How hot do you get, really ?

There are thousands of postings on heatsink temperatures and dissipations but nobody seems to also add info on actual mosfet temperatures. I get about 45C on my Aleph 5 which is very cool by most temps mentioned here and I was wondering if it can handle a raise in bias. However what worries me is that some of the mosfets when measured directly, on the hottest middle pin, are running at around 90-100C.Is such a difference normal? So how much more could they really take and be reasonably reliable ?
OK guys, what temperatures do you get on directly on your mosfets compared to heatsink temperatures?
First off, don't make the junction temperature go above 150 deg C. The middle pin is a reasonable approximation of tab temperature if the wire soldered to it is not drawing away too much heat, otherwise most brands of fet have a notch on each side of the case where you can get at the junction side of the tab with a thermocouple. Use a bit of heatsink past on the thermocouple. Second, look up the thermal resistance of junction to case (Rthj-c or similar) It will tell you how many degrees per watt the junction will run above tab temp. For a TO-247 device the Rthj-c would be in the range of 0.25 to 0.7 or thereabouts. So if you had a device with a Rthj-c of 0.7 deg C per watt then at 10 watts dissipation the junction would be 7 degrees above the tab temp. Hope this is of some use.

GP. - Professional temp-measurer (among other things) of big-slugger switchmodes.
Yes, but isn't 150C the absolute maximum beyond which you get break-down ? Are the fets meant to be run so close to max and how does reliability suffer? I mean if you run at 100C and the fet has let's say 20,000 hrs projected life span , what is the life span at 130C?
According to my calculations my heatsinks are dissipating normally. They are 0.33C/W and are dissipating heat from four mosfets at 32V X .55A= 17.6W each so 4 X 17.6=70.4 W let's say 71W. So 71W x .33C=23C above ambient. Since ambient in my place is around 22C , 23+22 = 45 C which is exactly what I'm getting and which proves I think that I've got the mosfets correctly positioned and fitted despite them being on an aluminium extrusion which is then coupled to the heatsinks. Or are there other parameters I forgot?
protos said:
Yes, but isn't 150C the absolute maximum beyond which you get break-down ? Are the fets meant to be run so close to max and how does reliability suffer?

In the RCA Receiving Tube Manual their tubes are rated by two methods - CCS and ICAS, that is "continuous commercial service" and "intermittent commercial &amateur service. This recognises the extra reliability you would get if you rate something conservatively. Conversely, if you run things a bit harder as in ICAS it doesn't mean it is going to quit soon. In fact with solid state stuff it may last only 50 years instead of 350. A diy amp for domestic use (I'm presuming that's what yours is) is similar to ICAS. Lives don't depend on it and the amount of hours logged isn't comparatively great. I don't expect you would ever have a failure in practice unless you made tens of thousands of amps and kept statistics.