Discretionary policy in NK model with debt - eigenvalue prob

[winter]

I try to solve the policy problem of a discretionary policy maker in a New Keynesian model with government debt using the corresponding Dynare routines (version 4.3.2). The policymaker decides about private consumption (-> nominal interest rate), government consumption and the income tax rate. Trying different calibrations, the routines always find some eigenvaues to violate the QZ criterium. This also happens if I use exactly the same setup and calibration as Leith, Wren-Lewis: "Fiscal sustainability in a New Keynesian model," who solve their model using the Matlab routines of Soderlind (1999).

I have attached the mod file. Your help would be very much appreciated.
Thanks!