You have to run a UECM and obtain the long-run parameters, that is by dividing through using the  coefficient on the one-period lagged level of the dependent variable, then use these obtained long-run coefficients to create an ECM term. Remember not the Engle and Granger residuals. Then reestimate the UECM using the same lagged structure identified by AIC earlier. 

thanks I will try that

 The answer of Chuck Arise is perfect. Keep it up!

Good luck.

Thanks Ehsan  for your time.It is working, thanks to chuck

Hi, is there a way to estimate two ARDL equations in a system? I am thinking of estimating two UECM form of equations in a system, but i wonder whether it makes sense. Thanks for your comments. 

You can execute a SUR estimator. One can also you Shazam or SAS to execute a three-stage least . I

Thanks Chuck Arize. I am trying to handle it in TSP. I guess TSP does not have a built-in command for ARDL.

what does UECM stands for?