PE1-2.5 Optical System Design

What this is and why it exists

Every topic so far described one component. This one puts them together and asks a single question. Will this link work?

The answer is two budgets. One for power and one for speed. A design passes only if both of them close.

The vocabulary

• Point-to-point link — one transmitter, one fibre run, one receiver.
• Link power budget — the accounting of every gain and loss between source and detector.
• System margin — spare loss deliberately left over for ageing, repairs and error.
• Receiver sensitivity — the least optical power the receiver needs for its specified error rate.
• Rise time — how long a signal takes to go from low to high.
• Rise time budget — the accounting of every contribution to the link's overall rise time.
• Quadrature addition — combining independent rise times by squaring, adding and taking the square root.
• WDM — wavelength division multiplexing. Several wavelengths sharing one fibre.
• Multiplexer and demultiplexer — the components that combine and separate those wavelengths.

The mental model

The power budget is simple accounting and it is the single most useful exercise in this subject.

Start with the power the source launches into the fibre, in decibels relative to a milliwatt. Subtract the fibre loss, which is the loss per kilometre times the length. Subtract every connector. Subtract every splice. What remains must be more than the receiver sensitivity, and the difference between them is the margin.

The margin is not spare arithmetic. It is deliberately kept for the things that will happen. Components age, a cable gets cut and respliced, and temperatures leave the laboratory range. A design with no margin is a design that fails in its second year.

The rise time budget is the one students under-rate, and it is why a link with plenty of power can still fail.

Four things slow the edges. The transmitter cannot switch instantly. Material dispersion spreads the pulse because the source is not a single wavelength. Intermodal dispersion spreads it further if the fibre is multimode. And the receiver has its own bandwidth limit.

These contributions are independent, so they do not add directly. They add in quadrature: square each, add the squares, and take the square root. The total must be comfortably less than the bit period, or consecutive bits blur into each other.

The two budgets pull on the same choices in different directions, which is what makes component selection a real decision. A cheaper LED source saves money and costs rise time. A longer wavelength cuts fibre loss and may worsen dispersion. Single mode fibre removes intermodal dispersion and makes every connector and splice harder. There is no ordering of components that is best for both budgets at once.

Wavelength division multiplexing closes the subject, and it is the idea that made the installed fibre plant last. A fibre carrying one wavelength is barely using its available spectrum. Send several wavelengths down the same fibre, each with its own transmitter and its own data, and the capacity multiplies. A multiplexer combines them at one end and a demultiplexer separates them at the other.

Nothing has to be dug up. That is the whole point. When demand grew faster than anyone could lay cable, this is what kept up, and it is why long haul capacity became affordable.

What you should now be able to explain or do

Build a link power budget from launched power, fibre loss, connectors, splices and receiver sensitivity, and say why margin is kept. Build a rise time budget from the transmitter, dispersion and receiver contributions, and combine them in quadrature. Say why a link can pass one budget and fail the other. Choose components against both budgets and name the conflicts. Explain what wavelength division multiplexing achieves and what components it needs.

Check yourself