In the past, antennas at tower tops were fed by coax from radio equipment at the base of the tower. Now, to reduce transmission losses, remote radio heads (RRHs) are placed at the top of the tower next to the antenna. Fiber optic cables carrying the signal information are run up to the RRH equipment, where the light signals are converted into RF power to drive the antennas. The RRH is powered by a low voltage power line (typically 48 V DC) run up to the equipment from a power supply at the base of the tower.
Towers can be very tall (over 2000 feet), and the lightning surges to these towers can have very high I^{2}t. Consequently, the potential for RRH damage due to surges on the power feed is high, and surge protection is needed. This article will discuss the factors that affect the design of surge protectors for RRH equipment.
Lightning Characteristics
There are basically 4 types of lightning as shown in Figure 1. The important parameters of these types are the manner in which the flash is initiated, and the polarity of the flash.
Type (a)
Type (a) flashes have a downward leader and negative polarity. About 90 percent of all lightning is of this type. It characterizes ground flashes and flashes to lower towers. Most of the existing data is for this type of lightning, and most equipment will be subject to it (unless the equipment is on a tall tower). The characteristics of Type (a) lightning were most recently described in CIGRE TB549 [1] (see Table 1 below).
Parameters 
Units 
Sample Size 
Percent Exceeding Tabulated Value  From Our Sponsors 


95% 
50% 
5% 

Peak current (minimum 2 kA) First strokes Subsequent strokes 
kA 
101 135 
14 4.6 
30 12 
80 30 
Charge (total charge) First strokes Subsequent strokes Complete flash 
C 
93 122 91 
1.1 0.2 1.3 
5.2 1.4 7.5 
24 11 40 
Impulse charge (excluding continuing current) First strokes Subsequent strokes 
C 
90 117 
1.1 .22 
4.5 0.95 
20 4 
Front duration (2 kA to peak) First strokes Subsequent strokes 
μs 
89 118 
1.8 .22 
5.5 1.1 
18 4.5 
Maximum di/dt First strokes Subsequent strokes 
kA/ μs 
92 122 
5.5 12 
12 40 
32 140 
Stroke duration (2 kA to half peak value on the tail) First strokes Subsequent strokes 
μs 
90 115 
30 6.5 
75 32 
200 140 
Action integral (li^{2}dt) First strokes Subsequent strokes 
A^{2}s 
91 88 
6.0×10^{3} 5.5×10^{2} 
5.5×10^{4} 6.0×10^{3} 
5.5×10^{5} 5.2×10^{4} 
Time interval between strokes 
ms 
133 
7 
33 
150 
Flash duration All flashes Excluding singlestroke flashes 
ms 
94 39 
0.15 31 
13 180 
1100 900 
Table 1: Parameters of type (a) lightning
Type (c)
Type (c) flashes have a downward leader, but positive polarity. About 10 percent of all lightning is this type. It characterizes ground flashes and flashes to lower towers. There is less data for this kind of lightning, but it still needs to be considered, especially for the protection of equipment on low towers. The characteristics of type (c) lightning, based on the 2005 data in [3], were discussed in the 2012 PEG meeting [2]. These are shown in Table 2.
Parameter 
Median 
Sigma 
Peak amplitude (lp), kA 
35 
1.21 
10 – 90 rise time (T10) μsec 
17.6 
1.23 
Fall time (tf) μsec 
230 
1.33 
Table 2: Parameters of type (c) lightning
Type (b)
Type (b) flashes have negative polarity, but an upward leader. This type of lightning occurs only with relatively tall towers. It is infrequent, and the database for it is small, most of it going back to Berger’s 1975 paper [4]. For tall towers, protection design should be based on the data (however meager) for this type of lightning.
Type (d)
Type (d) flashes have positive polarity and an upward leader. This type of lightning occurs only to relatively tall towers. It rarely occurs, but it often accounts for 5 percent of lighting, so the effects can be significant.
Because there is so little data on type (b) and (d) lightning, data for both are combined in one Table in CIGRE TB549 (see Table 3).
