I read the one post by Nikolios87 regarding his use of negative ptc's that get cold when a charge is applied to it and that he could use all the advice he could get..
Well, I do have real life experience with them and also the positive Ptc's that generate heat.
I once contracted at a place in Michigan called Caltherm. What their goal was to produce a solenoid that operated on a wax that had copper in it to determine its melting/ start up temperature. The application was for thermostats in Giant Detroit Diesel Engines.
I had the choice of building 1 of 2... or both circuits to control the temperature on the PTC's.
The First Circuit was an analog circuit that was basically configured as a window detector. *cruise control circuit* It worked ok however its response time to temp change was delayed up to a min or better using this analog circuit, so it was latent in its response to input change for output change.
The second circuit was a digital circuit. Again, this one was also a window detector/cruise control circuit however it was digital and alot more responsive and precise on keeping the target between the window voltages set so it remained constant.. not hot then a little too hot, shuts off the cools then turns back on. **BTW all this can be applied to the negative ptc too. Just change the Ptc Type)
The digital locks 2 base comparisons right below and above and holds steady by employing as the dc power source to the ptcs not just a plain dc voltage and current, but a Pulse Width Modulation circuit that sent the power to the ptc in that form vs just a plain +whatever voltage it ran at.
My findings were that the digital circuit worked way way more accurately and responsively as opposed to the analog circuit.
The Digital circuit used less power then the straight dc analog version due to it utilizing the pulse width modulation being oscillated on a ac wave to the ptc with the proper voltage riding on it, Plus, the ptc operated a factor of times better than the analog circuit. Iirc the wave for the pulse width modulation is a square wave. if i still have the circuit for that somewhere around here i can scan and get it to you but i have not seen it in years so am not sure if i have it anymore. At any rate if you understand all of what i explained and can make a digital version with the PWM circuit then i think whatever load transference method your going to employ will just work that much better. My recommendation would be to take a couple of small turbine fans that are the size of what you could mount in a case, that sucks air from the outside in and then goes to a box assembly like you indicated. How i would make the box would be insulated on the outside completely... metal frame for the box i would choose aluminum as its easy to mold to your needs and attach aluminium fins to it encompassed around the metal box which contains the ptc]s mounted to thicker aluminum 4 times than the box itself.. make the box somewhat thick too but the case has to be thicker and the mount brackets for the PTC has to be even thicker. each piece has to be interconnected with some kind of metal faster device that allows the cold to travel from piece to piece and into the bracket which in turn is connected to the box in many places mounted in the center air flow area and allows the fins to sit and be angled slightly forward to catch the air and transfer the cold into the blowing air. Might want to employ a temp sensor that turn on a Solenoid that powers the mini turbine fan when it gets super cold and then let the fan turn on. Allows the unit to start cooling sooner than later by slowly cooling from air temp the metal on the air exchanger. ALSO! I would have a plastic catcher with a tube that goes out into a disposal place somewhere as on hot days you will probably accumulate some humidity dripping off the cooler exchanger just like a regular ac unit so make sure you put that in too to take care of any water that may drip into the pan. Would also have microfiber cloth filters on the output going into the case to catch any debris that may get sucked in and also water droplets.
I think if you do it just like i explained... it will work Fab!. Only other hurdle is the PSU for it. I personally would build one into it using a large solid state step down ac to dc rectifier rated at the out put voltage and current you will need to operate the PTC's, fan and perhaps temp circuit and solenoid for start up and shut off. that should be Independent PSU wise and other wise period. If ya got any questions feel free to shoot me a message! and as always.... GOOD LUCK!