--- In LTspice@..., Andy <Andrew.Ingraham@...> wrote:
I think what's happening here is that your circuit has two possible stable
states, due to the fact that it has no real V+ supply voltage. If the
op-amp happens to start up yanking on the output pin, it will pull the
positive supply pin voltage down with it so that it goes low, and then it
settles on that second stable operating point, with the output pin
negative. In that state, the op-amp is open-loop and essentially broken.
If that doesn't happen, it lets the output pin (and the positive supply
voltage) go high, and the negative feedback works normally.
With it being a roll-of-the-dice which stable state comes up, a change to
one part of the circuit can cause a change elsewhere.
To prove that there are (at least) two stable states I did the following.
I added a .NODESET V(n010)=10V to the circuit where U1 was replaced by the
UniversalOpamp2, and this makes U2 come up "normally" again.
I tried adding a .NODESET V(n010)=-2.5V to the un-modified circuit. This
seems to cause LTspice to go into spasms where it can't find a clean
operating point to start the transient simulation with, but it runs anyway
and the transient simulation waveforms indicate that it doesn't ever
recover to the "normal" situation where there is a good positive supply
voltage. In other words, it finds the second stable operating point where
U2 is open-loop, and its output remains low.
This is a very poorly behaved negative feedback circuit. Highly
questionable design.
Regards,
Andy
[Non-text portions of this message have been removed]
That's great detective work and analysis! You've shown me a new tool to use in future troubleshooting attempts.
Yes, I know the circuit is strange . . . I inherited it from a customer and was running some simulations in order to understand the functions and problems. The reason for the strangeness is that it runs a the end of a long twisted pair, and the power line is also the signal line. I added the -5V negative rail in an attempt to get the simulation running, since the opamp I selected drew 5 times the current of the part they were using, unbalancing the design. I may try to give them an alternate power+signal 2-wire design that is inherently stable.
Thanks,
Carl