Parameters 
Units 
Sample Size 
Percent Exceeding Tabulated Value 

95% 
50% 
5% 

Peak current (minimum 2 kA) 
kA 
26 
4.6 
35 
250 
Charge (total charge) 
C 
26 
20 
80 
350 
Impulse charge (excluding continuing current) 
C 
25 
2.0 
16 
150 
Front duration (2 kA to peak) 
μs 
19 
3.5 
22 
200 
Maximum di/dt 
kA/ μs 
21 
0.20 
2.4 
32 
Stroke duration (2 kA to half peak value on the tail) 
μs 
16 
25 
230 
2000 
Action integral (li^{2}dt)s 
A^{2}s 
26 
2.5×10^{4} 
6.5×10^{5} 
1.5×10^{7} 
Flash duration 
ms 
24 
14 
85 
500 
Table 3: Parameters of type (b) and type (d) lightning
The Meaning of “Low” and “Tall”
In the previous discussion, the terms “low” and “tall” towers have been used. How do we know which type we have? One way to look at this is to calculate the percentage of downward and upward flashes as a function of tower height. The type of lightning to be expected as a function of tower height comes from the relation [5]
P_{u} = 52.8 × in(H_{s}) − 230
where P_{u} is the percentage of upward flashes and H_{s} is the structure height in meters. This Equation is valid only for heights ranging from 78 to 518 m, since P_{u} = 0 for H_{s} ≤ 78 m and P_{u}= 100 percent for H_{s} ≥ 518 m. Structures with heights less than 78 m (250 feet) are expected to be struck by downward flashes only, and structures with a height of greater than 518 m (1700 feet – there are some!) are expected to experience upward flashes only. There are no specific values for “low” and “tall,” but the Equation for P_{u} might suggest how robust the protection needs to be.
Use Cases
Today surge protection devices (SPDs) for RRHs are typically designed to withstand the 200 kA 10/350 surge specified in IEC 62305. However, SPDs, which are designed to withstand this surge are likely to be addressing the wrong threat. Why? A couple of use cases illustrate the issue. The details of these cases can be found in [6] and [7] . In both cases, a simulated tower with metaloxide varistor (MOV) protection for the power feed was set up in a lab, for example, as shown in Figure 2 (adapted from [6]). The simulated tower consisted of square steel tubes bent in a U shape to minimize the generator leads, and the feed wire was protected with MOVs at each end.
When the surge was applied the result was typically like that shown in Figure 3. Hmmm… What would generally be expected is that the waveshape of the current on the power feed would match that of the applied surge – clearly not the case!
To see why, it is useful to create a model for the use case tower and the power feed, and then see if the predictions of the model match the results of the use case. Agreement would suggest that the model is correct. For the model we can use an equivalent circuit for the tower and power feed (Figure 4), where R_{w }and L_{w} are the resistance and inductance of the wire, R_{T} and L_{T} are the resistance and inductance of the tower, R_{MOV} is the resistance of the 2 MOVs in series, and V_{MOV} is the sum of the voltages across the 2 MOVs.
The equivalent circuit can be used to calculate the waveform on the power feed. Then, to validate the model, the calculated waveform needs to be compared to an actual measurement. Fortunately, in the use case developed in [6], we have an actual measured waveform. So for the validation, we can compare this to the wire waveform calculated for the equivalent circuit.
Analyzing the equivalent circuit, the following expression can be developed for the current in the wire:
where
I_{W} = current in wire, and I_{f} (s) = current of flash
Let h = height of tower
R_{w} = h*(ohms per meter of wire
L_{W} = h*(inductance per meter of wire)
R_{T} = h* (ohms per meter of tower
L_{T} = h*(inductance per meter of tower)
And V_{MOV} has been taken as a unit step function.
The coupling factor k to use in Equation (1) is discussed in detail in ITUT K.97, and is given by k = α_{F}α_{T}, where α_{F} is the coupling effect of the cable tray, and α_{T} is the coupling effect of the tower. Essentially, these correct for variables not accounted for in the calculation.
In the present case, there is no cable tray, so α_{F} = 1. From ITUT K.97, α_{T} ranges from 0.1 to 0.3. In the present case, we can determine α_{T} as 0.25 from the ratio of measured to unadjusted calculated peak current in the power feed, which is within the ITUT K.97 range of values.
