Issues with the load share controller have come back to haunt us. We need to figure out if there’s a way to reduce the power loss we get using the load sharer or an supply ORing component.
A load sharing device would equally distributed the current drawn from each solar string. This is a major problem if one solar panel is facing the sun and another is only partially exposed. The power system would be limited to draw the same low current value from each solar string, resulting in major losses.
ORing Diode devices prevent current from flowing from one solar panel into another due to a voltage difference when both are connected in parallel. They use N-channel MOSFETs to replace Schottky diodes which saves lots of power. The power systems with MPPT studied in [1] and the power architecture in [2] use ORing Diode Protection.
The major risk with ORing Diode Protection is that if the output of an MPPT, that is not generating the most power has the highest voltage, that MPPT would be powering the satellite. Solar panels are sort of a special case, where over most of their operating conditions they act similar to ideal current sources, rather than ideal voltage sources which are more common [3]. Ideal current sources can be connected in parallel easily because of Kirchhoff’s current law. However, with each solar string having its own MPPT, they act more like voltage sources.
Figure 1. Typical I-V curve for solar panels generating different amounts of power.
Connecting all solar panels in series would not work because if one solar string is shaded, then the current would be limited. Virtually no CubeSats connect all their solar panels in series.
The MPPT which outputs the highest voltage (the voltage difference will be very small) will dominate and supply all the power. In many voltage sources, the dominant source’s output voltage would fall and eventually equalize with the lower output voltage sources.
The graph in Figure 2 shows two voltage sources with slightly different open voltage values (at the y-axis). When the current is increased, to Imax, the voltage of the parallel bus is equal to the drooped voltage from the higher voltage output source. Because of the droop, this voltage is less than the open voltage of the other (non-dominant) source. This allows the other source to contribute to the output current. The technique is simple, but does not perfectly get the maximum power from each panel. Though I believe it is much better than getting power from just one solar panel. Especially if the CubeSat is angled such that a corner is pointing to the sun and half of each solar panel is directly exposed.
Figure 2. I-V graph for two parallel voltage sources with slightly different output voltages [4].
With the current choice of MPPC (LTC3130) the datasheet does not show the how much the voltage falls when the load current increases. If we know how much this is, and how far apart the output voltages of the MPPCs are then we can figure out if additional components is required at all, or if we can just connect them in parallel. If the output voltages are too far, and the natural voltage droop is insufficient, then the simplest solution would be to add resistors to both ends to increase the droop. However, this inevitably leads to power loss.
Another solution would be to avoid the LTC3130 altogether and use a MPPT + Charger IC. These ICs such as the BQ24650 [8] do not set an output voltage during most of the charging cycle. Instead they mostly operate using a constant current charging phase, which allows us to OR the MPPTs using ideal diodes following Kirchhoff’s current law.