That’s why I said “None (or almost none)”. I didn’t want to make a categorical statement, but I don’t know of any and I don’t know why someone would build one - it’s better bandwidth wouldn’t help with anything that’s already out there so you might as well use some other standardized interface.
A lot of the experiments I’ve heard of people doing with TOSLink against other inputs are flawed in that they leave the other input connected while they listen to TOSLink. The beauty of optical interfaces is galvanic isolation: keeping things electrically connected while listening to optical connections throws the baby out with the bath water.
The experiments I’ve done between I2S, AES/EBU, S/PDIF and TOSLink from the PWT to the DS show that all of the connections sound essentially identical if you don’t disconnect all wires except the one you are listening to. I.e. leaving everything connected means that you are masking effects of differing shielding, grounding, etc. and since the DS is relatively input jitter insensitive the jitter differences between the various inputs doesn’t matter much.
I’m certainly happy with TOSLink when I don’t need to play anything greater than 176.4k. Unfortunately I broke the TOSLink output from my PWT 