Now assume that I_{f} is in the form of a double exponential. In the time domain
In the frequency domain
From Equation (1), so
The LaPlace transform of (2) gives the circuit response in the time domain:
Substituting values for a, b, c, d, k and I_{peak} into this Equation gives the waveform of I_{W}(t). From the use case [6], plus the calculated resistance and inductance for the wire, and resistance and inductance values calculated by Maytum for a 3leg tower [8]:
Lightning surge = 42.5 kA
Wire resistance R_{W} (ohms) = 0.068
Wire inductance^{1} L_{S} (µHy) = 18
Tower resistance R_{T} (ohms) = 0.0022
Tower inductance LT (µHy) = 5.8
MOV clamp V = 100 (each one, 200 V total)
Waveshape = 56.5/431, for which a = 1920 and b = 68600
The resistance of the MOV, R_{MOV}, depends on the current flowing through it, and changes as the current changes. It is difficult to take the instantaneous value of R_{MOV} into account. An approximation its average value can be calculated from
where W is the energy dissipated in the MOV (see Equation 4), and T_{max} is the time at which I_{W} goes to zero. In the present case, R_{MOV} = 0.13 ohms.
Using the values discussed for Equation (3), Figure 5 shows a plot of the calculated wire current due to the applied surge. Note that in this case, the duration of the wire current is much less than the duration of the flash (an observation first made by Maytum [9]).
Figure 6 shows the calculated waveform from Figure 5 superimposed on measured waveform from Figure 3. The correlation between measured and calculated waveforms is quite good. So it’s reasonable to conclude that the behavior of the MOV (and by extension, any clamping deice) is responsible for the waveform of the current in the power feed.
Examples of How the Model Can Be Used
The first example shows what happens to I_{W} in the use case if the height of the tower is varied and holding all else constant (Figure 7). As h goes from 10 m to 100 m, T_{0} goes from 86 to 258 µs. So tower height influences the waveshape of I_{W}.
The second example is based on the 5 percent probability of occurrence 80 kA 18/200 from Table 1 (Figure 8). Again, assuming the power feed protection is an MOV, this example uses the resistance and inductance values of the use case, with only the height as a variable. Note that the I_{W} waveshapes don’t look anything like an 18/200. They range from 11/59 to 16/101, and the peak amplitudes are much lower (shown x10 on the graph).
As a note, studying the effect of changes in variables shows that the waveform of the surge on the power feed is most affected by the waveform of the stroke to the tower and the inductance of the wire.
Practical Considerations
Energy
The reason for the interest in energy is that a common test (at least for MOVs) is to apply a rectangular wave of amplitude I_{2ms} and 2ms duration, and report the result as energy. So comparing the energy deposited in the SPD to a Table of I_{2ms} for 2ms ratings for a prospective SPD is a way of selecting an appropriate SPD for the application.
In general, the energy absorbed by an SPD of any kind is given by:
where T_{max} is the time the wire current falls to zero, v_{SPD}(t) is the voltage across the SPD and i_{W}(t) is the timedomain wire current. In general, v_{SPD}(t) is a complex function of i_{W}(t), generally increasing as current increases (e.g., for a MOV, v_{SPD} = ki^{alpha}). To simplify the calculation, let v_{SPD}(t) be constant, so that v_{SPD}(t) = V_{SPD}. Due to the variation of v_{SPD} with current, V_{SPD} needs to be chosen high enough to avoid underestimating J.
To calculate J, a reasonable choice for the lightning stroke waveform to be used in the calculation of i_{W}(t) can be found in the tables shown earlier for the various lightning types, and will typically be the value for a 5 percent probability of occurrence, for a margin of safety in the protector design. The Table to use depends on the combination of tower height and the accepTable risk of occurrence of the lightning stroke, as previously discussed. Once the variables for the expression for i_{W}(t) have been determined, J can be calculated.
