First some background: I am a mechanical engineer and work with heat transfer and mechanical systems. I also suffer from acute ODIYD (Obsessive Do it Yourself Disorder)
In 1990 I installed the A-B cold machine 200 as an ice box conversion in our Cal 31. Before doing it I did some ice melt tests to check out the insulation. After a first test recording ambient temperature, etc. I upgraded the insulation as much as I could using Styrofoam blue board and Great Stuff expanding insulation. This improved the insulation capability by about 38%. The A-B install was not too difficult. We used the large vertical evaporator mounted toward the back of the box near to top. It worked well and could keep the box cold for 2-3 days on our house bank of 2 golf cart batteries. The freezer part could make ice using the included vertical trays and could keep meats frozen. Ice cream was kept to the consistency of soft-serve. This worked well for the 10 years we had the boat.
In 2000 we bought a Catalina 42 that came with a Norcold system of 2 compressors and 2 evaporator (flat plates). It was an energy hog, used R12 and had no freezing capabilities. Hated it. My plan was to build up a custom system using components from Rparts.com. They use the same Danfoss compressor (BD50) that the A-B system uses. They did not have any standard evaporators that would fit our ice box shape very well. One of the members of our club was a retired refrigeration tech and had built his own evaporator system. After picking his brain and reading everything I could get my hands on I ended up building an evaporator coil from copper tubing with 5 sides that fit into the bottom of our box. The standard evaporator plates used by A-B and also from Rparts use a capillary tube like home refrigerators. The sizing of the tube is not a "calculate it and do it" exercise so I went with an expansion valve. Standard systems don't use these because of the cost and complexity and they can optimize their capillary tubes for the systems. I added some additional internal insulation to the bottom so we ended up with a 0.75 cu ft deep freeze (0-5°F). That is what the compressor controlleds. There is a pump-up fan that is controlled with a separate thermostat to maintain the refrigerator at 36-38°F. Another small fan run continuously the avoid stratification. (more later). The system uses 30-35 amp-hours/day in the pacific northwest. We can run 3-4 days on the 4 golf cart battery house bank before using the recommended 50% of the battery capacity measured by the Link 10 battery monitor. I also used Great Stuff foam wherever I could to maximize the insulation and covered over the ice box drain and insulated the area.
This was not a simple project but fun and successful. For a project like you are describing and based on my experience I would recommend:
-The A-B systems are fine. I cannot comment on other system manufacturers since I have not used them. Pick your evaporator to fit you box.
-Do everything you can to improve the insulation. I used Great Stuff with a piece of vinyl hose to access all the nooks and crannies I could reach.
-Put in a circulation fan. I mounted a small CPU cooler to the top of a piece of 1" light wall PVC that goes from near the bottom of the box up to within about 2" of the top. It blows downward and stirs the air. I took some temperature readings and there is only a 1-2° variation top to bottom. The little fan pulls very little current - about 3 a-h/day. Here is a link to one possibility. Make sure to get ball bearings.
http://www.newegg.com/Product/Product.aspx?Item=9SIA73M39H4553
-We mounted the compressor/condenser in a high part of the bilge, hopefully outside the flood zone. Our water up here seldom exceeds 70°F so the area is always cool which helps the efficiency.