Preventing differential follower loading?

[s3c]

I'm trying to use a differential follower in a circuit to measure the voltage over a component, I eventually tracked down the bug and was able to reproduce it using the following circuit (note the simulation works fine, it's the real world circuit that doesn't):



Now with only these components hooked up it works well enough until the Setable point drops below 0.6V, when this happens the output of U2 remains at 0.6V even though the + pin of U2 is below -. The kick is that when I unhook the follower from the differential amp it starts following normally and can reach to ground, I haven't tried it with another op amp yet but I'm still curious as to why this is happening.

When I remove the buffer the circuit works but the differential amp pushes the voltage at this point up slightly in a nonlinear way so the readings aren't accurate. Any ideas?

[toli]

How much current does it have to sink?

Just like the LM324, it can only sink about 50uA before the output jumps to around 0.6V.

[s3c]

Jip, that did it, thanks toli. I didn't even know to look for these specs. I'm slowly realizing that college courses don't focus nearly enough on practical aspects and that spice models aren't what they're made up to be.

[toli]

I'm still a student myself

Its just that I've had a very good professor who was in charge of the Analog Circuit course which was focused mainly on feedback networks and op-amp circuits.

[s3c]

Just read chapter four of Art of electronics to gain some insight about nonideal op amp behaviour, it mentions the lm358 by name and why it behaves in this manner, anyway...

[tecman]

If you need to go closer to the bottom rail, and not so critical about the high side, you can load the output to V- with a resistor.  This can pull the output lower as you are sinking some percentage of the current.

paul

[s3c]

Thanks techman, I got it working but could you give an example of this perhaps since I've read about it somewhere but haven't seen it?