Op amp questions - high input impedance inverting circuit, or direct coupling?

[composition4]

Hi

Using a 0-9v single ended power supply, as we do.  Trying a design and need to use an op amp to invert a signal, but require high input impedance without a large input resistor before the inverting input.

Is there a way to configure an inverting op amp stage so that it has a high (1M) input impedance?

Or alternatively if there isn't, I will place a non-inverting op amp stage buffer in front of it - but is there a way I can direct-couple the output of a non-inverting op amp stage to the input of a non-inverting op amp stage? Just so I don't have to stick a cap between them?

Please let me know if I need to clarify or give more details, it's 3:30am here and way past my bedtime.

Thanks
Jonathan

[ElectricDruid]

No, there are no really good solutions to high impedance *and* inverting op-amps. You can do one or the other, but not both, or not both *well*, anyway.

Yes, you can easily direct couple the two stages. The first one would be a non-inverting buffer (voltage follower) stage. It'd have 1M to virtual ground from the +ve input and a cap feeding the signal in. The output from that amp is therefore reference to that virtual ground level and will swing around that point. Consequently, there's no need to AC couple the signal to the next op-amp stage. The output of one amp can go straight to the input resistor of the inverting stage. You'll only need a cap agin when you get to the output, wherever that is...

HTH,
Tom

[composition4]

Thanks for the reply, appreciated. So I thought the same thing with the direct coupling, however my simulator is telling me I'm doing something wrong.

Top picture is with the post buffer output cap C6 in place, the output is 12x the input as expected.

Bottom picture is with C6 removed, look at the graph... I'm not sure what I'm missing, I'm sure it's simple.  Vref = 4.5V.

Cheers

[R.G.]

Well, there are no (1) good (2) commonly known solutions. The problems are that you have to use big resistors on the inputs, and even bigger feedback resistors for any gain, and the specter of noise is always looking over your shoulder.

But we often use 1M input resistors on (+) inputs, so a 1M film resistor should not put us off. So the trick is to use small resistors in the feedback path and still get attenuation that is more than a greater-than-1M feedback resistor.

That's most easily done with a T network. Two feedback resistors in series, and one resistor between them to signal ground to make the T. It's really better looked at as a resistor attenuator, with the first resistor from the output and the resistor to ground being a voltage divider, and the resistor from that tapped middle going to the inverting input for feedback.

It takes some math, but you really can use a 1M input series resistor to an inverting input, a T network for feedback, and get high(ish) input resistance plus gain in an inverting opamp stage.

[antonis]

What R.G. says... 

http://sound.whsites.net/dwopa.htm
(4.2 for scheme & calculations..)

[teemuk]

Yes, you can easily direct couple the two stages. ...output from that amp is therefore reference to that virtual ground level and will swing around that point. Consequently, there's no need to AC couple the signal to the next op-amp stage. ... You'll only need a cap agin when you get to the output, wherever that is...

AC coupling is required when DC offsets of the coupled stages are different, like for example 4.5 VDC reference for unipolar 9V supply vs. 0 VDC reference for a bipolar supply.

Theoretically stages with equal DC offsets can be direct coupled. In some designs direct coupling is even mandatory for biasing either one of the coupled stages. The trick is common for both discrete and opamp circuitry.

In practice, direct coupling is not always so trivial. A gain stage may feature a tiny DC offset (uncommon to general reference level) and when the stage directly couples to another gain stage the DC offset may get amplified. Then we're dealing with a bigger problem. In easiest-case -scenario the DC offset just passes along to the following stage with unity gain. So, sometimes AC coupling is introduced merely to prevent minor DC offsets of "real-life" circuits from passing along the signal path.

[composition4]

Thanks for the feedback guys, interesting methods. I stuck with the non-inverting buffer in front of the inverting stage, as it only cost me an extra opamp in the design.

However still puzzled as to what I'm doing wrong trying to direct couple the output of the non-inverting buffer to the inverting input of the next opamp.  It will just not work without that cap in between the stages.  Ah well it will only cost me three extra caps in the design if I can't find out what I'm doing wrong.

Cheers!

[amz-fx]

However still puzzled as to what I'm doing wrong trying to direct couple the output of the non-inverting buffer to the inverting input of the next opamp.  It will just not work without that cap in between the stages.  Ah well it will only cost me three extra caps in the design if I can't find out what I'm doing wrong.

You have a small DC offset that is getting multiplied by the inverting opamp, as teemuk explained.  A resistor on the non-inverting pin of the inverting amplifier might help to compensate, but why bother when a capacitor will do the job even better. 

regards, Jack

[Kipper4]

Nice RG I'm sure  could put this to good use.

[composition4]

Ah gotcha, thanks... sorry must have read past that. Managed to get my coupling caps down to reasonable values anyway, so no big problem to use them.