Notice the order in which the specs of the cameras are given by NASA here:
- Location on the rover
- Mass (which is always critical on a space probe)
- Power (energy consumption)
- Data return (the amount of data generated by the device)
- Color quality
- Image resolution
Function obviously includes an expectation of long term reliability. Mass, energy consumption, and data generated are also listed before the measurable performance in terms of color and resolution. The same list of attributes is probably followed in the same order for the descriptions of the other sub-systems that are part of Perseverance.
In the end, following mass, which is most critical in terms of being able to place a thing somewhere other than Earth where one wants to use it due to the energy costs involved in getting it there, almost all design decisions about interplanetary devices like Perseverance, whose life expectancy depends upon solar panels and rechargeable batteries, comes down to energy management vs. acceptable performance.
- The ultimate durability of the entire system (i.e. the entire rover) is dependent on preserving the energy supply system for as long as possible.
- The less energy each subsystem uses, the less stress it places on the energy supply.
- The less energy each subsystem uses, the greater the number of total subsystems that can be used for the same power budget, increasing the overall capability of the total system.
- If two competing sub-system designs can get approximately the same results and have the same durability, the one that consumes less energy will get the nod.
- In addition to consideration for how much energy it takes to capture a photo, with a camera on the surface of Mars there must also be consideration for how much energy it takes to transmit the data collected within the photo off the surface of Mars and back to Earth.
The last point above is the key one here.
Using a lower resolution sensor allows for lower energy consumption when lower resolution results are satisfactory. When higher resolution results are desired various techniques such as slight movement of the sensor or of the optics in front of the sensor in successive exposures can be used. Many of the cameras placed on similar probes in the past have the ability to shift the sensor slightly so that greater detail can be obtained using multiple exposures. The mast cameras have lenses with the capability to zoom in and out.¹ If high detail of a distant object is required the lenses increase focal length to make the object larger as projected onto the sensors of the stereo camera. This ability to zoom removes the need for a very high resolution sensor so that details of more distant objects can be seen in very small parts of the total image.
The higher computing power, and thus higher energy consumption, required to process the results of multi-exposure images can be terrestrially based after the results of each sensor actuation has been received on Earth. Ditto for the computing power needed to stich images together to form larger panoramic views. Ditto for any advanced color processing algorithms to produce "true color" results.
An image sensor of a given size isn't affected much in terms of mass by how many photosites it is divided into. That is, I'd be very surprised if a 24MP sensor with the same total surface area as a 2MP sensor using the same generation of technology has any increased mass compared to the 2MP sensor. Where it does differ is in the greater amount of data generated per photo. With the cameras most of us use while holding them in our hands and then taking the results home to transfer the results to our computers, the more data we collect the better, right? That's because the energy cost to us from using a 24MP camera instead of a 2MP camera is trivial compared to what the energy cost means to a designer of a Mars rover who needs to transmit that data via radio from the surface of Mars to Earth. The radio relay satellites orbiting Mars also have energy budgets that are affected by the amount of data per image.
Beyond the energy cost per photo:
When very high tolerance manufacturing is involved, as it is for any interplanetary space probe, a sensor with larger photosites also allows for fewer potential defects due to the increasing challenges of manufacturing chips using smaller and smaller die sizes which are needed to produce sensors with smaller and smaller photosites. Smaller die sizes do allow for lower power consumption, though, or at least they do with computing chips which don't suffer performance-wise from smaller total area. Ultimately the designers will choose a point where the miniscule gain in energy efficiency no longer outweighs the advantage of fewer defects in manufacturing. The ability to have a high reject rate probably also comes into play. When the design and engineering costs far outweigh the manufacturing cost for such a very limited number of sensors, the cost of making several orders of magnitude more examples and testing to find the one(s) with the fewest defects doesn't add that much to the total cost of the product. What's an extra $10-20K in the cost of production when you've already spent a couple of million designing the camera and it's going to cost a few tens or hundreds of millions more dollars to place the thing on the surface of Mars where you plan to use it?
Larger photosites may also make the sensor more resistant to the environmental challenges encountered by a space probe. Even though Perseverance is now safely on the surface of Mars and protected to a degree by the very thin Martian atmosphere, it spent months in the near total vacuum of space while journeying from Earth to Mars. If cosmic rays, for instance, strike a sensor it can affect the surface of the area where the particles impact it. If such a strike affects only 5% of a given photosite's total area, it would probably have less impact on that photosite's performance than if the same amount of surface area made up 50% of a smaller photosite's total area. Extremes in temperatures will also stress micro-electronics. Larger photosites and larger die sizes should be more resistant to such thermal stresses. Vibration resistance should also be better for larger photosites than smaller ones.
¹ The actual focal length range is 28-100mm, according to NASA. The sensor has a 4:3 aspect ratio with a diagonal of 14.8mm, giving it a roughly 3X crop ratio compared to FF. So 85-300mm FF equivalent. Thanks to BobT for the links!