I recently used fibre on a hi-rel project (had a simple light/dark relay logoc link) and was surprised to find the latter receiver (Agilent 1402) singing at UHF when anywhere near its threshold (as with dirty fibre connections!). Too late now, but in retrospect the TX/RX pair needed a run-time check that sensed the high-low thresholds and reported a bad 'eye' margin. That would catch assembly and cabling errors. Couldn't add it because we only built one end of the link.
Just patching a noise filter onto the RX wouldn't alert that a link had become weak.
Next time: change the logic link to use a modulation protocol for link-confidence checking (and monitor errors on the Gig E links as well).
we were the HMI company - you'd think the System Integrator (who did drill rig build and program for a living) would know this - not their first fiber link. TG it was plastic fiber; ease up on the clinch and bits come thru the hose!
When you introduce unfamiliar technology into an industry then it is absolutely necessary to educate on its quirks and limitations.
For example, boatyards are quite familiar with tying down cables and marine cables are designed to be very robust. If you give them something new that looks like a cable then expect them to treat it exactly the same.
anything that can go wrong will.
had a high power coax, you know the sort of thing 5cm diameter.
all went well for ages, till one day, full power and a bit was applied.
system failed 'spectacularly' after a while. turned out 'pipe' had been flattened 'just a little' to get it through a gap. aparantly standard practice on ships for pipes, but the dialectric of the coax did not agree.
moto. don't get high power rf 'pipes' installed in the ship yard by pipe fitters, even if they look and feel like pipes.
Good plan, William.
Long ago I needed a drawing made for attaching a sheath for an optical fiber to the metal tube extending from a single-mode LED assembly. My initial sketch had the note "Do not use excessive force to attach the sheath."
The drafting department laughed and said the note was not necessary. I was into the next project and did not need to waste time with an argument, so reluctantly I said they could eliminate that note.
You can probably figure out what happened. About a year later, LEDs started failing in the field. The metal tubes had been bent by excessive force by the factory workers while attaching the sheath.
After that I never let anyone tell me what or what not to put on my drawings.
It is amazing how poorly things can be made to work when the folks building things don't do it correctly. Sometimes they assume that the engineer was wrong and do it some other way, and then when it fails to work correctly the engineer gets the heat applied.
So for one project I had all of the released drawings stamped "System will not function correctly unless built according to the drawing".
I did get a few snide remarks, but when built to the drawing it worked as desired.
Blog Doing Math in FPGAs Tom Burke 24 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...