If you wish to maintain the +/-15V swing, squarewave over a range of frequencies, you will require an OPA that can operate from dual supplies of about +/-18V.
Can I ask what type of OPA's are you testing that require a +/-15V test input.???
I don't have the schematic with me right now. All I am trying to do is simulate an input to a controller in which one of its inputs uses zero-crossing detection to make frequency measurements. The real input when the controller is installed in the real system is sinusoid up to 2.5kHz and a varying amplitude up to +/- 30V. So, a simple square +/- 15V (or +/-12V etc) will work fine for software development and bench testing purposes. I have a separate microcontroller that I can use to generate 0-5V pulse trains of whatever frequency I wish. I am trying to take that ucontroller output to drive a circuit that will make +/-15V and feed that input to the controller that has the zero-crossing detection input.
See the attached block diagram.
I may be wrong, but it seems that there ought to be a very simple way to create that circuit (I just don't have enough specific expertise/experience to design it). The biggest limitation I have is that I only have a single unipolar DC power supply to use for this circuit between the controllers. So, it seemed some sort of voltage divider opamp circuit and a DC-DC converter IC (like the sample I provided in other posts) might work.
The circuit you have on post #14 would work perfectly; just change the +15V on the comparator/op amp to +18V, and likewise change -15V to -18V. The reason for this is that these devices don't reach their rails perfectly and need a little overhead.
But this circuit violates the constraint of a single unipolar supply. So you need to somehow generate -18V from what you have. A "simple DC-DC converter" to the rescue. An example is the circuit which KMoffett offered in post #6.
Use the circuit in #6 to supply operating voltage o the circuit in #14.