest, there is no difference.

Any Vref impedance appears in series with the R-C from the inverting input and changes the gain.

Also, a subtle point. Any other circuits using Vref are counting on it being zero impedance too. To the extent it is not, the currents the other circuits source into Vref appears as an error signal driving the R-C to the inverting input. It's one way to get crosstalk.

Of course, the same applied to "ground". If the ground wires connecting the R-C have a high impedance, it gives the same flaws.

[tonmann]

Thank you very much R.G. for a comprehensive answer.
I imagine that connecting multiple stages and/or high gain (high current) stages to Vref would source more current into Vref, producing a voltage change at Vref. This would affect errors also at the non-inverting input as R1 is connected to Vref.

[R.G.]

I came up with a trick a ways back. TI has a part that is intended to make an active Vref, a supply splitter. This is a TO-92 transistor looking thing that is actually an opamp inside with an equal resistor divider chain. The three pins are V+, V-, and Vref. Vref is almost exactly halfway between V+ and V-. It's very low impedance, all the way down to DC. I believe it's the TL2425.

They're hard to find. I figured out the following:
The LM386 power amplifier chip is an 8 pin DIP. It has the unusual property that its inputs are ground (that is, V-) referenced by resistors inside the chip, and it self-centers in its power supply. So set it up with a power supply of, say, 0V and +9V, and its output is accurately half the power supply, and also very low impedance, up to big fractions of an ampere of current. Noise is comparable to resistors. So you can take an LM386 and use it for a Vref, no other parts needed for bias, decoupling, etc. It works well.

[cjlectronics]

That's interesting using the LM386. Thanks for the tip.

I've used the TLE2425 and the TLE2426 T0-92 package in a few designs and they are readily available from Digikey. The 2425 is a 2.5v output Vref and the TLE2426CLP is 1/2v rail splitter from 2 - 20v output.

CJ

[R.G.]

Yeah - that was it, the TL2426. Handy. What do those cost now?

[Enzo]

$1.30 at Mouser. TLE2426 stock 595-TLE2426ILP Also comes in a DIP-8.

[tonmann]

Thank you for the LM386 tip - I will try it in my next build