Re: OLTC primary loss measurement

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: "tesla by way of Terry Fritz <twftesla-at-qwest-dot-net>" <tesla-at-paradise-dot-net.nz>

Greetings Jim et el

I think you rasie a very valid point regarding the variability actual
thermal resistance of the sink to air figure and this would certainly lead
to significant errors. I do suspect that heatsink degC/watt figures are for
a set of defined conditions however and would apply if they were accurately
reproduced in Terry's lab (and thats probably not possible)

I have a friend who has worked professionally assessing heatsink efficiency
and modest setup or environmental changes do lead to significant changes in
degC/watt for a given sink.
But in controlled environments ie stll air and limited room gradients good
repeatability is possible in his experience.

A simple method (short of full calorimetric measurements) to get more
accurate results might be to run pure DC thru the IGBT array and to
calibrate the degrees centigrade per watt figure for the system. This could
be done at about the same power loss level expected for the operating tesla.

While the system was still warm the operating tesla system could then be
substiuted for the DC.

If draughts are excluded and orientations of elements are not changed
reasonable accuracy could be obtained.

All this said as the IGBT losses are believed to be small compared with
other losses the matter may not be worth too much effort.

Rgds
Ted L in NZ

> >Few Further thoughts on last post re this:
> >
> >1) The only thermal resistance you need to know is heatsink to air if you
> >measure heatsink temp. This is normally published for each heatsink type
>
>
> But that number is only a "design guide", used for sizing the sink in the
> first place, not an actual measured or reliable number.  A huge amount of
> things can affect the sink<>ambient resistance.. small amounts of air
> movement (as from thermal gradients) dramatically reduce the thermal
> resistance.  If you think electromagnetic stuff is complex, it's nothing
> compared to heat transfer, where the transfer medium is non-linear, has
> varying flow regimes (turbulent/laminar), and all sorts of other weird
effects.
>
> Putting it in a big box (to reduce the effect of external air currents)
and
> doing a power replacement calorimetric measurement would be your only hope