When a clamping device is used for the SPD, i_{W}(t) can be calculated from Equation (3). Substituting Equation (3) into Equation (5) and performing the integration:
where
For the use case, J = 48 J total, or 24 J per MOV, in which case a 14 mm MOV is suitable. But the use case does not represent the type of lightning flash that generally occurs.
The Issue of Multiple Strokes
Up to this point, we have considered a flash having only a single stroke. These are rare. Generally, the flash will consist of a first stroke followed by several subsequent strokes and a continuing current, something like that shown in Figure 9 (adapted from [10]):
Depending on the interval between strokes and the thermal time constant of the clamping device, the energy due to the strokes might accumulate in the SPD. Added to the energy from the strokes is the energy deposited by the continuing currents (if any). In fact, the energy due to continuing strokes might be more of an issue than the energy from the associated strokes.
Maytum [11] gives an example of this case. A paper by Yang et al [12] reported that a 40 kA SPD stopped functioning at lightning currents no greater than 1.6 kA. Why? Analysis of the data suggested that the cause might be accumulation of energy delivered by the 350 A, 0.5/29.5 first stroke, followed by eight return strokes ranging from 0.22 kA to 1.64 kA, with a geometric mean waveshape of 15.6/63.3. By itself, this accumulated energy might not have been a problem. But added to this energy was that from several impulses which showed a continuing current content lasting some 3 ms and delivering 100 J.
This added 100 J probably pushed the SPD beyond its capability.
In the case of multiple strokes and continuing current, Equation (4) still applies, but now i_{W}(t) includes all strokes and any continuing current. For an example, see post #20 by Maytum on the SPDC web site entitled, “8/20 burst MOV energy.”
So, to summarize, with a clamping device, the important thing to consider is the amount of energy deposited in the device by all components of the flash, that is, first stroke, subsequent strokes, and continuing current. A protector chosen without consideration of energy may be undersized, as the example just cited shows.
References
 CIGRE Working Group C4.407, “Lightning Parameters for Engineering Applications,” CIGRE TB 549, 2013.
 A. Martin, “Lightning Induced GPR Characteristics and Comments,” in The Alliance for Telecommunications Industry Solutions Protection Engineers Group Conference, March 13 – 15, Huntsville, Alabama, 2012.
 Lightning and Insulator Subcommittee of the T&D Committee, “Parameters of Lightning Strokes: A review,” IEEE Trans. ON Power Delivery, vol. 20, no. 1, pp. 346358, 2005.
 Berger, K., R. B. Anderson, and H. Kroninger, “Parameters of Lightning Flashes,” Electra, vol. 41, pp. 2337, 1975.
 Eriksson, A. J., “The Incidence of Lightning Strikes to Power Lines,” IEEE Trans. on Transmission and Distribution, vol. 2, no. 3, pp. 859870, 1987.
 Zafiris, P., “Lightning Current Through MOV Based SPDs in RRH Applications,” Raycap GmbH, Parkring 11, 85748, Garching bei Munchen, Germany.
 Xiong, Y., “Lightning Current Waveform onDC Power conductor Along Tower,” ITUT document TD 94rev1, Geneva, Switzerland, January, 2013.
 Maytum, M.J., Remote Radio Unit (RRU) DC Feed protection, Working document contribution to IEEE PES SPDC WG3.6.3, 2014.
 Maytum, M. J., “Towering Powering Protection Problem on Remote Radio Head (RRH) Cellular Systems,” in The Alliance for Telecommunications Industries Solutions Protection Engineers Group, Huntsville, Alabama, March, 2012.
 Rakov, V. A., “Lightning Parameters of Engineering Interest: Application of Lightning Detection Technologies,” in EGAT, Bangkok, Thailand, November 7, 2012.
 Maytum, M. J. , “Technical Bulletin (TB) 549 (2013), ‘Lightning Parameters for Engineering Applications’,” in Alliance for Telecommunications Solutions Protection Engineers Group Conference, Littleton, Colorado, March, 2013.
 [B11] YANG, ShaoJie1, SD CHEN, YJ ZHANG, WS DONG, JG WANG, M. ZHOU, D. ZHENG, and H. YU Hui., “Triggered Lightning Analysis Gives New Insight into Over Current Effects on Surge Protective Devices”.