Don't know if we're talking about the same thing,but CrystalDisk has a entry on my C400-
BD-Factory bad block count-Raw value-52
He said he's running the Linux "badblocks" tool. Per the output, there are (at least) 32 unreadable sectors.
I've always assumed NAND has spares to replace bad blocks at manufacturing level.
Yes, it does. But these are dealt with internally; if bad blocks are visible at the user level in a write-read-verify test (what badblocks does), the drive has a problem. A new drive should never have trouble coughing up data which was just written to it.
What's the brand?
The described behavior would be incorrect regardless of brand.
Is the capacity right?
It's certainly worth verifying the drive info and health as reported by SMART, to see if there are any anomalies. Since OP is apparently using Linux: smartctl -a <device>
(you may need to install an additional package from your repository first, e.g. in Debian/Ubuntu you need "smartmontools").
I would also run the same test with a known good drive connected to the same port and power connector, to rule out system level problems.