Before making a radical change to the PA transistor (BS170s) thermal design it is necessary to fully understand the the thermal parameters.? As Hans ha pointed out:
- The six layer board conducts heat very well and is designed with lots of copper in each of its layers.? You may of noticed how much harder, time/temperature, it is to solder parts than the QCX and QDX kits.
- Under the flat surface of the BS170s is a large heat sinking layer.
Background:
? ?BS170 Thermal Resistance
? ? ? ? 220C/W? ?(package only)
? ? ? ? 175C/W???(plated through-holes)
? ? ? ? 145C/W???(plated through-holes and?0.25?inch sqr ?per lead)? With the multi layer board this is probably equivalent.
It appears that the thermal path from a TO-92 (BS170) package is lowest through the pins. This is demonstrated by the significant improvement to 145C/W with good pin heat sinking. (From my investigations it appears that more than 50% of the conduction is through the pins). FYI conductivity of copper is ~400x that of epoxy.
With a fast electrical transient thermal capacities may attenuate any observable rise outside the transistor chip before it fails.
Adding an additional structure under the transistors is questionable:
? ? ? ? Added interfaces to between the transistors and the PC board/case primary heat sink
? ? ? ? In the cramped area of the heat sink there will be little air flow.
? ? ? ? I would strongly recommend against adding anything between the pins and the board.
Test
Good engineering practice mandates measuring the temperature of the current heat sink under the BS170s and comparing the temperature of the proposed heat sink.? Must be done in a full assembled QMX.?
