High speed communication "system"

[Gruber2]

Hello.

Quick idea : I am trying to find a way to do send a defined data to a optical output in very fast ( in general ) speeds.

the idea detailed :
i have a laser diode and photodiode ( coupled together ) capable of transmitting/receiving up to 20 nS optical signals.
The data i want to send is around 1k ( or some small amount ) and it will be changed from time to time to see the dynamic characteristic of the optical systems ( i.e. a regular PWM wont work,because it will be transmitting everytime the same consecution of signals ).

The idea of the project is to "test" the optical system ( the actual system of the laser diode,photodiode and lens is already done ). The system itself is complicated mostly from lens' point of view,but its not the my "job" to take care of it

The mentioned speed of 20nS is not needed,i am aiming at 500nS signals being send to the laser diode.The receiver's circuit is not my "problem" too at the moment The fastest achievable will be OK so no minimum/maximum speed is defined ( although it is highly recommended for me to be under 1uS )

The idea i have in mind is to use a pic18f entirely dedicated of storing the data into its RAM and sending it to the communication port. When the system is "initialized" the tested data is send it into pic's RAM ( or EEPROM ) then on external signal it starts sending it in a sequence of 1 and 0 into the communication port of laser's diode.
I am worried if the pic will be able to achieve ~500nS output signals even if its entirely dedicated to job ( i.e. no LCD/buttons/other stuff is operated by it ).
I hope by using the pic for "read from ram -> send to port x data" will be able to achieve 500+nS signals,although even using fastest 40Mhz PIC will do around 250nS at instruction time, which might be slow if i have to implement alot of reading/sending loop.

I guess the "idea" doesn't sound quite logical and it has big flaws, but this is the result of combining optical stuff and microcontroller into one thing...

I am wondering if a CPLD with some logic will be able to read the data ( the question is from where ) and to send it into the output,while the pic/mcu will provide storing the data into the temp location, starting and stopping the sequence,etc. I cant think of an idea how such fast in general logic will be able to read the data in consecution and then send send it "as-is" to the output.
Maybe i have to say that the data wont be needed to be processed in any way - i.e. all it needed is to be read and send to the laser diode. Thats why i hope that fast microcontroller or CPLD will be able to do the job.

Any idea or suggestions will be greatly appreciated
Unfortunately the idea seems crazy and there are alot of questions - but its normal for out university - nobody knows why something is needed,but it has to be done somehow

 

[Papabravo]

I think the PIC has a relatively small chance of success for this application.

I used to build network communication testers that had to test the physical layer before the ASICs reached first silicon. I even have a patent on one, #5,289,474. I used FIFO memories. They were loaded from a byte wide interface on a PC. Upon command they would transmit their contents up to 1K, and could be commanded to transmit the same data repetitively. It was all controlled by a finite state machine. It was slicker then green stuff on a doorknob.

 

[RadioRon]

It was slicker then green stuff on a doorknob.

I hadn't heard that one before. Out of curiosity I googled this phrase and would you know, there are many variants! These include:

slicker than snot on a brass doorknob
slicker than buggers on a glass doorknob
Slicker than Greases Owl ****.
Slick as Snot AND "Colder Than a Witch's Tit in a Brass Bra" OR "Colder Than a Well Digger's Butt
slick as owl sh** in the moonlight
slicker than deer guts on a door knob
slicker than a goose booger on a wet doorknob
slicker than WD 40 on a doorknob
Slicker than a sycamore sapling
Slick as a cow pie on a flat rock


and I just couldn't resist also adding this one, which has nothing to do with things being slick:

I've been to two World's Fairs and three goat-f**kings and I've never seen anything like that!

 

[Roff]

I hadn't heard that one before. Out of curiosity I googled this phrase and would you know, there are many variants! These include:

slicker than snot on a brass doorknob
slicker than buggers on a glass doorknob
Slicker than Greases Owl ****.
Slick as Snot AND "Colder Than a Witch's Tit in a Brass Bra" OR "Colder Than a Well Digger's Butt
slick as owl sh** in the moonlight
slicker than deer guts on a door knob
slicker than a goose booger on a wet doorknob
slicker than WD 40 on a doorknob
Slicker than a sycamore sapling
Slick as a cow pie on a flat rock


and I just couldn't resist also adding this one, which has nothing to do with things being slick:

I've been to two World's Fairs and three goat-f**kings and I've never seen anything like that!

Snort!!!

 

[Papabravo]

So sue me for having a sensitive side and wanting to be less offensive then you two neanderthals.

ROFLMAO

 

[Gruber2]

thanks for the offtopic guys

I am thinking of getting pic/atmel filling external RAM, then a CPLD to read off the memory and using parallel to serial internal logic to send into the output.

That seems to be one of the fastest way to get the data at the laser diode under 1uS

 

[Papabravo]

thanks for the offtopic guys

I am thinking of getting pic/atmel filling external RAM, then a CPLD to read off the memory and using parallel to serial internal logic to send into the output.

That seems to be one of the fastest way to get the data at the laser diode under 1uS

My post was on topic. The control logic will be much simpler if you use a FIFO. Trust me on this. If you use an ordinary RAM chip how will you mediate two things trying to access the RAM at the same time?