Hi,
In the real life circuit we cant depend on the transistors in the current mirrors all being EXACTLY the same. There will be some imbalance. You could check this by using small resistors in series with the current mirror collectors. See how different they are with the real life circuit.
Im not sure if i still have IC's with multiple transistors inside or i would try this myself, at least to test the current mirrors without the rest of the circuit as that is my guess of what the problem could be right now. They may even have to be matched. It would be pointless to try this with separate transistors. The voltage/current characteristics have to be identical, and if the temperature changes the voltage/current characteristics have to change the same in both transistors, so we're asking a lot.
I am still intrigued by this circuit though (at least with one set of current mirrors) because it is able to increase the output slew rate, which is very important for the large signal response (and therefore the usability) of the op amp.
We could look into what to do about such an imbalance. Some sort of current mirror offset adjustment maybe, but there is a chance that using the offset null of the 741 might do it too. Depends on how much range you can get and if that is functionally a good enough approach to begin with.
The 741 inputs should be ok as long as they are within about 4 volts of either rail. So any input should be between say -11 and +11 volts with a plus and minus 15 volt supply.
The outputs should be within about 2v of each rail i think, and no greater.
For testing CMRR you have to use resistors less than 10k, but 1k resistors are the most common to use i think, so using 1k resistors should give good results.
Come to think of it i dont think i have any 741's around anymore either as i stopped getting them long, long, long, long time ago. I still get LM358's once in a while, and some specialized extremely low input offset types. Dont have to use them much anymore though except for the occasional measurement.