I took the design from Peter Oakes design here
I designed and made up some PCBs for it but was having real issues, so I decided to go back to start from scratch.
I understood the Zener Diode reduces pull up voltage down because MOSFET turns full on at around 4V so I kept it the driving stage the same.
I took out the voltage sensing Op Amp because I don't need it and removed the buffer because I am using an MPC3428 ADC and it can read negative voltages (I thought one less source of error)
This option also allows me to use one TLE2442 per control circuit (voltage and current)
With the sense inputs shorted I get 6.7V. (I have not been using a DAC in this test circuit, just controlled voltage from an divider circuit)
The LT1054 should provide 100mA from the data sheet and the Op Amp is drawing far less.
I haven't checked the stability of the control loop yet as I have been at home and not had a scope.
Does this make it clearer and suggest anything else I may be doing wrong?