Series vs shunt linear voltage regulation for small solar-photovoltaic power supplies

Article By : Stephen Woodward

This Design Idea presents circuits for—and compares the advantages of—shunt versus series regulation for small solar power arrays.

Solar-photovoltaic arrays seem to get cheaper and more efficient every day, making them increasingly practical in renewable and/or remote power supply applications. Still, the voltage generated by any given array varies significantly with loading, incident light intensity, and temperature, so some form of regulation is often necessary.

Array performance can benefit significantly from Maximum Power Point Tracking (MPPT) and switch-mode regulation, as illustrated in an earlier Design Idea: Solar-array controller needs no multiplier to maximize power.

But for small arrays, the extra complexity of MPPT and switch-mode circuitry might not seem justified, making linear regulation the simpler and better choice. This Design Idea addresses such systems, focusing on the relative advantages of series versus shunt regulator topologies.

Let’s start with a hypothetical small solar array optimized for 12W output (in full direct sunlight ~1kW/m2) of 1A at 12V, a 20% light-to-electricty conversion efficiency, and therefore a nominal area ~0.06m2 = ~100in2. Then add linear regulation circuitry to maintain a constant 12V output against load current variation from 0 to 1A.

Figure 1 illustrates a suitable series regulator while Figure 2 is a comparable shunt topology. To ease comparison of the advantages of shunt versus series regulation, both regulators employ identical sense/control circuitry based on the venerable LM10 combo reference + op amp.

Figure 1 A suitable series linear regulator for small solar arrays.

Figure 2 A suitable shunt linear regulator for small solar arrays.

Referring to the figures, the LM10 200mV internal reference (pin 1 + 8) drives the op amp inverting input (pin 2) via R1 = R2R3/(R2 + R3) that provides input bias-current compensation, while the noninverting input (pin 3) connects to Vout via the 60:1 R2:R3 voltage divider (Vsetpoint = 200mV(R3/R2 + 1)). Thus op amp output (pin 6) will slew negative when

Vout < Vsetpoint and positive when Vout > Vsetpoint.

In Figure 1 (series regulator), pin 6 connects through current-limiting R4 to the base of the D45 PNP pass power transistor, increasing drive and load current when Vout < Vsetpoint, decreasing them when Vout > Vsetpoint.  In Figure 2 (shunt regulator), pin 6 drives the base of the D44 NPN shunt transistor, routing more array current to ground when Vout > Vsetpoint, less when Vout > Vsetpoint.

So, which type of regulation, shunt or series, is better, and when, and why?

To answer this general question, three specific categories of circuit performance will be considered:

  1. Regulator efficiency (the maximum fraction of array power delivered to the load at peak demand)
  2. Thermal management challenges (primarily determined by the required thermal capacity of the power transistor heat sink, in turn determined by maximum transistor power dissipation)
  3. Effect of regulation type on solar array temperature and thereby on array conversion efficiency

Regulator efficiency

Full-load (1A) efficiency of the series topology, when the D45 pass transistor will be ON and near saturation, is limited by three factors:

  1. Current draw of the LM10 and R2R3 voltage divider = 312uA(typ)
  2. Base drive for the D45 @Ic = 1A = 10mA(typ)
  3. Saturated voltage drop of the D45 @Ic = 1A = 100mV(typ)

Summing these losses produces an estimated typical efficiency factor of 98%.

By contrast, in the shunt topology the D44 power transistor is completely OFF at full load and the connection between array and output is direct, leaving only one of the three factors outlined above to compete for output current:  #1—the 312uA LM10 current. This results in near-perfect 99.97% efficiency.

Conclusion: For efficiency, series is very good but shunt is (practically) perfect. Note that this result differs from the general expectation that series regulation’s efficiency usually trumps that of shunt’s.

Thermal management challenges

Maximum heat dissipation of the D45 series pass transistor is ~1.33W, occurring at 0.66A load current, which can be accommodated by a small clip-on heat sink.  Maximum dissipation of the D44 shunt transistor, by contrast, occurs at zero load current, and is much greater:  ~4.5W, requiring a substantial and bulky extruded sink to acceptably limit temperature rise (~40oC) under conditions of natural convection and radiation.

By this criterion, series regulation is the obvious winner by a (cool) factor of >3.

Effect of regulation type on solar array temperature

Total solar energy absorbed by a solar array can go only two ways: 1. Conversion into electrical power delivered to the connected circuit; or 2. heat dissipated by the array. The first law of thermodynamics dictates that the sum of the latter two must always exactly equal the former.  Consequently, the less electrical power that’s accepted by the connected load, the more that must be dumped by the array as heat, which inevitably increases the temperature of the array.

Series regulation causes most of the power that’s not accepted by the load to be dissipated by the array (remember how cool the D45 stays), while shunt regulation dissipates rejected power in the D44 transistor and R4. Thus, at partial load, a 20% efficient shunt regulated panel runs cooler than the series regulated panel, by as much as 10°C. Solar array conversion efficiency declines with temperature rise by 0.3% to 0.4%/°C, so that a shunt regulated panel may be 3% or 4% more efficient than a series regulated panel in some circumstances.

By this criterion shunt regulation is obviously superior.

To sum up, we see a mixed bag: Does shunt regulation win the design derby by beating out series in two ABCs out of three? It depends. Balancing conflicting criteria in the designer’s choice of regulator type will depend on competing priorities as they sort themselves out in the detailed requirements of a particular application. That’s why we design engineers earn the big bucks! Ahem.

This article was originally published on EDN.

Stephen Woodward‘s relationship with EDN‘s DI column goes back quite a ways. In all, a total of 64 submissions have been accepted since his first contribution was published in 1974.


Leave a comment