Digital Communication question / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Digital Communication question

1 2 next

Move Thread

LAN_403

HV Enthusiast

Fri Jun 23 2006, 11:12PM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

Had a question regarding digital communcation . . .

I have a fiber optic communcations link which i'd like to send a serial stream over. On the transmitting end, I have a 12-bit A/D and on the receiving end a 12-bit D/A. Note the fiber optic link may contain up to 3 lines for the serial data communication.

Is there a way to have this interface work without the need for a microcontroller etc... For example a dedicate off-the-shelf chip(s) which would transmit and receive this information serially?

Thanks

Sulaiman

Sat Jun 24 2006, 01:41AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

Have a look at Maxim MAX170 and MAX543 12-bit serial adc & dac

I've not tried, just looked at them.

ragnar

Sat Jun 24 2006, 01:49AM

Registered Member #63 Joined: Thu Feb 09 2006, 06:18AM
Location:
Posts: 1425

A kit like this might be a quick + dirty solution:
Fibre optic communication kit (AUD$45.95)

"It contains all the components needed to build a fibre optic data link, both transmitter and receiver, including one metre of fibre optic cable and quality double sided plated-thru PCB's. A 26 page booklet is included which gives you assembly instructions, theory..."

"When you have built this kit you will be able to transmit and receive high speed data in serial format across one metre of fibre optic cable.
- The transmitter is powerful enough to actually transmit up to 16 metres."

etc ^^

...

Sat Jun 24 2006, 02:39AM

Registered Member #56 Joined: Thu Feb 09 2006, 05:02AM
Location: Southern Califorina, USA
Posts: 2445

You didn't say anything about bandwith, so I will assume the fiber link's is >>>a/d's

The first thing to come to mind would be a shift-out register

Carbon_Rod

Sat Jun 24 2006, 05:21AM

Registered Member #65 Joined: Thu Feb 09 2006, 06:43AM
Location:
Posts: 1155

EVR this thread sounds a bit like an echo of an old.4hv.org thread.

However, as before I was unclear about the reasons why 3-wire and 2-wire busses cannot be so easily interfaced I will try to explain. The 2-wire communication protocol uses a bi-directional data line. You would need at least one micro controller and some sort of primitive encoder/decoder to get this to work. The 3-wire SPI specification usually uses a packet structure with an addressing scheme and cyclic redundancy check. Again, a minimum of one micro controller is needed even to enable the a/d or d/a chips internal configuration.

The solution is simple if you canâ€™t program PLDs:
1.) Use the SPI emulation Printer port hack
2.) Use the I2C RS232 hack
3.) Get an SPI protocol dongle adapter for USB (kind of buggy to program as they usually come with VB libs)
4.) Adapt an existing RS232 based 12-bit data logger to use 2 optical communication lines.
5.) Ask someone to program a streaming data logger for you (a $5 chip has a multiplexed 12bit a/d unit built in.)

If you don't have a programmer:

a.) What is your intended use? Temperature/Size/SOIC/DIP
b.) Where are you located?
c.) Do you have a Paypal account?
d.) Do you have a FedEx account?

The US requires a confirmed residentsâ€™ information on the shipping label for import of certain electronic devices now.

Cheers,

Bored Chemist

Sat Jun 24 2006, 07:40AM

Registered Member #193 Joined: Fri Feb 17 2006, 07:04AM
Location: sheffield
Posts: 1022

Just a thought. If the data rate is low and the distance is short then you can transmit 3 signals down a fibre link using a multi colour LED and beamsplitters/ filters at the other end.
12 bit resolution is going to be a challenge for anything other than pretty slow data.
I'm sure that providing us with more information wouldn't just satisfy our (or at least my) basic nosiness, but might also help with suggesting a solution.

