Power electronics on a chip: The Solar Charger
One area of electronics that has seen intensive integration yet is not at all well-suited for total integration is power electronics. I need not recount the long list of linear voltage regulators, PWM and PFC controllers, battery chargers, motor-drive controllers, and gate drivers that can be found in integrated form. This integration has been taken about as far it can be, with high-voltage processes allowing the power switch to be integrated into 5-pin off-line switching power converters in a power IC package. The question that confronts the innovative mind is which IC functions have yet to be integrated in power electronics. How about solar chargers?
Solar chargers for off-grid systems add very little benefit most of the time; they guard against battery overcharge and do little else. They cannot charge the battery bank based on some prescribed charging procedure because sun, clouds, and solar-array size determine charging rates. Many simple solar-power systems connect the panels directly to the batteries. The following block diagram shows where the solar charger fits into the overall off-grid scheme.
In view of this, the choice of solar-charger integration at first seems to be a marginal possibility from a marketing standpoint because it is able to be implemented (and is) with low-cost μCs in semi-discrete designs. Solar power is growing very rapidly as China builds 240 W solar panels that are sold at a price of $300 US - little more than $1/W and are competitive with grid-sourced power in places where electric rates are high. The newer panel technology based on a continuous (rather than batch) thin-film process has projected prices well under one dollar.
As solar panels become low in price, the need to maximize efficiency of solar-panel output power is reduced and the solar charger problem is less critical. On the other hand, the need to maximize efficiency of electric power with the ever-higher volume of its global usage implies an increased need for solar chargers which maximize power transfer from the panel to the battery.
The newer chargers have maximum power-transfer control, referred to in advertising literature as “maximum power point tracking”. The power-point referred to is on the v-i curve of the solar panel, as shown below. There is a family of curves at different solar input power densities. The overhead sun on a clear day delivers about 1100 W/m2. When smog, chemtrails, and other factors are taken into account, a useful working design value is 1 kW/m2.
The curves given in solar-panel specifications are based on this maximum optical power input. The solid curve shows that the voltage remains nearly constant until the short-circuit current value of I0 is reached. Consequently, over most of the curve, as current increases, power increases linearly until the maximum power point is reached, near the short-circuit current.
To find a maximum in a μC, it is easy to store prior values of v and i and invoke a peak-seeking algorithm. There are some less than obvious methods for analog-circuit regulation of peak power, though a more obvious one (suggested to me once by Bob Pease) is based on the observation that battery voltage is very close to constant for long time scales. Consequently, measure output current to the battery and servo the input current to maximize it. Not only does this simple scheme maximize power to the battery, it avoids the complication of also taking into account the charger efficiency as a function of power. By measuring output power, it is already in the loop. Battery voltage is easy to acquire with a μC, but the current for an IC requires some consideration.
The amount of current that a typical residential charger handles is in the range of 20 A to 120 A. With a 24 V battery bank (25 V nominal), this corresponds to 500 W to 3 kW of peak charging power. This much current is usually handled by one or more power MOSFETs, and for a more complete integration, the MOSFETs, which are PWMed by the on-chip controller, controls the effective input resistance of the charger.
The maximum-power transfer theorem applies because the panel-charger-battery system can be modeled as (nearly) linear incrementally around the maximum-power operating-point of the panel. While charge rates for a given battery technology such as lead-acid (or preferably nickel-iron) cannot be optimized along with maximum transfer power, it is of secondary design priority to power. The power MOSFET is PWMed to maximize output (battery) current. How does it satisfy the maximum-power transfer theorem?
The slope of the panel v-i curve at pmax is some negative value (negative because it is a source). If the charger presents an equal value, maximum power is transferred. The duty ratio, D, of the switch controls the average input resistance of the charger by controlling the average switching-cycle current while the voltage across the charger (from input to output port) remains nearly constant. Thus PWM control with a common-passive (buck) PWM-switch configuration can maximize battery input power from the solar panel.
A 60 A charger is a good IC design goal for a company having power MOSFET integration capability. Usually, power MOSFETs are optimized with their own process but additional circuitry can be added, as several companies have been doing for some time. What complicates the design somewhat is that current from the battery must not be allowed to flow back into the solar array at night and discharge the battery. Consequently, MOSFETs, with their body-drain diodes, are usually placed in series, with the diodes back-to-back to prevent reverse current flow. On a single chip, two MOSFETs at the rated current (in series) are required.
The solar charger challenge is the integration of a μC with power components. Project risk is reduced, perhaps, by finding an analog peak-power scheme. The problem is that previous power (or current) values need to be stored for the time it takes to change the PWM duty ratio and let the circuits settle. This is not an excessive amount of time for an analog peak detector circuit. Consequently, the solar charger IC looks like a good prospect for integration. The market volume in the off-grid power sector is increasing rapidly.