What's so special about a low attenuation interconnect?

Low optical attenuation interconnects in data centres are required to increase flexibility by using structured cabling and can lower the total power dissipation, but how are they measured and made?

First it’s important to note that optical attenuation, also known as transmission loss, is the reduction in intensity of the light beam (or signal) and can be quantified using the following equation: attenuation (dB) = 10 x log10 (input intensity (W) / output intensity (W)). With this formula 3dB means losing 50 percent of intensity of the light beam, 10dB means losing 90 percent, and 1 dB means losing 21 percent. Optical interconnects of today often have an attenuation performance which is far better than 1dB. Examples are 0.7dB which means 15 percent attenuation and 0.5dB resulting in 11 percent attenuation.

Another important item to note is the way optical attenuation is specified. Different manufacturers use different ways of specifying by using “average”, “maximum” and “typical”. For the first two its clear what is meant, but what does “typical” mean? I would say it’s something that is normal or not unusual, but what does that mean for the attenuation of an optical interconnect? That’s why I prefer to use “average” or, even better, “maximum”. It all relates to a large series of optical interconnects and what you can expect in optical attenuation when connecting two of them together.

The last item to note is how optical attenuation is measured. In general we need a second interconnect to measure the first, but how do we specify this second interconnect? We can use a random interconnect of a larger set or we can use a “reference connector”. A reference connector is a specially manufactured or selected interconnect of which we know that the optical fibre is very close or exactly in the centre of an optical interconnect. So when we talk about the attenuation of an optical interconnect, the most severe way of describing it is “random mated max”. The disadvantage is that it’s difficult to measure this in a manufacturing environment and normally determined by distributions and statistics. That’s why it’s more common to use a maximum attenuation value against a reference.

So we now know how to specify and to measure an optical interconnect, but do we know how to make one? In general, two items are very important. First is that the optical fibres are really touching each other at all times, which is often referred to as “physical contact”. Second is the X and Y-alignment of the optical fibres relative to each other, as well as the angle, if any, between the fibres. Especially when the amount of fibres per interconnect is increasing then this is more difficult to achieve. It’s all about the quality of the components, the termination process and the proper connector end-face finish. Please keep in mind that we are talking about XY-alignment at the micro-metre level and fibre height control at the nano-metre level.

Last time I promised to write about low attenuation, as well as density, so it is clear density is the subject of my next column. I hope to ‘meet’ you again!