Charge controllers are often equipped with a load switch that protects the battery from deep-discharge. If supported by the charge controller, the load switch can also be used for advanced control of connected loads, e.g. switching on lights during the night.
The following image shows the layout of a typical MPPT charge controller with the DC/DC power stage between the solar panel and the battery and a switch between the battery and the load output.
Even though it might seem very simple to switch on and off a load using a MOSFET, a reliable load switch control can get quite complicated. The load switch needs to protect itself and the charge controller from overcurrents and from exceeding its thermal limits. In addition to that, it must protect the wires from short circuits, acting like an electronic fuse. These different protection features are described below in more detail.
High-side vs. low-side
- Circuit diagrams
- Advantages and disadvantages
Inrush current during turn-on
A load with high capacitance will generate an inrush current after the switch was turned on. This current spike can be limited with two different methods:
Slew-rate limitation using gate capacitor
Instead of switching on continuously, a PWM signal can be applied to the load switch, so that the capacitance is charged in several small quantities over a few milliseconds. Depending on the PWM frequency, this method requires stronger MOSFET gate drivers.
This method is also described in detail in the following blog post by Victron Energy.
The maximum continuous current the load switch can handle is mainly determined by the thermal limits of the MOSFET. The relevant temperature is the junction temperature
As the junction temperature cannot be measured directly, the following thermal model provides a method to estimate it based on the measured current flow and control the load switch accordingly.
The heat flow
For simplicity reasons, the on-state resistance is assumed to be independent of the temperature, so the worst case resistance should be considered for a safe estimation.
The dissipated heat is partly stored in the thermal capacity of the MOSFET and its surrounding PCB area, considered as a single thermal mass
Using an energy balance, the junction temperature
With the thermal time constant of the system
The time constant depends on the PCB design, the used components and attached heat sinks. It is normally quite small, in the order of a few seconds. It can be determined from a measured temperature response after a load step or estimated based on experience with similar designs.
As the circuit should be designed for a certain maximum constant current
The maximum junction temperature
This leads to the final temperature model equation:
For the implementation in the microcontroller, the previous temperature
Normally, there should be a fuse installed at the battery terminal to protect the wires from short circuits, which also protects the load output. However, even if the fuse rating at the battery is not much higher than the rating of the load output, fuses react quite slow. Typical automotive blade fuses for example need more than 10 ms to melt even for approx. 10x the rated current.
MOSFET datasheets normally state a so-called safe operating area, which shows the maximum allowed current vs. the drain-source-voltage. The following example is from the Nexperia PSMN5R2-60YL datasheet:
As long as the maximum junction temperature is not exceeded, this MOSFET can handle 100A continuously. However, under real operating conditions with passive cooling via the PCB, the continuous current will be in the range of 20A. During a short-circuit, the current rises very quickly, only limited by the impedance of the battery, the wires and the PCB.
According to this study, a dead short circuit of VRLA lead-acid batteries in the range of 100 Ah can reach currents >2000 A within less than 10 ms. So the MOSFET could get destroyed before any fuse reacted.
Inrush current when connecting capacitive loads
It would be a simple task to just switch off immediately as soon as a current above a certain limit is reached. However, this sort of protection creates false positives when connecting a load with a high input capacitance. A capacitance acts as a short circuit in the short moment of connection, but the current decreases quickly after the capacitor got charged.
The following simplified RLC circuit can be used to analyze the behavior of the system:
The wire inductance slows down the current increase and reduces the current peak during a failure. So the worst case is a short wire. The self-inductance of two parallel wires with a conductor radius
This results in an inductance of approx. 1 µH per meter for a pair of 4 mm² wires.
The following interactive graph shows the current vs. time with worst-case assumptions as default values:
- Total wire length of 1 m (calculations assume 4 mm² cross-section)
- A good quality large VRLA battery at full state of charge with 5 mΩ internal resistance plus another 5 mΩ contributed by the charge controller terminals and PCB.
- A low-ESR electrolytic capacitor of 1000 µF with around 20 mΩ ESR (e.g. Panasonic FR series)
The default values are realistic for what a typical 10-20A charge controller needs to detect and handle. In order to protect the MOSFET, the charge controller should switch off within less than 20 µs in case of the steep current increase of the short circuit. You can change the values to get an impression of your particular system and the general influencing factors on the maximum current.
Digital solution with fast ADC
In order to distinguish between inrush current for capacitive loads and an actual short circuit, the microcontroller has to take multiple measurements and not only consider the absolute values, but also the slope of the current.
For a fast detection of the steep short circuit current increase, a sample frequency of less than 5 µs is necessary. This is at the limit of what lower end microcontrollers can do, if the ADC needs to measure not only the current, but also other values continuously. In addition to that, the delay and bandwidth of the current measurement have to be considered.
Analog solution with filter and comparator
A comparator can be used to trigger a load switch-off signal as soon as a certain threshold of the current measurement signal is reached. As the comparator does not consider the slope or the length of the pulse, but just the absolute value, this approach can lead to false positives as described above if the signal is not filtered.
The current measurement output normally contains a low-pass filter (RC circuit). The filter time constant should be chosen such that the maximum allowed pulse caused by a capacitor just does not yet trigger the comparator, i.e. the damping at the frequency of the pulse is high enough to get an output signal below the threshold. In order to validate the correct filter layout, SPICE simulation of the circuit and laboratory tests should be used.
The comparator can be a dedicated part on the PCB that pulls down the MOSFET gate and switches the load output off. If the used microcontroller has an internal comparator, this can be used to trigger an interrupt and switch off the load output via software. In some MCUs, the comparator can also directly control an output pin without any interference of software.
If the battery voltage rises above allowed limits, e.g. because of a fault in the charger or because of transients, the load output should be switched off to protect connected consumers from overvoltage. This is a trivial task and will work reliably as long as there is enough margin to the maximum
However, switching off an inductive load like a motor or a relay can result in voltage peaks at the load switch. This is the reason why most charge controllers don't allow switching inductive loads at the load port.
In order to allow the current to decay slowly, a free-wheeling diode
A freewheeling diode in the charge controller should only be seen as an additional protection measure. Ideally, the diode should be located as close to the load itself as possible.