Help with AD620 instrumentation amplifier

[Steve Zissou]

Hi guys.

Im trying to make a circuit using the AD620 in-amp, but Im getting some unexpected results.
I have attatched a sketch of the circuit. The problem is that with no resistor to adjust gain I would expect the gain to be one, but I measure ~100mV on the output instead of ~1mV.
Any ideas on what is causing problems?
Thanks

 

[MikeMl]

Try this: (off the data sheet)

 

[KeepItSimpleStupid]

Try putting a 100 ohm resistor in series with ref.

 

[Steve Zissou]

Thanks, it is now fixed.
But I have a new problem. Im making a temperature sensor with the help of the AD620 and Im getting, again, some unexpected results. I've attached a skecht of my new setup.
The problem is that when I have no temperature sensors, i.e. a short circuit instead, I expect the voltage difference between + and - to be zero. But on the output I measure 8V which is way higher than expected. Rg is not connected so the gain is 1. How is there an eight voltage difference when the gain is 1? I could sort of understand it is the gain was ~1000 but not this case.

Thanks

 

[KeepItSimpleStupid]

That's the same problem. The input bias current has to have something to drop across. So even if it's picoamps and a very small resistance, it's a large voltage. I'll use zero, but it's still a big number. V = Ib/0. So, the usual thing to do is to add series resistors to the inverting/non-inverting inputs. It's a tradeoff of noise and accuracy. Higher value resistors mean higher noise and higher accuracy. 100 to 10,000 is OK. Offset/null may be able to get rid of it, but the resistances are safer.

I ran into it too playing with IA's too.