Date: Fri, 11 May 2001 16:50:52 -0600 (MDT)
From: Craig Loomis
Subject: Re: 2.5m obslog 20010510 (MJD 52040)
  Summary: we have three problems with the Paralans.
  1) The Paralans are supposed to be powered up before the associated SCSI
devices (i.e.  the Photometrics A2S converter and the Mac.) In other words,
the SCSI channel provided by the Paralans should be in place before the SCSI
devices connect to it. The power-up order may not alway be a problem, but
should be kept in mind.
  Both the engineering camera cartridge and the spectro guider box use a
single power switch to power both the Paralan an the A2S. This should be split
up or a delay for the A2S added.
  2) The Paralan tech. insisted that the allowable loss over the fibers is
only 2 dB. This seems surprisingly low, and is lower than the pretty high
2.5-3.0 dB loss that we have on that run.
  An unnecessary patch cord on the cartridge has been removed, and I can
remove another in the computer room after I order a long ST-SC patch cord.
With some cleaning, I think we should get near/under 2 dB.
  3) The Paralan tech. also stated that some generation of FX-50 converters
was prone to over-heating (due to an improper socket on the card?) While
testing today, we noticed that a cool Paralan always established a FO
loopback, but that warm ones didn't. Furthermore, the signal measured on the
TX fiber was always -15-ish dBm on powerup for cool Paralans, but was at
-60-ish (i.e. no signal) about half the time for warm ones.
  We moved the Paralan in the computer room away from heat sources as a quick
fix. All our devices were purchased together, so they're all suspect. I'll ask
for identification tips from Paralan tomorrow, and will plan on rotating them
back to Paralan for repair.
  Finally, I'll urge that the "guider" (either the engcam or the spectro
guider or both)  be added to any list of systems that must be checked out
during the day, especially during shakedown. "Checked out" would now mean
successfully executing a "gcam doread" and getting a plausible image on the
'host disk. The systems are pretty reliable when left running, but any
swapping/power-cycling makes them completely suspect.
 - craig