I have to admit to being strongly biased against urethane anything. Too many times I've seen urethane decay into toxic, highly corrosive messes, even worse than rubber. It's great while it's new, not so much after a few decades.
Rubber can be OK if it's properly vulcanized (eg tires) but that's a chemistry adventure. Also interesting to see some youtubers have no idea whatsoever what 'vulcanizing' means.
Silicone rubber varieties seem to be very stable, so I'm interested in seeing what's available in 2-part silicone, for molding. Like I said, machining a mold is no problem, and I also have vacuum pump(s) for bubble removal from a mix.
Another possibility is rollers from old office machines. I have accumulated a large shopping bag full of mixed rollers from stripping street-tossed machines. Didn't think there was much chance of finding something size-suitable and silicone rubber, but was surprised when I looked through the bag.
See pic. Found one just about perfect, so long as I can cleanly cut it into sections for the 4 rollers. Would need to make new hubs, with lips to prevent sideways creep.
But of course that's a one-off. I'm still going to try for a generic, repeatable solution that works for any future pinch roller needs.
Edit to add:
drussell, do you know of any instruction texts on heated press vulcanizing?
BurningTantalum & tpowell1830, yeah, machining rubber. I had figured some kind of high speed abrasive or die-grinder setup might work. I did know of liquid nitrogen freezing, and do have a large dewar. Otoh I try to avoid anything with high consumables cost, which this would have. Still maybe...
It would probably work with silicone rubber too.
An amusing variation:
http://everist.org/NobLog/20170321_machining_pool_noodle.htmAnyway I'm going to try molding first.
Muttley... oh god, no. If I used those, every time I look at the card reader an image of gangrenous testicles would come to mind. Besides, latex is used because it rapidly decays.
CatalinaWOW, yah, if I try this I'd go for fat O-rings, to get enough compression range. Also cheap O-rings have pretty crappy durability, so maybe Viton Fluoroelastomer, or neoprene.
As for cards, someone on cctalk has been generous, so I have a smallish stack on the way here. Plus I have an IBM 026 card punch (still to restore) and a couple of boxes of blank cards. But still looking for more. Both new and punched. A several inch thick stack of punched cards with known data content (as a file) would be ideal.
Ha, someone I'm corresponding with says he recently threw out some boxes of punched cards. I'll mention your "prices were 5-10 cards for $10 US", see if he will kick himself even more.