Update: I found the source of the problem and a solution!
Like Dan suggested, it's definitely a reflection issue.
Even without anything attached to the FW (no telescope or lenses), the bright ring was still present in my test images.
So, it had to be something in between the filter and the sensor.
I noticed that the filter edges are beveled, and I could see a colored reflection in this edge.
The color was also different in each NB filter.
This is likely the source of the problem.
I found a post on the ZWO forum, where a user reports the same problem:
http://zwoug.org/vie....php?f=8&t=6833
He made small rings to block the filter edges.
So, I cut out small paper rings and placed them on top of the NB filters.
That did it!
Paper is probably not the best material for these rings though, so I'm going to (try to) make new rings out of a thin plastic sheet.
I don't use your camera or filters (I use Astrodon's - worth every penny paid), but a thought to consider:
What light source are you using to make your flats? Narrowband flats require good illumination in their bands, and longer subs. If you took too short of a sub, then you'll stretch harder and amplify minor variations more. My flats are 15-20 seconds for unbinned (1x1) subs. Also, normal lights tend to drop off near the IR region (where Ha and Sii are located) .. make sure you use a full spectrum light.
My old flats were only exposed for 2-3 seconds to reach 12k ADU, but now I know that this is due to the bright reflection ring.
The new flats needed 10-12 seconds with the same light source, so that's as expected. Thanks!