# Unexpected saturation with the rail to rail instrumental amplifier AD623

I'm trying to use the rail-to-rail instrumental amplifier AD623. Here the datasheet.

simulate this circuit – Schematic created using CircuitLab

First experiment: AD623 set as a voltage follower. Gain = 1 obtained by removing the Rg resistance between Rg- and Rg+ pin.

Everything goes right until I reach 4.3 V (read with VM1 and VM2) 4.3 V is the maximum amplifier output even when the difference between inputs is 5V. But this is a rail-to-rail amplifier...

Second experiment: AD623 set with gain = 2 (using a 100K resistor in according to datasheet).

Everything goes right until I reach 2 V of difference between inputs.(read with VM1) In this case 4.0 V is the maximum amplifier output.

Under this value the amplifier works good (if in input I have 1V in output I read 2V)

Other information:

• +Vs = +5V
• -Vs = GND
• REF = GND

How is it possible?

Can anyone help me?

Thanks for help (and sorry for my poor english)

• Are you providing a path for input bias current? It's not clear that you are from the schematic you provide. See page 19 of the data sheet. – Alfred Centauri Oct 12 '13 at 2:20
• you are right! I completely forgot the bias current. 3 hours ago there was another answer below, but now is disappear. They say that AD623 isn't a rail-to-rail Amplifier. The presence of bias current can produce, in this case, a difference between theorical and real output of few mV, it's true? – Grigno Oct 12 '13 at 11:51
• I think the now removed answer was that the input isn't rail to rail but that wouldn't account for your result with gain 2 so I'm guessing that's why it was removed. – Alfred Centauri Oct 12 '13 at 12:27
• Did you figure it out? – Anthony Dec 30 '17 at 17:40
• @Anthony I think so, that was an old project and, at the end, was a success. In the 2nd prototipe the problem was no longer present. I think the problem was a simple mistake but after 5 year I don’t remember... I should have written it here. Are you experience the same problem? – Grigno Dec 30 '17 at 18:29