Dark Secrets of Light Budgets
How to calculate light budgets and select the best fiber tap for the job.
In using passive fiber taps Engineers are often asked to verify the light budget. So what is a “Light Budget” and how is it calculated?
At either end of a fiber optic link there is a transceiver. A transceiver both transmits data over light at a given power level and also receives the light. On reception the light must be at a sufficient level to maintain the Bit Error Rate (BER) of the link. Having too low a light level causes the error rate to increase. If the light level is too low, the link will fail to work at all. Most data sheets for transceivers will specify the required receiver light levels (Sensitivity) for a given BER. For example a BER of better than 10EXP-12 might have to have better than -24 dBm of power. If the same transceiver has a average launch power of say 4dBm this means that the total losses across the transmit path must be less than 28dB. This is the Maximum Loss Target.
The actual loss is made up of the losses introduced by the fiber itself, any connectors in the path (including patch panels) and any passive fiber taps that are introduced into the path. In addition there are chromatic dispersion effects that can occur on fiber runs that further degrade performance and these should also be included in any loss calculations. The summation of the losses is known as the loss budget and the loss budget should always be less than the Maximum Loss Target.
So lets look at all the parameters that make up the light budget calculations…
1. Transceiver Loss Target - the launch power levels and receiver sensitivities will vary by technology and vendor. The following are transmit power and receiver sensitivities for a leading manufacturer of 1G and 10G transceivers.
Note that the Target Loss must be less than the difference in the worst case (ie. Minimum) average transmit power and receiver sensitivity. As you can see from the table the specifications for transmit powers can vary widely (often more than 6dB). Ixia’s experience is that most transceivers purchased have transmit powers a lot closer to the best case power levels rather than the worst case, so using the worst case tends to be a very conservative approach.
2. Cable Losses - cable losses vary by both the wavelength and the transmission mode of the light. The difference between a high quality supplier and a low quality supplier can also be significant. Losses will vary by the distance traveled.
Typical losses are:
Multi-Mode – (850 nm) 3dB/km – OM4 fiber type is closer to 2.5dB/km and OM3 is closer to 3.5DB/km
Single-Mode – (1310 nm) 0.4dB/km (OS 2 fiber types)
Single-Mode – (1550 nm) 0.3dB/km (OS 2 fiber types)
So within a data center, say with a 100m fiber run, the losses for Multi-Mode are significant, but for Single mode there are not as important – Moral – if you are in a data center and you are up against light budget issues use Single Mode technology. Also use higher quality (OM4) to minimize losses.
3. Connector Losses - connector losses depend on how and when they are mated with the fiber. Lowest losses occur when the connectors are ‘fused’ to the fiber in a high quality environment. Again the losses will vary according to the quality of the supplier. Typical losses are:
Multi-Mode – (850 nm) 0.3dB
Single-Mode – (1310/1550 nm) 0.2dB
“field’ connections that are hand made can often have higher insertion losses, maybe up to 1.0 dB. Moral – avoid hand made connections if possible, and keep the number of connections in a path to a minimum.
4. Splicing Losses - sometimes its necessary to splice fiber cables together than use connectors. Losses for multi-mode fiber are similar to mechanical connectors (around 0.3dB), though fusion splices that are often used for single mode cables can be better than an impressive 0.1dB. Fused splices are also better in extreme conditions.
5. Keep the fiber clean - a single spec of dust can severely degrade the performance of the optics. Its therefore important that the dust covers are kept on the connectors, transceivers and taps.
6. Other losses - for long distance fiber runs of single mode cables chromatic dispersion must be allowed for. This is the impact of there being different propagation delays for the frequency of the light either side of the nominal frequency. This impact can usually be ignored within a data center.
Once the overall light budget has been calculated, the correct tap split ration can be selected based on the insertion loss.
Lets consider two worked examples within a data center. Case 1 involves a run of 5m within a single rack using multi-mode cables at 1G, and Case 2 involves a run of 100m using single mode cables at 10G. In the first case assume that the only connections are the ones at either end of the fiber runs that connect to the transceivers, in the second case assume that the fiber run goes via a patch panels, so a total of 4 connectors are used.
Case 1: MM 10m at 1G
A. Light Loss Target (from Trans. Analysis) 9.00 dB
B. Cable 0.03 dB
C. Connector losses 2 x 0.3dB 0.60 dB
D. Total Loss ‘Margin’ A – B – C 8.37 dB
In this case there is 8.37 dB of ‘light’ budget available for use by any taps that need to be inserted into the light path. The insertion losses of the Ixia Flex Taps are shown on the data sheet on the ixia web site. In this case the insertion loss of the 1G 50/50 split multi-mode taps is 4.5dB (excluding any connectors). So in this case there is no problem in using this standard 50/50 tap.
Case 2: SM 100m at 10G
A. Light Loss Target (from Trans. Analysis) 4.40 dB
B. Cable loss 0.04 dB
C. Connector losses 4 x 0.2dB 0.80 dB
D. Total Loss ‘Margin’ (C-D-F) 3.56 dB
In this case there is only 3.5 dB of insertion loss that can be allocated to a fiber optic tap. At 10G the insertion loss of a Single Mode tap is 3.7dB which is just above the allowance. In this situation its therefore wise to go with a tap with less of an insertion loss. The 60/40 insertion loss SM Flex Tap only has a loss of 2.8 dB on the network link, so going with this tap minimizes the risk of too low a light level on the network link.
The down side of this fiber tap though is that loss on the tap monitor port is 4.8 dB. This is 1.24dB higher than the loss margin, so there is a risk that the light level will be too low to meet the BER rates on the monitor port. However if you look at the transceiver specifications you can see that the difference between the best and worst case transceiver power is 4dB with this transceivers. There is therefore a very good probability that there will be no problem with this tap and the light being received by the monitoring tool.
So, to wrap things up in an optical network the amount of light available for various activities is finite and is decreased by a number of things including the fiber itself, connectors and taps. When building a network, in order to maintain performance it is important to stay under your Maximum Loss Target. Avoiding field connections and selecting the right taps can help.