(I tried searching past posts, but either I'm totally missing the appropriate keywords, or this hasn't been discussed much.)
Clearly a traditional hardwired system like LiteJet would have most of its heat contained in the panel. I'm assuming that On-Q/iLine would be comparable with UPB/Insteon as far as how much heat is contained in each j-box.
The questions then are:
1) It would seem like it would be better to contain all the dimmers/relays/controls in one panel that could have auxilliary cooling applied if necessary, or would you argue that it is better to distribute this heat load to all different places in the house, assuming you observe any necessary derating and such to keep from setting your walls on fire?
The 40C rating is presumably at full rated output, at which the heat generated by the dimmers is at a maximum because current through the dimmers and resulting "IR" ( =amps * volts) power loss is at a maximum. Either derate the dimmer panel if you anticipate greater than 40C = 104F -- or (only partly tongue-in-cheek) turn down the lights!
The heat dissipated by the solid state dimmers is small compared to the heat generated the lights themselves at high light output. Approximately:
Dimmers: 1.3 volts * amps ~= watts dissipated
Lights: (115- 1.3) volts * amps ~= watts dissipated
One puts lights (and the >98% of the thermal load) where one needs illumination. Put the dimmers (and their <2% thermal load) wherever dictated by other factors. See below (fourth paragraph from end of this post) for more discussion on the relative heat dissipation.
2) If the AC goes out in the house, do you still answer the same to question 1?
Yes, assuming AC = Air conditioning and not Alternating Current.
3) Is my assumption that On-Q, iLine, UPB and Insteon are all comparable when it comes to heat output correct, or are some known to be more/less efficient than others?
Most dimmers use phase-controlled TRIACS or back-to-back SCRs. The latter are particularly common in the Solid State Relays (SSRs) typically used in panel-based (centralized) dimmers. All are comparable with respect to the amount of heat that needs to be dissipated.
Some more recent panels use reverse-phase dimming and potentially more efficient output devices (MOSFET's) and IGBT's (Insulated Gate Bipolar Transistors ). If there aren't yet reverse-phase MOSFET and IGBT wall-mounted dimmers, I assume that there eventually will be -- in part stimulated by stricter codes and regulations concerning creation and propagation of electronic noise, and in part because of potentially higher efficiencies and reduced heat and resulting need for de-rating multi-gang, wall-mounted dimmers.
4) Just how toasty does it get in and around the panel in a hardwired system? Lite jet lists the upper operable limit as 40deg C. That seems high enough until the AC goes out...
Recognize that air conditioning typically only reduces ambient temperatures by say 100F -70F = 30F, and that 30 F is small compared to the delta T (Junction-Ambient) of the semiconductor junction temperatures. But yes, ambient temperature absolutely needs to be considered in design.
The total maximum heat dissipation of a panel might be conservatively estimated by multiplying the (total maximum amperage of all VAC power supplied to the panel) x 1.5 volts. So for 3-phase x 20-amp panel ( a common industrial/architectural configuration) the total dissipation would be 3 x 20 x 1.5 = 90 watts. Compare this to the implied wattage of the lights themselves: 3 x 20 x (115-1.5) = 6810 watts.
Assuming that no heat is lost out of the panel through the back and sides (eg, built into well-insulated wall), 90 watts is the amount of heat that would need to be dissipated (long term) through the front (eg worst case) of an enclosed panel.
One can calculate the temperature rise inside the panel based on empirical or physics-based formulas if the amount of heat created (what we just estimated), outside temperature, and the panel-to-air thermal resistance is known or estimated.
Here's my home-brew dimmer panel which measures 24 X 18 x 6 inches:
Mounted _on_ a masonry wall, it can dissipate beaucoup heat compared to switch box installed _in_ a wall with a plastic (poor conductor) cover plate. Note that many industrial and architectural dimmers and some large residential dimmers have aluminum faceplates/heatsinks to better dissipate heat.
The black object in the center to which the SSR's are mounted is a large extruded alluminum heatsink that is open at the top and bottom. There is a fan operated by a 50 degree C bimetalic thermal switch that to my knowledge has never been triggered.
Compare this to the amount of heat that has to be dissipated through (mostly) the front of a multigang wall dimmer and the need for derating of dimmers placed side-by-side in a much smaller switch box becomes obvious.
The preceding was a more direct answer to rmac's first question than offered earlier in this response.
If one is concerned about dissipation from a switch box dimmer configuration, one can simply derate the dimmers even more than recommended/required. Because dimmers are sold in incremental amperages, as a practical matter it means buying the next higher size or leaving one more empty space in the box.
That said, if I were worried about high ambient temperatures, I'd prefer to have the loads in a panel rather than crammed into a switch box but its a third-order concern in a typical home situation. There may also be countering factors such as availability of space for a utility closet with adequate ventilation/space conditioning.
Central panels also have the advantage compared to switch-mounted dimmers of providing a much larger volume compared to wall-mounted dimmers that is potentially available for inductors ("chokes") to reduce EMF and other electronic noise.
( EMF has been even been credited with producing "headaches, tiredness, poor short term memory, fatigue, depression, nausea, rashes, irritability,[and] even chronic fatigue syndrome " -- along with other more probable, conventially recognized side effects ;-) Fer grins, visit http://www.lessemf.com/suppress.html
HTH ... Marc