Second, conduction: molecular electron excitation via contact. Convection: moving along with fluids (air or any gas is actually considered as "fluids") forming "currents", optimized by the proper airflow, both speed and quantity, as well as the layout of the fins, which need to be as large a surface area as posssible. In basic terms thermal transfer relies on 3 separate forms. I also expect that 3000 grit lapping is a lot cheaper, and good enough. I expect that there may be an optimum wedge shape for maximizing surface contact, and capturing heat energies. But that is only theory.Īccoustical studies done way back when the first anechoic chamber was built, they found that there the shape of the wedge, it's angle, made more impact on sound absorbtion than the material.įor lasers, there is a specific angle of about 37 degrees that end glass is placed at so as not to reflect, nor accidentally capture any of the photonic energy. If the lapping is done in the same direction on both parts, it is possible that the peaks will fit into valleys, actually increasing the contact area. Two highly polished and exceptionally flat surfaces will may have greater contact, or may not. Heat transference is a function of surface area. Heat conduction is a function of cross-sectional area. The finer the grit, the shallower and fewer the valleys, the less sharp the peaks, the more difficult to effect a direct metal to metal contact, when using compound. The hypothesis for why the difference had to do with the metal's ability to pierce the compound and make actual contact. The discussion was over what grit to use for lapping. When using compound, the perfect mirroring was worse. A guy had done some research and found that a "perfectly" mirrored surface did not work as well as a not a perfect surface match.
I remember a thread about this last year, either here or on some other forum. You're better off organizing cables and installing a filtered airflow system.
Mirror paste not working clo3d full#
Breaking your back to lose a degree and a half Fahrenheit at full load is not worth it. This apparently makes up for the marginally increased distance between the CPU die and the HSF assy.īasically, do what you want to do, as long as you apply thermal paste, correctly mount the aforesaid HSF assembly, and provide adequate cooling within the case, you'll be OK. Essentially the argument for the latter states that if you coarsely grind the HSF in a circular pattern, then the surface area in contact with the thermal paste is greater. In fact, the merits of lapping as opposed to simple circular grinding the HSF have been discussed before. Arctic Silver) then this isn't necessary. Obviously if you're using a thermal paste (i.e. The crucial point has already been stated, that mirror polishing brings as much of the HSF assembly into contact with the CPU die as possible. Assuming that you're not placing your HSF assembly four inches away from the CPU die, you're transferring heat by conduction, which therefore overrides any sort of radiative effects of heat transfer.