Multisim's 3T_Virtual Op-Amp
I ran into a seemingly odd result the other day in a simulation I was running on Multisim 12 using the three terminal virtual op-amp. This sketch shows what I found:
Clearly, something was amiss. A voltage follower cannot be made with the op-amp's non-inverting input tied back to the op-amp output. An effort was undertaken to understand this result as follows.
It has been suggested that since everything here is ideal, since the op-amp has zero offset voltage, zero bias current and mucho-mucho gain, that this apparently stable condition would be like the theoretical stability of an ideal pencil balanced on its sharpened point.
In theory, that pencil could stay upright forever, but even the slightest perturbance would make it fall and thus, even the slightest offset would make the output of a virtual op-amp get railed.
I gave it a try and here's what happened.
The pencil on its point analogy seems to apply. Even the U2 op-amp in the middle that was not given a deliberate input got railed anyhow, apparently an artifice of numerical accuracy in the program itself.
The moral of the story is to check for this condition. You can introduce a disturbance as shown above or you can temporarily replace the virtual op-amp with a real world part's model and see if latching occurs or not.
The '741 model can serve nicely for that as we see in the following case of the TL431 model that was recently shown in this blog.
These two '741 results show us that one of the ideal, 3T_Virtual op-amp results shown above is wrong and more importantly, the '741 results show us which one is wrong.