AD623 unstable and not amplifying

Question

I need a x18 amplifier to go from a sensor that provides 0-200mV and get 0-3.6V at the output. I have made the following circuit: 
 
 I have applied -1.5V with a battery between pins 4 (-1.5V) and 5 (0V). The circuit is supplied with 5V between pins 7 (5V) and 5 (0V). Between pins 2 and 3 I have applied 100.0 mV to adjust the resistance between pins 1 and 8 to the exact gain, but it has not amplify the signal. I have used different chips in case some disaster occurred when connecting it, but the same thing has happened with all of them. I've checked the diamond diagram and there doesn't seem to be anything abnormal: 
 
 After not finding the error and seeing that erratic voltages appear at the output (for example from 30mV to 40mV) I ​​have tried to decouple pins 4 and 7 from ground, adding two separate 10uF capacitors, but the situation has not changed. Anyone ever happened something similar? Thank you.

EB3BNJ · Accepted Answer

Thanks Achim. I will measure all again because something, maybe a bad soldering or so it's giving me all my headaches. 
 
 Thanks again.)

Achim · Answer

Hello Juan, 
 EB3BNJ said: I found 0.25V in the common mode 
 The input bias current of the AD623 can go up to 25nA. If your multimeter has an input resistance of 10MOhm (a typical value), then the 0,25V would have been created by 25nA*10MOhm. So the source was indeed floating and you "grounded" it with the 10MOhm of your multimeter. That fit's nicely. 
 EB3BNJ said: I put two 1MOhm resistor beetween ground and both inputs, but the problem is still here. 
 Ok, that does not really fit. 1MOhm should be low impedant enough to provide the input bias current. If the problem persists, I have no idea based on the description you gave. I could only imagine, that something in the described setup and the real setup does not match perfectly. Maybe theres an open connection/a bad soldering. 
 In your case, I would once again measure the voltages at every single pin with reference to pin 5. I'd measure directly at the pin of the chip (not at some trace which "should be connected" to the pin). Maybe a table of the measured voltages could give a hint for further failure analysis. 
 best regards 
 Achim

Electronic Design Services - PartnerZone

Products Mentioned

AD623 unstable and not amplifying