Measurements on a 1206 resistor (actually a platinum thinfilm RTD, to be precise)
theta, K/W thermal tau
in free air, 275 6 sec #30 ww leads
on pcb, normal 69 1.5 s kinda traces
on pcb, about 35 0.5 s
Do these numbers seem reasonable? Anybody got data for other sizes?
John
A quick look at a range of trimmable Vishay resistors
throws up about 220K/W for 1206 mounted on a PCB, 425K/W fpr 0805 and about 600K/W for 0603.
The laser-trimmable area is on the top of the chip, which may explain the difference between your 70 and and their 220, or they might be doing something evil, like using beryllia rather than alumina - though you'd get a government health warning on the part if it did.
The performance of Pt resistance sensors is fundamentally determined by their thermal resistance to ambient (as I mentioned in my Peltier paper, with a reference back to Larsen's classic paper) so the manufacturer may make an effort to minimise this thermal resistance.
The thermal time constants sound reasonable. Data sheets on temperature sensors often offer this sort of information.
I would expect that a 1206 would thermal very poorly. The aspect ratio sucks and so does the wing loading.
On the bright side, you won't need a dethermalizer.
This looks like a straightforward linear decline in tau with increasing thermal conductance--the first two values give 4x lower tau with 4x lower thermal resistance, to about 1% accuracy. The last one deviates from that behaviour, but that seems to be because the transient behaviour is governed by conduction into the copper, whereas the steady-state behaviour is governed by conduction through the FR4, which is very much poorer.
The thermal mass is primarily due to the alumina substrate, because ceramics have a much higher specific heat capacity (c_p) than metals, and there's more ceramic than metal in a 1206 even counting the solder fillets. Alumina has a c_p of 850 J/kg/K and a density of 4000 kg/m^3, so a 1206 (3.2 by 1.6 by say
1 mm) would have a thermal mass of about 0.07 J/K. With a thermal resistance of 275 K/W, that gives a time constant of 19 seconds, so I conclude that your RTD is about 0.3 mm thick. Am I close?Cheers,
Phil Hobbs
The datasheet
says 0.6. Not bad, given the likely accuracy of my input data. Their outline dims are a little smaller than the numbers you used, too. And I did solder on some #30 wires for the free-air measurement.
The data came from eyeballing a transient from a photograph of a scope waveform, just applying a current step to the RTD and looking at the voltage drop...
_____________ ______/ ____/ ___/ __/ | | | | | | | | ___________|
For the last case, the step was only about 0.2 cm, so the numbers get seriously un-serious. I'm checking to make sure it's OK to use this RTD on a pcb, as a thermocouple reference junction temp sensor. With some nice copper wings and reasonable (say, 5 mA, 1/10 duty cycle) excitation, it looks pretty good.
The "1% accuracy" thing is obviously dumb luck, one of my most valued talents.
John
Have something to add? Share your thoughts — no account required.
Ask the community — no account required