The negative feedback loop is probably confused...
With the transistors missing, and so the PSU not producing any output, the control loop has probably increased the pulse width to its maximum - I'm surprised that it's not more than 25% though.
With the IGBTs removed you'll be able to test that the drive to them is ok (right voltages, nice clean edges etc) but you'll not be able to test the control loop.
You may be able to vary the pulse width though by injecting a signal into somewhere appropriate in the circuit to fool the feedback loop into thinking the output has increased.
Apologies for the terse response - time to get back to work!