Looking at the size, the width of the solar part of the array seems to be pretty close to 1m. This would, together with the images, put the length at around 10m, for about 10m^2, or 12kW of solar incident power.

Or perhaps 3000W.

This would conservatively place the weight in the 30kg range.

The area, assuming that it is randomly spinning and 10m^2 would be about 5m^2, so the effective sectional density is 6kg/m^2.

https://spaceflight.nasa.gov/realdata/sightings/SSapplications/Post/JavaSSOP/orbit/ISS/SVPOST.html gives the area of ISS as 17659ft^2, or ~2000m^2, and mass of 420000kg. Or 210kg/m^2.

ROSA decelerates at about 210/6 = 35 times the rate of ISS.

ISS started June at ~405.2km, and will end at 404.7 or so. A drop of 500/30 =16 meters a day, ~1m/orbit.

ROSA should lower in altitude by ~35m per orbit, 560m/day.

This would alter its period by one part in 10000 or so.

Handwaving stuff that I don't understand, this would lead to a miss by of the order of ~5s*8000m/s, or 40km, after one day.

However, it's going to come uncomfortably close for a significant period. An old source says the collision avoidance box is 1km 'deep' from ISS towards earth, so to drop out of that box would take ROSA a couple of days.

It's at least possible that proper understanding of the drag model, and release time could mean that there is no risk to ISS.

500m a day, it could easily be a large fraction of a year before it reenters.

(Above orbital calculations according to Wil.E. Coyote's book of orbital mechanics, where all orbits are linear or square as convenient. I'd love someone to do it properly).