A week of rain meant i got to move along with this project. The circuit shown in previous post turned out to perform poorly. The top end swing would clip at around 4.3V. Increasing the voltage to 9V allowed the top end to work but the Zener was clipping the 'negative' voltage swing. Not how I thought it would work but guess thats bench testing for you...
I had considered a H-bridge, but the circuit spends about a 3rd of its time at 0 volts, which with some jitter could cause the H-bridge to switch rapidly back and forth...
So figured this was going to require some proper centre tap. Found a +/- 5V 100mA DC-DC converter on element14: PEME1-S5-D15-S. This would provide me with a proper ground reference. With this range, the DAC could output its full 8-bit voltage range, and now all that was needed to to massage this output to get the required -1 to +1 voltage swing, which is a perfect task for one of the old LM358 OpAmps...
See attached but basically R3 and R4 voltage divider (from the -5V supply) and the DAC output are used with the OpAmp as a summing amp to add (subtract) -2.5V from the DAC value, this centres it around the 0V reference. R6 is inverting amp feedback from the push-pull amp, setting the gain at 0.66. This gives a 5V signal from the DAC an output voltage of 3V (or +1.5V).
Except for a bit of distortion around the 0V reference (Which i think might be issues with the breadboard setup), it seems to do as required. So have turned it into a prototype board and send it off to China for Fab. Hopefully it'll be here in a week or two...
Yellow is an input from function generator, 5V P-P centred offset 2.5V
Cyan is the output from the OpAmp (Pin 1)
Pink the the output from the Push-Pull amp.
Love KiCAD 3D Views...