Sanity check please...

[karbomusic]

Simple non-inverting op amp buffer assuming a TL072...

9V single sided supply + 100k (or 10k) divider for bias then hanging a 2M2 resistor (post input cap) to Vref to set input impedance.

The 2M2 is going to drop my reference voltage way below the 4.5 VDC the divider is supplying as virtual ground.

If I set this up (ignoring AC for a moment) and probe the DC on the output, I get my 4.5 VDC. If I then feed it an AC signal, it buffers properly and doesn't clip until I get close to 1/2 the power supply on the input = good.

Can I assume this is due to self-biasing of the op amp to 1/2 the PS? I'm going with yes until someone scolds me and sets me straight.

[amptramp]

You have set up a divider that gives you 4.5 volts and you have a 2.2 meg resistor to raise the input impedance.  There is almost no bias current in a TL072 amplifier input - it is rated as a maximum of 200 pA which, multiplied by your 2.2 meg resistor, gives you a voltage shift of 440 µV or 0.00044 volts.  This does not drop your reference below 4.5 volts much - in fact the drop should not be noticeable.  The non-inverting amp drives the inverting input to match the non-inverting input so if you have a gain of 1, there is a solid wire going from the output to the inverting input and the amplifier has sufficient DC gain to force both inputs to nearly match.  It has nothing to do with the power supply - the reference point can be adjusted up or down and the DC average on the output will follow.

[karbomusic]

This does not drop your reference below 4.5 volts much - in fact the drop should not be noticeable.

When I add the resistor, pin 3 goes from 4.5 volts to 2 something. Maybe I'm missing the entire boat then. I mean it works fine but looks like I'm missing the underlying idea. I'll do some readin'.

[samhay]

^When I add the resistor, pin 3 goes from 4.5 volts to 2 something. Maybe I'm missing the entire boat then. I mean it works fine but looks like I'm missing the underlying idea. I'll do some readin'.

Are you using a cheap DMM with an impedance << 2.2M perchance?

[anchovie]

Are you using a cheap DMM with an impedance << 2.2M perchance?

This.

My DMM is so cheap it could have come free with the batteries it runs on. Measuring Vref for opamps drops to between 2 and 3 volts if the biasing resistor is 1meg or higher.

[antonis]

If I set this up (ignoring AC for a moment) and probe the DC on the output, I get my 4.5 VDC. If I then feed it an AC signal, it buffers properly and doesn't clip until I get close to 1/2 the power supply on the input = good.

When I add the resistor, pin 3 goes from 4.5 volts to 2 something.

The above two situations aren't consident...

If you had "real" 2 something volts at Vref then you should have a negative cycle clipped signal at 2 something minus something volts...

I don't also believe that the problem lies to your DMM's impedance...
( a 2 something voltage drop with an initial 4.5V is kind of HUGE one...!!!)

[samhay]

^The above two situations aren't consident..

That's not how I see it:
The op-amp's output impedance is very low, so will not be loaded down by the DMM impedance. If this measures about half supply voltage, then the op-amp must be biasing to this voltage via the 2.2M, unless the output is tied to Vref somewhere downstream. However, if this were the case, the buffer is unlikely to work as expected / have the headroom described.

[karbomusic]

^When I add the resistor, pin 3 goes from 4.5 volts to 2 something. Maybe I'm missing the entire boat then. I mean it works fine but looks like I'm missing the underlying idea. I'll do some readin'.

Are you using a cheap DMM with an impedance << 2.2M perchance?

I don't think so, it's 600.00 combo board but maybe...

http://www.digilentinc.com/Products/Detail.cfm?NavPath=2,842,843&Prod=EEBOARD

but I will scope it as DC instead to rule that out?  Will report back after double checking everything this evening.

[antonis]

I totally agree with you Sam...
(but - speaking for a simple buffer - I can't find a point at which the OpAmp's output could be tied to Vref...)

That's why I refered to "incosinstency"..
(sorry for my trully bad English..)

[slacker]

Data sheet says the input impedance of the digilent thing is 1.2Meg so the 2 point something volts measured will be due to meter loading.

[GibsonGM]

That's an awfully low input Z!  Any way to beef that up, using (sorry) an opamp or something?

Heck, my $45 Extech is 7.5Meg...

[karbomusic]

That's an awfully low input Z!  Any way to beef that up, using (sorry) an opamp or something?

Heck, my $45 Extech is 7.5Meg...

LOL, I have a house full of meters so I'll try some of the others and report back.

[amptramp]

Measure the 9 volt battery you are powering the unit from.  Measure Vref to ground.  Then measure the +9 V supply voltage to Vref.  The latter two measurements should add up to the 9 volt battery voltage.  If they don't, the meter is the culprit.  I suspect you will get 9 volts, 2.5 volts, 2,5 volts so you know meter loading is part of the problem.