I was in the lab all day today and measured every possible point with a DSO. I feel quite stupid as the Arduino was the culprit. I am very sure it was fine when I was in the lab yesterday as I had checked it with a DSO. It probably started malfunctioning when I reached home yesterday and since I had already tested it in the lab, I assumed it cannot be the issue. It was not properly turning off which was causing the issue.
Anyhow, the driver is working beautifully and I am getting flawless output. I think I am going to stick with ADUM3223 for a long time as it is so perfect - 4A peak current rating and the isolation is built into it so no need to connect isolation externally (optocoupler or digital) it's **** fast and draws less than 0.1uA from the inputs and so can be directly connected to a DSP. Hopefully it is robust and will last long, in which case it is quite perfect.
After spending months in the search for something good, having tries IR2110, IRS21867, IR2332 and TLP250, I think it is safe to say that the ADUM3223 pisses on all of those.
I am sorry RSchnell as I know you went through a lot of trouble but in the end the problem lied in my inputs.
Regards and Thanks.