fiber optic transmitters and receivers

yarisboy

I'm working on the BOM for a propeller application board. I need good noise/EMI immunity on this application. Does anyone out there have good recommendations for fiber-optic transmitters and receivers I can mount directly on the propeller board? Do my fibers have to custom made and of equal length? Do they make fiber-optic ribbon cable yet? Is this cost prohibitive?

localroger

Fiber optic hardware ranges from the very cheap where short range and low bandwidth are needed to the very expensive where long runs, high bandwidth, and environmental durability are needed. The least expensive route would be to use raw fiber strands, which should be fine for runs up to 10-20 feet at the bit rates you get from fullduplexserial. You will need dedicated strands for transmit and receive. They make multi-strand cable but it's very expensive. For each link, you will use a LED at one end and a phototransistor at the other. To make the connection, you can get expensive kits or you can drill a hole the diameter of the strand through a small piece of wood or opaque plastic, then drill a blind hole the diameter of the LED or transistor from the other end, stick the transistor or LED in that end and the fiber in the other. I have also heard of people using heat shrink tubing, but it won't shrink enough to grab a raw fiber strand. You could also drill a small blind hole in the LED or transistor itself the diameter of the fiber and insert the fiber.

T Chap

Here is a company that offers xmitters and receivers that are easy to use.

www.i-fiberoptics.com/

You can get some of their parts off digi:

parts.digikey.co.uk/1/1/91494-led-fiber-optic-660nm-red-if-e96.html

search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=FB133-ND

In practice you can use one single fiber for both xmit and received if you use a decent splitter on each end. The light going one wasy doesnt care that light may be going the other way. On a decent splitter, the back flow from the same sides xmit is often low enough to not interfere with that sides receive. If so just add another strand and have isolation.

yarisboy

Things have come a log way since I worked for the people we don't mention. Back then if someone cut a fiber you had to call in a tech to bond on a new sapphire. My applications will always be less than 20'. I'd prefer a fiber with a bonded on coating for durability. I don't really want be directional communication on any fiber. The whole point of going to a prop is to be executing simultaneous interdependent processes with immediate redundant feed back with fast error checking. Do many of the available objects deal with multiple PID loops and fuzzy logic auto tuners?

Beau Schwabe

yarisboy,
I don't know if this will help or not, but if you wanted to use a standard pair fiber·cable then this might work for you...
One of the first things I did when I worked at National Semiconductor was to design a Fiber-Optic Ethernet card based off of an existing Ethernet over copper design.· The Fiber-Optic connector used was an HFBR-53D5.
·
I believe this connector has·auto negotiation·to determine if there is·an optic fault in the connection.· Some of the connectors that were used could detect the RX/TX polarity and automatically switch if necessary.· Because of the bi-directional nature the connectors could also switch between normal (full-duplex mode) or High speed (half-duplex mode).· Sorry, I don't have a part number for those connectors but you could probably do some digging around at Agilent Technologies.· I would suspect that other bi-directional·fiber connectors·would also have the ability to pair up to use both of the fiber ports as ... TX/TX·... RX/RX ... or ... RX/TX· as deemed necessary.
·
Reference:
http://www.datasheetcatalog.org/datasheet2/4/081a2czpcl7aawqot831lhlw6yky.pdf


▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Ale

Sorry to revive this thread but I was looking at exactly this for a couple of days and wondering what if we use this TOSLink connectors ? and ready to use and cheap fibre... the receivers are cheap and transmitters are not that cheap [noparse]:)[/noparse].
Any thoughts on this ?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Visit some of my articles at Propeller Wiki:
MATH on the propeller propeller.wikispaces.com/MATH
pPropQL: propeller.wikispaces.com/pPropQL
pPropQL020: propeller.wikispaces.com/pPropQL020
OMU for the pPropQL/020 propeller.wikispaces.com/OMU
pPropellerSim - A propeller simulator for ASM development sourceforge.net/projects/ppropellersim

Clock Loop

Using an LED, I transmitted stereo 44100hz PCM SPDIF data through a standard audio optic cable found at any electronics store.

You can even receive the data using the LED AS BUTTON(detector) approach.

Ale

Can you please clarify what "led as button" means ? do you mean a photodiode ?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Visit some of my articles at Propeller Wiki:
MATH on the propeller propeller.wikispaces.com/MATH
pPropQL: propeller.wikispaces.com/pPropQL
pPropQL020: propeller.wikispaces.com/pPropQL020
OMU for the pPropQL/020 propeller.wikispaces.com/OMU
pPropellerSim - A propeller simulator for ASM development sourceforge.net/projects/ppropellersim

Clock Loop

With RCTIME(waitcnt), you can use a standard LED as a photodetector, and a very fast one at that.