Carbon_Rod

Sat Jun 24 2006, 09:28PM

Registered Member #65 Joined: Thu Feb 09 2006, 06:43AM
Location:
Posts: 1155

Well itâ€™s a good idea â€“ only 2 problems:
1.) Most fibre optic communication line is not multimode (usually just infrared.) If he were using multimode acrylic plastic fibre then this would not be an issue.
2.) LEDs are not very fast or monochromatic.

IIRC there is some high-end Audio products that have a low power optical transceiver that looks like a stereo jack. Perhaps some sort of hybridization would remove the need for A/D and D/A chip programming. Also, 3m makes optical gigabit network cards and reasonable plastic fibre.

Again, a context for application would be required in order to make a decent recommendation.

Multimode acrylic fibre communication lines do exist though â€“ see below:

...

Sat Jun 24 2006, 10:30PM

Registered Member #56 Joined: Thu Feb 09 2006, 05:02AM
Location: Southern Califorina, USA
Posts: 2445

Um, you can shine just about any color of light down a singlemode fiber... I have gotten plenty of green for a comunications link through 1550nm singlemode fiber. Just because LED's are not monochromatic it does not mean that they can be used for communication, but if does limit the number of channels that you can have in a link (you would probably have to limit yourself to ir, red, and green possibly a blue and yellow...). Led's can also be run well into the khz (atleast) before they run into problems.... It wouldn't surprise me if you can run them in the mhz ranger either.
If you actually read the first post it sounds like EVR had a link that gives up to 3 channels of data, and just needs a way to compress the 12 channels (and probably a clock or 2) of data from the a/d into a serial stream, then uncompress it back into 12 parallel channels on the other side, preferably without a computer in there-just like in the old days...

BTW; feeling lucky EVR with your 777 posts

Carbon_Rod

Sun Jun 25 2006, 04:44AM

Registered Member #65 Joined: Thu Feb 09 2006, 06:43AM
Location:
Posts: 1155

â€œUm, you can shine just about any color of light down a singlemode fiber... â€œ
Um, no you canâ€™t for the Kevlar reinforced acrylic-coated glass fibre -- I know because I have tried making compatible diagnostic hardware for the transceiver plug sets. They have a specific range for efficient operation.

â€œI have gotten plenty of green for a comunications link through 1550nm singlemode fiber.â€
Likely it is the 100% acrylic fibre of which you speak. (See picture above for an example.) 3m uses the multimode fibre.
(

)

â€œJust because LED's are not monochromatic it does not mean that they can be used for communication,â€
True, I was referring to BC suggestion of using a beam splitter would mean having overlapping noise bands would be counterproductive.

Holographic splitters are quite common for the $40 000.-- backbone hubs.

â€œIt wouldn't surprise me if you can run them in the mhz ranger either.â€
Anyone who has built variable PWM output circuits for controlling LED based displays could answer that assertion with a no. However, you are correct that normal LEDs will function into the kHz range depending on the type. An IRDA transceiver and driver chip may be around though (they are the cheapest units for optical communication.)

â€œwithout a computer in there-just like in the old days...â€
With or without a computer is irrelevant â€“ a MCU of some type will be needed. Besides in the old days one built the PLD too if it was necessary â€“ but since parallel busses were all the rage it was easy to build a latching buffer and use a UART based chip pair.

â€œIf you actually read the first postâ€
I did â€“ even the last time this was posted too.

Cheers,
(If EVR was more specific and owned a programmer-- In the time it took to write this EVR could have a workable chunk of code.)

Dr. Slack

Sun Jun 25 2006, 07:04AM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

The easiest physical way to arrange things will be to have a single fiber, then there are no worries about possible skew between data, clock, strobes etc. There are a whole bunch of chips going under the general name of SerDes (serializer-deserializer). Look on the National Semicinductor site

, the DS92LV16 takes 16 bit parallel data at a clock rate between 25MHz and 80MHz, serailises, adds enough redundancy for the clock extraction to work at the Rx end, and spits it out on LVDS at several Gb/s, which is a good standard for interfacing with fiber diode sets. Is that fast enough, or way too fast?

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob