RS232 Fireoptic Link (was FSK...) / General Science and Electronics / Forums

Forums

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

« Previous topic | Next topic »

RS232 Fireoptic Link (was FSK...)

1 2 next

Move Thread

LAN_403

Extreme Electronics

Sun Jun 28 2009, 08:08PM

Registered Member #74 Joined: Thu Feb 09 2006, 09:17AM
Location: Nottingham UK
Posts: 99

I need to create a short distance (~10M) RS232 link over fibre optic to send and receive a data rate of 115Kbaud (MAX). for Tesla Coil Control and monitoring.

Im looking to use cheap fibire optic transmitters and receivers using the 1mm SFH756V FIBRE OPTIC EMITTER SFH250V FIBRE OPTIC DETECTOR

Both of these devices will permit the use of a 500Khz or higher carrier which I will frequency shift and demodulate with a VCO / PLL (CD4046 is my thought at the moment)

I'm looking for any Ideas, suggestions, IC's or circuits that my be useful, especially on the receiver/demodulator side.

Thanks Derek

Carbon_Rod

Mon Jun 29 2009, 02:48AM

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

irda with plain acrylic fiber may work over short distances.

Extreme Electronics

Mon Jun 29 2009, 07:57AM

Registered Member #74 Joined: Thu Feb 09 2006, 09:17AM
Location: Nottingham UK
Posts: 99

I'd thought about an IRDA Enc/Deenc, but I couldn't find a chip that wasnt attached to a IRDA IR Trancever which is not good for firing into a fibre.

Following your link gave me a referance to the MCP2120 which looks ideal, assuming I can find a supplier.

Thanks for the link...

Derek

Steve Conner

Mon Jun 29 2009, 10:59AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

You might want to check out the Wavefront ADAT transmitter/receiver chipset. These look like they could send any Tesla coil control signal you ever wanted down a fibre optic... RS232, interrupter pulses, MIDI, PCM audio, all at once. They work with the Toslink fibre optic parts.

Also have you considered a Bluetooth RS232 dongle? The Bluetooth frequencies might be high enough that the Tesla coil doesn't jam them.

Extreme Electronics

Mon Jun 29 2009, 08:01PM

Registered Member #74 Joined: Thu Feb 09 2006, 09:17AM
Location: Nottingham UK
Posts: 99

Steve,

I can't seem to find much information ADAT apart for ma interesting amatur radio design... Do you have any links ?

RS232 will be fine especially if I can acheive 115Kb, At present I cant get teh data from the coil back fast enough at high breakrates. Even averaged over a burst.

I don't really want to start adding on to the processor any software to accept Ethernet, IRDA stacks, Bluetooth Stacks or any other complex protocol. I would prefer to keep the device simple (ish). Rs232 is supported in hardware, Easy.

With Bluetooth, my worry about any form of RF is the noise issue. Even Bluetooth has to be mixed to lower frequencies to be decoded. Even Fibre could be fun. Much of my desire to get a proper RS232 fibre driver is that at the moment I can take out a USB RS232 converter at 2+ feet with an 8" strike to earth. Even with the isolation I already have.

I like the idea of an optic link, its contained. I picked FSK so I could have a simple coil shutdown in hardware if the RX control carrier was lost. So If you are really in trouble, just pull out the fibres. Even if the processor is off playing solitare...

Derek

GeordieBoy

Mon Jun 29 2009, 09:07PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

All sounds okay, but you might have to use an FSK carrier frequency significantly higher than the baud rate. I can't remember the rule now, but you can treat the two carrier frequencies for binary '0's and '1's like two Amplitude Modulated carriers. The FSK spectrum is the sum of these two AM (or ASK) spectrums each centred around it's own carrier frequency. As long as there is not too much overlap your PLL should be able to seperate the symbols with a minimum of error rate.

Obviously if the carrier frequency isn't high compared to the bit rate, then it will be very hard to see any change in frequency in the carrier over less than a complete cycle. In this case PSK might work better.

-Richie,

Extreme Electronics

Mon Jun 29 2009, 10:40PM

Registered Member #74 Joined: Thu Feb 09 2006, 09:17AM
Location: Nottingham UK
Posts: 99

Richie,

I haven't had too much time to check the specs, but I can go up to (at a push) 1Mhz carrier. So I was thinking something like a 500Khz Carrier for '0' and a 700Khz Carrier for '1' This gives me >5 cycles to get a PLL lock.

I need to brush up on my transmission theory and to look in more detail at the MCP2120. If I acheive only 56K Ill be happy, its still 3 times faster than what I have at the moment.

Extreme Electronics

Sun Jul 19 2009, 12:14PM

Registered Member #74 Joined: Thu Feb 09 2006, 09:17AM
Location: Nottingham UK
Posts: 99

So finally had time to sit and play. I've gone down the MCP2120 IRDA route.

Quickly adding in a MCP2120 into the PC end of my existing un-encoded RS232 link and looping back the fibre. Nothing. Spent about 2 hours trying to work out why. I then realised that IRDA is a half duplex protocol and there is a hint in the MCP2120 that the chip only works half duplex. So I built another board with two MCP2120's one for RX and One For TX. I managed to get looped back comms over a 1M fibre.

After upping the pulse current on the transmitter and some tweeking with my photodiode and crude MC4049 amplifier, I have now got a 115.2Kb data link over 3M half duplex to the Tesla coil with intermittant errors.

I am convinced that these errors are caused by my crude photodiode amplifier so my next step is to buy a combined phototransistor and amplifier. this has two benifits.
1. Its likley to work properly.
2. It should be much more tolerant to noise.

The MCP2120 has two downfalls in this app. The half duplex means that I need 4 chips to get full duplex, although I may not bother, most of the data is command/responce in nature so shouldn't be too much of a hastle.

The other problem is I was looking at FSK so there would be a carrier. The Idea was that lack of carrier would shut down the HV power of the TC in an emergency. If all else fails pull out the fibres!!

Lack of a carrier means that I will need to devlop some PC software that will give me a blank byte every 500mS or so as a 'keep alive' for the TC, which is a bit of a pain as at the moment I can run with a standard RS232 terminal emulator on either a PC or Linux Box.

Never mind it works for now.

Derek...

Carbon_Rod

Mon Jul 20 2009, 08:36AM

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

Extreme Electronics wrote ...

Never mind it works for now.

Derek...

Sounds great,
note long run fiber-optic communication generally has all sorts extra stuff to improve noise rejection.

In this case it may help reducing the baud rate, ignore flow control, and add a CRC check to the keep-alive bytes.

Cheers,

Extreme Electronics

Mon Jul 20 2009, 09:49AM

Registered Member #74 Joined: Thu Feb 09 2006, 09:17AM
Location: Nottingham UK
Posts: 99

Cr,

Flow Control ???

I'll put in another fibre for RTS/CTS

The amount of data is small and (now) the data rate is high enough for flow control not to be an issue. At the TC end I have a TX and RX 32 byte circular buffer to take care of any processing delays in either direction.

I'm aware I'm pushing the link, my aim was 57K, but If I can get 115.2K running over 5M then I have the option to go own to 57K if I have noise issues. If I aim for 57K now, Ill get 19.2K when I get the noise issues, if you follow my logic.

I'm convinced that the link is being held back by the receiver. The circuit I have used in the past for slow pulse detection isn't up to the task. It should be a nice transconductance amplifier and a shmitt, but as you can get all of those in a package with a photodiode in for Â£5 (~$7), why bother.

Ive attached the circuit just for a laugh...

Although if there is a better, simple circuit I'd love to have go whilst my new parts arrive ?

Derek

1 2 next

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