| Electronics > Projects, Designs, and Technical Stuff |
| thermocouple junction area/diffusivity? |
| (1/2) > >> |
| coppercone2:
So when you melt thermocouple wires together, you usually do it in a little bead with a hot torch. When you melt the two metals together, I assume you are basically fusing them with some penetration of 'dendrites' into the dissimilar materials that hold it together and some alloy formation from diffusion, so you get a composition mix gradient and longer salients that may not have as much mixing. I imagine its kind of like yeast on a small scale. If you take the thermocouple to absurd dimensions, will it work? Say two examples: 1) giant disks that are thin and bonded together along the surface where the bond is traditional in regards to penetration/alloying but it is just mixing 2) you weld two big cylinders together some how, say forge welding, where there is more diffusion/thicker inter metallic layer forming and possibly like propagation of softer metals into formed cracks If standard materials are used, how will their behavior deviate from normal expectations? I am having trouble imagining the behavior of this '3d' model. Could you possibly get better behavior by separating the dissimilar materials by a known distance and using silver to bridge the gap rather then fusing them directly so that weird alloys are not formed? |
| Kleinstein:
The important point in a thermocouple is not the junction, but the part of the wire that sees the temperature gradient. Ideally the junction is in area with little gradient, so this is the area that does not matter how the alloy actually is. So it is also possible so solder a thermocouple - if soldering works and the temperature is not too high. The extra solder in between does not alter the temperature reading. Similar the cold side usually also is with 2 junctions: on both sides the contact from the thermocouple wire to copper. |
| coppercone2:
I want to know from a microaccuracy standpoint in regards to physics not what works are you saying even if you managed to get the best amplifiers ETC it would not matter so long there is conductivity there? And the main idea is thermal symmetry, i.e. flat bond that's exactly 50/50? In that case there is a inherant error due to the different thermal conductivity of the wires and how much heat they add or extract from the system? can this be eliminated by varying the thermal mass of the wires? since they have different compositions and different thermal conductivity you don't really have symmetry there. |
| coppercone2:
do you just consider the intermetallics/dendrites to form a bunch of other thermocouple voltages that are proportional in magnitude to their thermal mass which is microscopic in comparison to the thermal mass of the heat effected zone ? how would you mathematically quantify their effect? can you flatten out the junction to be very thin, shift things around and consider them series junctions? |
| Kleinstein:
The junction zone does not matter, as long as there is no temperature difference. It is the physics that shows that the thermal EMF is coming from the regions with thermal gradients. Even a tunnel contact should be good enough, though maybe a little high impedance to get low noise. There is no need for thermal symmetry at all. Thermal conductivity only matters in that it can lead to a thermal gradient near the junction. The really unusual things come up, if the range with thermal gradient is a non cubic single crystal or highly textured material with less than cubic symmetry. In this case the thermo-power is no more scalar but a directional property. |
| Navigation |
| Message Index |
| Next page |