Re: Optical pyrometers, was R = V/I ?

[Mark Sylvester <msylvest@SPIN.IT>]

At 14.29 09/05/00 -0700, you wrote:
> >These limitations are recognised... and there are others. We compensate for
> >the conductive/convective energy loss by extrapolating the vaguely linear
> >start of the power - T curve as a straight line, and then for several T
> >values we take the difference between the curve and the extrapolated line
> >below it as Pr, the radiated power. The glass envelope is kept immersed in
> >cold water to justify neglecting the energy radiated back to the filament
> >from the surroundings (I must check sometime to see just how much
> >difference this makes).
>
> You are to be complimented for even thinking about the convective heat loss.
> I have no idea how important it is, but the conductive and back-radiated
> terms are surely negligible compared to the radiative loss. The fact that
> convection occurs can be seen in old light bulbs that have burned in a
> constant position for a long time. They develop a dark spot of deposited
> tungsten at the top of the envelope. The tungsten got there by convection,
> of course. I would skip the cold water bath. The difference, if any, will
> be due to the convection process, not the radiative. The best way to check
> this is to do the measurement with and without the bath.
>
> >Strangely, the result from the log(Pr) vs log(T) graph is very consistent:
> >it invariably comes out as a straight line with slope somewhere between 3.5
> >and 4.5. A serious error analysis is probably too hard at IB level.
>
> It has to behave in some way, and radiation is the dominant process. What
> value do you get for the Stefan Boltzmann constant? What is your highest
> inferred temperature using your linear resistance model thermometer? What
> is the melting point of tungsten?

The highest inferred temperature is typically under 2000K, overrunning the
rated voltage slightly. This is well below the mp of tungsten which is
somewhere over 3000K.


> >I regard the lab as a valid exercise: it could be improved by trying to
> >quantify better the difficulties that we bridge with simplifying
> >assumptions. I do have the feeling that there may be bigger errors than we
> >think, which compensate for each other to give a result which is a bit too
> >good.
>
> As with the light bulb demonstration I described last week, I think there
> is much to discuss here, and the discussion will be valuable. I just don't
> like to see students told something we know is untrue. In this case the
> tungsten resistivity data are available. I haven't searched very hard to
> find (in Kate & Laby) the values I did find. If you find values for the
> complete range then you probably can use the resistivity of tungsten as a
> thermometer with much greater confidence.

I have to agree, as mentioned in previous post.


>  You must still take a look at
> the emitting area problem to get a value for the Stefan Boltzmann constant
> but you may then have something to say about convective losses.

I've regarded the problem of the filament area as too hard, so we've never
calculated a value for the S-B constant. Apart from finding the length and
thickness of the coiled wire, there is the question of how the coiled coil
behaves: I imagine it is in part like a hot cavity inside, with not all of
the surface facing outwards.

Mark






_____________________________________
Mark Sylvester
United World College of the Adriatic,
34013 Duino TS, Italy.
_____________________________________