Key OTDR Specification Relates to User's Real Needs

With these problems in mind, Bell Communications Research (Bellcore, now called Telcordia) has introduced some specifications (GR-196-CORE generic requirements for OTDR-type equipment) that explain an OTDR’s performance from a user’s standpoint. First, the calibration and test setup relate directly to field operation. And, second, the specifications consider both hardware and software by characterising the quality of an OTDR’s display trace and its measurement accuracy.

Dynamic Range Still Key Spec

Dynamic range remains one of the most important characteristics of an OTDR because that spec determines the maximum observable length of a fibre. The higher the dynamic range, the higher the signal-to-noise ratio and the clearer the display will be, and the better the event detection. Historically, vendors have made life difficult by using a variety of methods to describe dynamic range.

You can define dynamic range as the difference between the extrapolated point of a backscatter trace at the near end of a fibre (taken at the interception between the extrapolated trace and the power axis) and the upper level of the noise floor after the fibre end. The level is expressed in dB and is averaged over three minutes.

To complicate matters, there are several ways you can define the upper level of noise (see Figure 1).

First, you can regard it as the upper limit of a range that contains at least 98% of all noise data points (as in specification IEC 61746).

Second, some vendors commonly use what is called the RMS noise (or S/N = 1) method. This method relies on the usual way of representing the magnitude of a fluctuating variable as its root-mean-square (RMS) value.

Figure 1. Both the 98% and RMS methods of defining dynamic range are valid but the RMS method gives “better” values.

Figure 2. Dynamic range specifications based on noise levels close to the noise floor don’t give you any information about measurement accuracy in this area of operation.

Figure 3. Attainable measurement range on a 0.5 dB splice is less than the theoretical dynamic range vendors have previously quoted in specifications.

In practice, the 98% method seems more representative because it shows a level where the signal visually meets the upper noise signal envelope. The RMS method has more attraction though because it allows you to quote higher dynamic range specifications. As soon as one vendor started to use this definition, others were forced to follow in order not to appear at a disadvantage. To create further confusion, vendors may use different values to extrapolate results from 98% and RMS methods ranging from 1.5 to 4 dB.

Whatever noise definition you use, the dynamic range defines only an attenuation loss between two levels on the OTDR trace (from maximum signal level to noise floor level). However, the closer the signal is to the noise floor, then the noisier it becomes. For example, if you visually inspect an OTDR trace 6.6 dB above the noise floor, you’ll see the noise formed on the trace has a 60.1 dB amplitude. If you observe at 3 dB above the noise floor, you’ll see a noise amplitude of +0.5 dB -0.6 dB. The theoretical local noise level according to the signal-to-noise floor attenuation margin appears in Table 1 and Figure 2. The important point is that dynamic range specs based on noise levels in this area of operation give you no idea of measurement accuracy.

So, What Is Measurement Range?

You could consider measurement range to be equal to the dynamic range reduced by a SNR margin where non-linearities or noise result in measurement errors that exceed preset limits. According to Bellcore GR-196-CORE (issue 1, rev 2, Dec 98) standard, an OTDR’s measurement range is the maximum (one-way) attenuation that you can place between its optical output port and the “event”, and for which it can “accurately” identify the event. The standard defines four measurement ranges:

• splice loss range,

Today, vendors most commonly use the splice loss measurement range, assuming an insertion loss of a 0.5 60.1 dB event and a reflectance of less than -40 dB. A successful measurement assumes accurate measurement results from four out of five attempts. Figure 3 compares the typical difference between dynamic and measurement range limits.

Measurement Range Has Advantage

Adopting a measurement range instead of dynamic range specification has several advantages because the spec takes into account the overall measurement domain. For example, a measurement range spec considers signal-to-noise, linearity, distance resolution, event detection and measurement algorithm, and repeatability. In the case of a 0.5 dB splice loss, the OTDR has to detect the splice, make a local attenuation measurement, and provide 60.1 dB accuracy.

Bellcore defines a measurement range test setup based on fibre lengths of 1 to 4 km before an event and 1 to 5 km after the event. This setup results in poor performance using 20 ms or more pulses and explains the difference between achieved measurement ranges and quoted maximum dynamic range specifications at maximum pulse widths. The fact is that operators don’t use OTDRs at a 20 ms setting in the field (except for special applications such as on submarine cables).

Adopting a Test Setup

Bellcore includes procedures for the four measurement range test setups, but you can also use a simpler arrangement (see Figure 4). This arrangement still needs fibre reels, a low reflectance variable attenuator (to simulate long fibres), a calibrated reflectance and insertion loss coupler, and a reflective or non-reflective fibre end. You’ll also have to repeat measurements and to check the final insertion loss with an optical test set. If your OTDR has no low reflectance input connector, you can place a fibre between the OTDR and the low reflectance attenuator. Additionally, you can use this fibre to initially measure global attenuation (with the attenuator set to minimum) between an OTDR and a test splice. Don’t think you can do the measurement without a low reflectance setup because multiple reflections, multiple backscatter (ghost signals), and attenuation dead zones will unavoidably overlap and disturb the area of fibre where you’re trying to measure the splice.

Figure 4. The simplest possible test setup for measurement range still requires a low reflectance variable attenuator, a calibrated reflectance and insertion loss coupler, and a non-reflective fibre end.

If you focus on the splice loss measurement range, you can consider 0.5 dB insertion loss as a realistic value for fault detection but it may be too large if you want to qualify or commission real splices with smaller than 0.1 dB. Don’t forget, when you build the 0.5 dB insertion loss event, check the value with an optical light source and power meter, as well as an OTDR (from one end).

Splice loss measurements depend upon the relative value of the Rayleigh backscatter coefficient of the two spliced fibres. To minimise this uncertainty, use fibre from the same reel and check the direction dependence of insertion loss with the OTDR used under favourable signal-to-noise conditions (so there is a small optical loss budget compared to the dynamic range).

You can consider a reflectance of less than -40 dB as a realistic value, although this value could also mean that an event appears as a reflective event using short pulse widths, and as a quasi non-reflective event at the largest pulse-widths. For example, a -45 dB reflectance event leads to a 13.5 dB peak amplitude above the local backscatter level for the 5 ns pulse width test, and only 0.5 dB for a 10 ms (at 1310 nm) pulse. So, your choice of a particular reflectance value could optimise the event detection capability.

These last points show how difficult it is to define a universal and field work representative test setup that is able accommodate applications from submarine cables to passive optical access networks. Figures that OTDR vendors give can vary between dynamic range and measurement range from 6 to 14 dB and could indicate that users have misunderstood Bellcore’s standard. You have to carefully analyse measurement test conditions. Note that some users could adopt an alternative setup based on a fibre looped between one output and one input port of an optical coupler. This controversial setup generates an infinite impulse response that simulates a non-ended fibre link pattern, and could convey a rough but rapid idea of a “measurement range”. T&ME

André Champavère is an R&D project manager with Wavetek Wandel Goltermann, Saint-Etienne, France. He specialises in fibre-optic measurements, opto-electronics, and DSP. He joined WWG in 1981.