There are cases where a bad checksum is generated, or created, and it is consistantly bad. This can be due to (poor) multiplexer design, or other reasons. While it would be best if the bad checksum were not being generated, as long as the data values are plausible, and agree with other measurements (or common sense), it's probably OK to use, at least until the checksum problem can be properly solved
I will probably be posting
the new version of NavMonPc tomorrow (Sunday) -- I decided to write up a User's Guide Supplement, since there are several significant changes. The Checksum Repair option works well.
I'm not recommending blindly using data with bad checksums, but if you've got a solid hookup that otherwise works reliably, I think you should be able to use the data for a while. If you want to worry about errors, you can always have bogus data with perfect checksums. In fact, I think that's a much more likely scenario.