I've just gotten back to working with our KE237 sourcemeter. I just discovered a couple of new bugs in the LV drivers posted here and at Keithley. It's a pretty unique old instrument, and we're lucky that they still make it, but it has some rough edges. The drivers are a free download, so when we find bugs, the best we can do is tell Keithley and share what we know. Caveat emptor. Beats starting from scratch.
In the Keith 237 Basic Config vi, The enumerated type control for integration time has the text for 60 hz line cycle, and 50 hz backwards. This is hard to spot, since the instrument only displays "LINE CYCLE INTEGRATION" on the front panel once this has been programmed. Happened to look at the actual string going out to the instrument and noticed it.
Easy to fix, just change the enum text labels for selection #2 to read 60HZ, and #3 to read 50HZ. So, if you've noticed that your instrument seems noisier running under Labview, this might be why.
While I'm at it, I also found a non-optimum feature of the SMU Device Errors query vi. It will overwrite any upstream error messages in the LV error cluster. For instance, if you call it twice along the communication chain connected by the error cluster, if it finds an illegal instruction error in the instrument in the first call, since the first call cleared the error condition in the unit, the later call sees nothing wrong and wipes the error cluster clean, even though the unit wasn't successfully configured. i sprinkle these error queries through my apps pretty liberally.
Easy fix is to add an outer case structure around what's already there to simply pass through upstream errors with no new action taken.
