unexplained behavior with op amp

[danjel]

So this is the second time on a design I have encountered a strange phenomenon with a TL07x
used as a summer.

I have a tl074 with the following:
power +/-12V

see diagrams:

on the bottom circuit with output B, sweeping the pot from fully CCW to CW gives a continuous linear voltage sweep. The only issue is that the opamp is badly oscillating and requires a stability cap to put on it. This is not a big deal but I actually need to use this circuit with different, lower values and it created a totally different problem.

In the top circuit (output A) I get a strange phenomenon: at fully clockwise I get a large positive voltage, as I sweep the pot the voltage gets smaller until there is some point where it suddenly drops a lot and then starts to grow again until a max value when fully CW.
I have no explanation for why there is this non-linear behavior and this is the second time I have seen this on a design.
When I look at the output pin I do not see any oscillations (which was my first guess).
Is this a known design issue with opamps? If so, what is it called and how do I avoid it? The only thing that created change was using different summing and feedback values but I would rather understand what the cause of the problem is.

 

[danjel]

is this possibly the symptom of phase reversal as the summed voltages go too close to negative supply rail?

 

[mvs sarma]

if in the first diagram, if you scale down the bias pot also to 1K as you did all others except output series resistor,
even the first one would behave the same way, i presume.

 

[crutschow]

.......In the top circuit (output A) I get a strange phenomenon: at fully clockwise I get a large positive voltage, as I sweep the pot the voltage gets smaller until there is some point where it suddenly drops a lot and then starts to grow again until a max value when fully CW.....

You may be looking at the resolution limit of the pot. What type of pot is it?

Edit: Looking at the circuit I don't see how the output can go to zero and then increase again. With R2 connected to -5V, the output should go from a positive voltage to zero (or near zero), since when the pot is at the +5V end, its voltage should just counter-balance the -5V at R2 (with R2 equal to R1). If you get otherwise I would suspect an incorrect connection or component value.

 

[audioguru]

Your supply is plus and minus 12V so the maximum allowed input common mode range is from +8V to -8V. These opamps are unpredictable when this input voltage range is exceeded.
The TL07x and a few other older Jfet-input opamps have the problem called, "phase inversion" where the output suddenly goes high if an input voltage is too low: