More fixes today:

• reseated Bulgin/Bulgin connection (missing a purple ring) for rad.20m.rel
• changed trh.8.5m.rel to RS232 mode (rather than RS422)
• replaced CSAT3A at 6m.init with CSAT3A from 1.5m.a2.rel
• replaced CSAT3A at 1.5m.a2.rel with a CSAT3, requiring a new cable, new mount to be invented, new serial port in RS232 mode.  Note that the new mount puts the path about 5cm higher and 3cm further north.
• installed a new sensor for Tsoil.grass.uconv
• also tried to organize cables at rel.a2 and grass.a2, but am not really happy with the job I did.  I really should have started by disconnecting the ethernet cables.

I <thought> that fixed all sensor issues, but now see a dead CSAT3A at 1.5m.rel.  I can't win and the stupid CSAT3A's keep on dying and Campbell hasn't fixed the batch of 5 or 6 bad ones that I shipped to them a month ago....

I still don't know why pmons at uconv, P3 and P5 aren't working.

gps.uconv2 is a known problem that Gary has worked around.

Dry run IOP just starting now (9pm).  The PI team is pretty excited, and so are the mosquitos.

@Gary: I'm not sure I did the best change-over of the sonic.1.5m.rel2.  I changed the port, (usb1 → usb5), id (1010 → 1016), and name (.a2 → .a2n), which made ck_xml happy, but now we have another set of variable names.  I <thought> there was a way that we could, with one config, process both the data up to now and the data from now, into the same variable names.