Industry isolated communication

and from a non optical approach, TI has some isolated tranceivers for RS485 for a few $$$ each

http://www.ti.com/lit/sg/slyt484a/slyt484a.pdf

these provide full galvanic isolation upto about 4KV

Either way you still need the Parallel to serial conversion of your data or many isolated traceivers, it is how you get from location A to location B that is the descussion here

Fiber will add a couple of hundred $$ to the cost or more depending on where you shop and may need ethernet as the connection at each end which is also a viable option btw and ethernet also provides isolation due to its termination transformers at each end of the cable

Ethernet will provice about 1000V of isolation on its own

from a cost, Ethernet and RS485 should be the cheapest, optical more expensive but providing superior isolation if needed

Plastic fibre will add far less than $200 to the cost and no way does it need ethernet.

You can connect low cost optical transceivers directly to the UART ports of the micro.

TX HFBR-1522Z (Farnell 1173129 £8.80) - will need a driver chip but it's not expensive

RX HFBR-2522Z (Farnell 1173132 £8.88)

The data sheet from Avago is helpful on these parts.

20metre coil of cable (Rapid Electronics 58-0710 £20.81)

Ethernet (using the usual cables and transformers) is an option but needs much more software and more expensive and complicated micros at each end. There are some TI launchpad thingies with Ethernet built in which might make a simple way to go.

(BTW Ethernet transformers are supposed to be tested for 1000V breakdown - they are not usually rated for safe use at 1000V and most pcbs are not tracked to provide serious HV isolation. RJ45 connectors may or may not comply with the regulations for creepage distances so any solution with magnetics built into the connectors should be considered as an earth loop avoidance solution but no good for safety isolation (unless you have evidence that they are OK))

100Mbit Ethernet using only basic Ethernet protocol (no IP stack or RTOS) can give very good performance at low cost - any kind of OS and using off the shelf IP software will add a lot of latency and slow things down significantly.

MK

Plastic fibre will add far less than $200 to the cost and no way does it need ethernet.

You can connect low cost optical transceivers directly to the UART ports of the micro.

TX HFBR-1522Z (Farnell 1173129 £8.80) - will need a driver chip but it's not expensive

RX HFBR-2522Z (Farnell 1173132 £8.88)

The data sheet from Avago is helpful on these parts.

20metre coil of cable (Rapid Electronics 58-0710 £20.81)

Ethernet (using the usual cables and transformers) is an option but needs much more software and more expensive and complicated micros at each end. There are some TI launchpad thingies with Ethernet built in which might make a simple way to go.

(BTW Ethernet transformers are supposed to be tested for 1000V breakdown - they are not usually rated for safe use at 1000V and most pcbs are not tracked to provide serious HV isolation. RJ45 connectors may or may not comply with the regulations for creepage distances so any solution with magnetics built into the connectors should be considered as an earth loop avoidance solution but no good for safety isolation (unless you have evidence that they are OK))

100Mbit Ethernet using only basic Ethernet protocol (no IP stack or RTOS) can give very good performance at low cost - any kind of OS and using off the shelf IP software will add a lot of latency and slow things down significantly.

MK

I stand corrected, Nice Michael.

I was looking at standard optical units through google and most of them had an ethernet port accociated with them

A few years ago I did several very low cost links using the Avago interface bits and plastic fibre but since then it's become much harder to find the twin plastic fibre at a reasonable price.

I've recently worked on a very high performance analogue link (true analogue - no adc or dac, > 100dB S/N) using plastic fibre but the fibre cable was made for the job so out of range of a one off project. We made our own fibre to LED/PD mechanical bits.

Amplicon used to make and sell a very low cost RS232 to fibre converter system but now they only do the Moxa (or similar) ones and they do cost about £100 per end.

MK

Hi Michael,

Very cool to hear about the analogue link. If it is not proprietary information, may I ask, does it use FM? If so, what technique was used for the demodulation - was it quadrature detection?

No - it's baseband linear - done for analogue audiophiles. Two LEDs in push-pull like an audio amp (one does +ve half cycle and other does -ve) two fibres, two receiving photo diodes and recovery amps. Bandwidth about 50kHz, S/N ref about 2% distortion is approx 102dB. I never thought it would work either. It is patented but you can't buy one yet.

I thought there might be industrial applications but haven't thought of one yet.

MK

Hi Michael,

That is very interesting. Thanks for the information!

If I can think of industrial use-cases, I'll let you know. I'm thinking there are bound to be scenarios where it is too expensive or difficult to process or convert data close to the machine or process - you could remotely do this by transporting sensor signals to a safer or easier environment, e.g. to a control room. Or perhaps automotive scenarios too?

How about remote speakers.

In our case the workstation is often in the equipment room and the KVM is remoted to the cab (above), but they have an issue with speakers for the audio alerts.

Because these installs generally have VHF/UHF radios, keeping the RF out of the audio is a problem.

Mark

plastic fibre but since then it's become much harder to find the twin plastic fibre at a reasonable price.

We were using plastic fibre for lighting effects.

They were available in a range of sizes from 0.5 - 6mm and larger.

Not sure what sort of loss there was, but the fibre was very cheap.

Some were single fibre and others were multiple fibres in a sheath (always use the loose sheath otherwise it can crack the fibre or the coating).

Mark