flipping cdmd to 01 will enable misfire diag. quick look in a 559d file, it seems to be disabled.

Aww crap. I was afraid it would be something like this. Looks like it's old fashioned fault finding then.

Hmm... That'd be one option I guess. I have a 559A box coming next week, I'll use that as a guinea pig and leave the original as a backup.

I'm guessing You have a 558S binary just laying around somewhere...  
Scratch that, I found the bin.

I finally got around to digging into this. I replaced the cats with straight pipes about a week ago, replaced O2 sensors with Bosch OEM ones at the same time. I have 4D0907558S binary that I downloaded from chiptuners iirc but the file has errors according to ME7Check. Neither Winols or Me7Sum seems to be able to correct the file. I'm thinking what other options there are for testing? Would 4D0907559G from a manual S8 work, as in does it have the necessary measuring blocks for misfire detection?

I'm starting to accept that the engine is meant to misfire. During the last year and a half I've had the heads done with some new valves, guides, machining etc the whole deal. New plugs. Coils are tested by replacing them with spares one by one. New filters. SAI delete with block-off plates. Compression check OK. Injector output test OK. No vacuum leaks. Timing belt job done with correct tools and rechecked a few weeks ago just to be sure.

Starts up instantly, pulls and revs real good but puffs at idle. Now that I took the cats off and the exhaust gas is more visible even with a warm engine, You can clearly see when it coughs every now and then, maybe once every two to three seconds. If I take injector connectors off one at a time when the engine is running, it always get noticeably worse so it doesn't run on 7 pots all the time when idling. Raising the idle rpm to 800 helped a little but the misfire still remain. Now I'd really really like to see if it's one single cylinder misfiring or is it randomly on all cylinders.

Any suggestions?

I haven't logged anything or even looked at the ltft's with VCDS after the de-cat and new O2 sensors. Before they were looking normal, 3-6% if I remember right. I'll get back to You with the trims.

Heads should be good as new. Timing belt snapped february last year and I ended up buying a whole replacement engine. When the "new" engine anyway was out of the car, I had the heads inspected at a shop where they did all sorts of work.

Thanks. I had completely missed your reply, I'll give that 558S bin a go. Fuel trims from channel 032 are -2.4% -7.8% -0.2% -4.7%

Do the negative values mean that it's running too rich and tries to lean it?

I flashed the 558S bin. It wasn't drivable but I got to the right measuring blocks. Channels 014, 015 and 016 show misfires. Cylinder 1 is misfiring @ idle.

What I've tried so far:

Swapped coils from 1. to 2. cylinder. Still misfiring with cylinder 1.
Swapped spark plugs from 1. to 2. cylinder. Still misfiring with cylinder 1.
Swapped fuel injector from 1. to 2. cylinder. Still misfiring with cylinder 1.

I'm thinking maybe a valve isn't seating properly? The more I give it throttle, the better it feels. The misfire isn't felt anywhere else but low engine speeds. Any comments?

Just to verify I'm not being an idiot, cylinder 1. is the first one counting from timing belt end on the right side of the vehicle. Right side being right looked from the drivers seat.