[time-nuts] SMPS or conventional?

[ed breya]

Regarding linear versus SMPS, I always prefer linear where possible. As 
others have said, for assessing a SMPS, try it and see. If it does the 
job sufficiently without too much noise, then go with it.

Regarding Rb cooling, I don't know about the newer, smaller ones - I 
only have three big old Efratom M-100s - but if they have provision for 
heatsink or chassis cooling, then use it. The MTBF info Neville put up 
shows the benefit. As far as I know, the "ambient" temperature sometimes 
seen in Rb specs is actually the base plate temperature, assuming use of 
a heatsink large enough that they are roughly the same.

Each of my M-100s came with a very nice black-anodized, extruded Al 
heatsink, perfectly fitted, hardware-wise. In my portable reference 
project, I put the Rb on the inside of the Al back panel, and the 
heatsink opposite on the outside, for convection cooling, and of course, 
a generous film of conductive goop between each interface surface. The 
screws go through the heatsink and back panel, into the base plate, so 
in reality, the back panel is just pinched in between.

The other two units are planned to be together in a dual system, with 
both mounted on a single, heavy Al plate, which is mounted on rubber 
isolation hardware. The thermal plan is to surround everything except 
the plate with foam sheet insulation, and let them self-heat to 
something a bit above the highest expected ambient temperature - maybe 
40-42 deg C. When this range is reached, a variable speed blower will 
feed cooling air to the plate through ducting, to provide some degree of 
temperature regulation, depending on how well I design the arrangement 
and its control system, and squeeze everything into the limited space. 
In anticipation of too cold an ambient temperature, there are also 
heater resistors on the plate for just in case, and of course a thermal 
cutout switch to shut it all down if things go very wrong. Fans and 
blowers are the weakest link in this sort of stuff.

This is all experimental, and only partially built, into an old rack 
style HP instrument carcass. Some day I may finish it. The best approach 
though, would be true regulation at some lower temperature for better 
MTBF, like with TEC heat/cool capability, but this gets to be much 
bigger and more complicated.

Ed