Advances in solar array performance are envisioned: a) near-term: 150–200 W/kg, b) mid- to far term: 200–250 W/kg.

And https://www.nasa.gov/smallsat-institute/sst-soa/power

Flexible and thin-film solar cells have an extremely thin layer of photovoltaic material placed on a substrate of glass or plastic. Traditional photovoltaic layers are around 350 microns thick, while thin-film solar cells use layers just one micron thick. This allows the cells to be flexible and lightweight and, because they use less raw material, are cheap to manufacture. In 2014, FirstSolar announced a flexible solar cell design with an efficiency of 20.4%, closing the gap on single-junction solar cells. Flexible solar cells designed specifically for space applications are available from United Solar and have an efficiency of 8% (in 2014) on 1 mil polymer giving them a specific power of 750-1100 Wkg-1. 

Need to add some ROSA info here

So say 1000 W/kg = 1KW/kg = 1 MW/1000 kg = 1 MW/MT = 100 MW/Starship ... on a 1 mil polymer.

So I imagine large 10 m x 100 m rolls of this inside a Starship (1,000 m^2 each) - about football field size

Say this can be packed so this is effectively 2-3 mil thick ... this creates a footprint in the cargo bay (in landed vertical orientation) of 2m x 2m = 4 m^2. Given that the bay has 50 m^2 of floorspace lets assume that packing will allow for 40 m^2 of space ... so it can easily carry 10 10 m x 100 m rolls ... maybe 15 with more optimal packing.

So at least 10,000 m^2 of these thin solar arrays can be packed in a Starship, maybe 15,000 m^2.

To figure out the mass per kg of these thin solar arrays ... if you get 8% efficiency (as stated above) you get 60W per m^2 on earth ... and the material provides about 1000W/kg

So that material is 0.06 kg/m^2 -> 6,000 kg per 10,000 m^2 (packed Starship bay)

Clearly Starship is volume vs mass constrained ... this material could be 2-3 times more dense and there should be no problem in launching it and working with it. Rollers, deployers and other needed equipment could add another 10-20 mT for this Mars Solar Powerplant. 

The solar flux onto a 30 deg tilted solar collector at 40 deg on Mars is at least 200W/m^2 over a 10 hours a day

If you get 20% efficiency (I assume the 8% will be 20% in 2024) its 40W/m^2

So an average of 400Wh/m^2 day per on Mars at mid-latitude, assuming no dust issues.

At 10,000 m^2 you have 4MWh per day ... At 15,000 m^2 you have 6MWh per day ...

For deployment one would need a "laydown" machine (LM) to accept rolls from the Starship and take them to a hillside pointed in the right direction, and roll them out (see below). Proximity of a hillside might be part of the base site criteria. The LM would need to also lay cable and connect them up to the core cylinders that contain the needed electronics. Small NASA heli-drones would come by and blow dust off if needed. We might go with 5m high rolls instead to reduce the size of the LM.

Why the first Mars power grids will be Low Voltage Direct Current (LVDC):

1) Mars power will be primarily multi-MW scale solar (DC) buffered by lithium-ion batteries (DC). There may be a backup MethLOX generator (fueled by stored MethLOX from the MethLOX factory) for long dust storms which will be AC. That AC -> DC conversion will likely be no more 93% efficient.

2) As a Mars colony distribution grid will likely be within a 3 - 4 km radius circle, AC's advantage of long distance transmission does not apply.

3) DC power cabling is more mass efficient than AC with Heavy Duty AWG 4: 0.2 kg/m, 200 kg/km. The power distribution system cables mass may be as low as 2 mT.

4) If needed, DC -> AC (97% efficient) conversion is more energy efficient than AC -> DC (87-93% efficient). Changes in DC voltage are very efficient at 99% (and the equipment does not appear to be large or heavy).

5) While DC -> DC power conversion equipment is more expensive per KW than AC -> AC, that price different won't be significant in the cost of a Mars colony. DC -> AC Inverters will be needed in edge equipment that does not have a good DC alternative, and at the MethLOX Factory to convert AC -> DC for emergency backup power during long dust storms (of durations that exceed the the capacity of the distributed battery network).

6) Using Tesla Powerpack 2 4HR (used in Power Grid application, Capacity: 210 kWh, Op Temp: −30 to 50 °C, Mass: 2,160 kg, Size: 218.5 cm × 82.2 cm × 130.8 cm provides 960 VDC 66 A 55 kW) the 2 mT battery mass looks like the main limitation. Assume 2100 kWh ... so 10 of these (20 mT total) distributed around colony. Note that batteries may be based placed within areas that are human occupied since the operation temperature is only as low as -30C, which is much higher that average Mars surface conditions.

7) It exists now: LVDC has already been adopted as a medium of distribution in many applications such as data centers and telecommunication power systems.* With the rise of residential solar power DC is becoming the most efficient choice even for the mixed DC/AC home.

8) More standards are emerging and lower cost equipment to meet them. Emerge Alliance, together with the partnership between Japan and South Korea’s governments, with academic institutes, and industrial groups, have also made a significant development on the standardization and technological development of DC-powered buildings and microgrids. So far, the standards for using DC with AC in hybrid settings in commercial buildings are established by the Emerge Alliance, which uses 48 V DC for low-power applications and 380 V DC for high power applications.* Note that 48 V, 220 V, and 380 V DC are the default voltage choices.

References:

* DC distribution for residential power networks—A framework to analyze the impact of voltage levels on energy efficiency: https://www.sciencedirect.com/science/article/pii/S2352484719313757

Why DC cable is better than AC: https://www.zmscable.com/new/The-difference-between-DC-cable-and-AC-cable-in-power-cable