So you could use a single optical link between two props, using the LED's as both transmitters and receivers.

http://forums.parallax.com/showthread.php?p=589699


I use LEDS as buttons in my BlackBox audio sequencer.

http://forums.parallax.com/showthread.php?p=831833

the LED code looks like this:

PUB ReadLed (n, p)

   '---READ led-----
  dira[noparse][[/noparse]n] := 1             ' make n pin an output   

  outa[noparse][[/noparse]p] := 0            ' p - low      'reverse bias led        
  outa[noparse][[/noparse]n] := 1            ' n - high     'reverse bias led 

  dira[noparse][[/noparse]n] := 0            ' n - make n pin an input
  waitcnt(readwait + cnt) ' wait for pin to change state   
  ledstate16[noparse][[/noparse]n] := ina[noparse][[/noparse]n]      ' read state of n pin

A proper driver would need to be made to use the led as both a transmitter and receiver, but it could be done with not too much effort.

The voltage decay on the LED's N leg takes time depending on how much light is falling on the DIE of the LED.
This time, and voltage change is slow enough that the prop can detect it, and make use out of it.

Another thread about this same thing.

http://forums.parallax.com/showthread.php?p=856132

Post Edited (Clock Loop) : 7/15/2010 5:41:46 PM GMT

Beau Schwabe

Clock Loop,

What speeds can you get?

I was doing some experiments with standard Photo-NPN transistors and was only able to get about 500kHz before my signal started to drop off. It was in open air though, a direct fiber link might make a better difference and allow for a higher communication speed. The problem was that the C-E parasitic capacitance was becoming a dominant filter. I wonder what limitations the the PN junction on the LED would have with parasitic capacitance effects.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Clock Loop

Beau Schwabe (Parallax) said...
Clock Loop,

What speeds can you get?

I was doing some experiments with standard Photo-NPN transistors and was only able to get about 500kHz before my signal started to drop off. It was in open air though, a direct fiber link might make a better difference and allow for a higher communication speed.

I suppose i shouldn't say "and a very fast one at that."

Because I have not tested a led as a receiver. (except for the use of buttons)

I have only used a led as an SPDIF transmitter. (spdif runs at 3.1Mhz)


I am curious now, how fast a standard led CAN work at as a receiver.

Using a clear led might help too.

Using a "luxeon" "rebel" led might result in some BLISTERING speeds. (for a led)

Post Edited (Clock Loop) : 7/15/2010 8:31:30 PM GMT

KenBash

I recently developed a fiber optic sensor for the medical industry. Using fibers just off the spool, one thing I discovered fairly quickly is that the way your ends are CUT, alignment, ( and focus distance if going into a coupling lens) can have a HUGE effect on the amount of light that gets in and out of the fiber. ( this could easily affect the signal/noise ratio on a transmitter/receiver pair. ) If using a plastic fiber, the polishing/alignment is a bit less critical, but still needs to be paid attention. ( you can't expect as much out of plastic either) If using glass fibers, you need to use a diamond cleave and get a nice flat break. I found that cutting the ends with scissors or clippers and testing the output, I got only 1-5% of the light in and out that I got using the diamond cleave. The actual process of gluing and doing an optical polish is a little bit art and a little bit science, but I found that if I was careful with the diamond cleave, I got enough signal through the fiber not to need to do a full polish.

The main point here is be sure you read up a bit on optical coupling if you are going to do it yourself or stick to pre-made fibers if in doubt. It would be a shame to have a great project not work because of something as simple as how the ends of your fibers are cut.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
" Anything worth doing... is worth overdoing. "

··············································· ( R.A.H. )
····································

rpdb

I found some 3COM SuperStack network switches with a type LC (Lucent Connector) transciever module in the back for less than $5 a piece at Wierdstuffwarehouse.com. They are good for 100Mbs. You should be able to find a pair of surplus network card or switch with similar transciever for less. These switches have slow 10Mbs copper ports on the front·and most companies are upgrading.

The transciever modules have CMOS level I/O and are relativly easy to harvest.

Buy a LC patch cable is the easiest cable solution. Other wise you need a scribe and buffer stripper.

Belden.com makes a pre polished connector with index matching gel in the back side. You scribe (barely scratch) the fiber after stripping off the coating and buffer with a buffer stripper (kind of like scoring glass before you snap it). You pull straight up (not side to side) on the fiber and it (hopefully) propagates the scratch straight across. Get a laser pointer and point it into the far end of the fiber, point the end you just cleaved at the ceiling. Look at the side of the cleave, ie 90 degrees from the ceiling while slowly spinning the fiber. If you cleaved it correctly you will NOT get a "Light House" effect ie any light from the sides. You put the fiber into the gel side of the connector and slide a retaining tab and your done.

Check out the connectors at Belden.com, look for "FiberExpress Brilliance Connectors". They have a promotion going on right now and will send you a free sample with fiber and two connectors for one cable. You will need two cables for Tx and Rx so get a fiber buddy.

Don't stare at the laser light and becareful with the glass fibers.

Post edit:As long as the length does not differ between the tx/rx by 5% on a short run (less than 2000 meters you should be ok. This is done on "multi-mode" fiber in a "campus" type setting with LED xmitters and photo diode recievers. For longer "hauls", single mode fiber with laser light source is used with PIN diode reciver for distance of 40 to 60 Km.

Post Edited (rpdb) : 7/16/2010 2:45:13 AM GMT